1 --- Log opened Mon Mar 17 12:20:15 2014
3 12:20 -!- Irssi: #ipdf: Total of 2 nicks [1 ops, 0 halfops, 0 voices, 1 normal]
4 12:20 -!- Irssi: Join to #ipdf was synced in 0 secs
5 12:20 -!- sulix changed the topic of #ipdf to: Angry Strawberry Summer
6 12:20 -!- mode/#ipdf [+o matches] by sulix
7 12:39 <@sulix> This is sensible discussion about the project.
8 12:39 <@matches> The suspense
9 12:39 <@sulix> Anyone up for lunch?
11 12:40 <@sulix> drumroll...
12 12:41 <@matches> I don't think it worked
13 12:42 <@matches> Yes I did it manually
14 12:42 <@matches> Maybe cron is broken on motsugo
15 12:43 <@sulix> We shall find out next time at 12:40, I guess.
16 12:43 <@sulix> (Or the next time cron emails errors out, maybe)
17 12:52 <@matches> Let the record show that it didn't work because I didn't install the crontab
18 --- Day changed Wed Mar 19 2014
19 10:59 <@matches> Some guy asked me if he could use the whiteboard, so I let him rub out our timeline
20 10:59 <@matches> And all he did was write "hello world"
22 19:10 <@matches> Testing
23 19:25 <@matches> So the IRC commits should have a different author
24 19:25 <@matches> So that I don't get a ridiculous number of daily IRC commits
25 19:26 <@matches> I could rebase the first three of them but I think that might break things
26 19:45 <@matches> Ok, and now it will only commit when there is more than one new line, so we won't get one every day when no one says anything
27 19:54 <@matches> Alright, and in theory
28 19:54 <@matches> If I spam this some more
29 19:54 <@matches> There will be a commit soonish
30 19:56 <@matches> Any second...
32 19:56 <@matches> Now that is sorted out I can attain maximum productivity
33 19:56 <@matches> (Sorry)
34 --- Log closed Mon Mar 24 01:12:25 2014
35 --- Log opened Mon Mar 24 08:40:12 2014
37 08:40 -!- Irssi: #ipdf: Total of 2 nicks [1 ops, 0 halfops, 0 voices, 1 normal]
38 08:40 -!- Irssi: Join to #ipdf was synced in 3 secs
39 13:26 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
40 13:27 < matches> Uh oh
41 13:29 -!- Netsplit over, joins: @sulix
42 13:31 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
43 13:32 -!- Netsplit over, joins: @sulix
44 13:40 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
45 13:40 -!- Netsplit over, joins: @sulix
46 16:19 < matches> Well that will make a nice automatic commit
47 --- Day changed Tue Mar 25 2014
48 21:33 < matches> So, I am pretty much free all day tomorrow
49 21:33 < matches> Also Thursday
50 21:34 <@sulix> I'm pretty busy Thursday, but I'm free all tomorrow so long as I can get some maths done at some point.
51 21:34 -!- mode/#ipdf [+o matches] by sulix
52 21:35 <@matches> I'm supervising a test from 2-3pm but that's it
53 21:35 <@matches> So, I will try and get to UWA as early as I can
54 21:35 <@matches> It might be a good idea to have some kind of start on a code base
55 21:35 <@matches> Even though we're supposed to be focusing on a literature review, I don't really like not having any code done yet
56 21:36 <@sulix> Yeah. I agree.
57 21:36 <@sulix> Code + fixing git stuff + preparing for this "talk" on Friday.
58 --- Day changed Wed Mar 26 2014
59 11:32 <@matches> I am in the super secret room
60 11:36 <@sulix> Okay, I'll head in shortly. Had a small "parents fell for a scam email and entered passwords on dodgy websites" crisis here this morning.
61 11:37 <@matches> Oh dear
62 11:42 <@matches> The phone is ringing :O
63 11:43 <@matches> Phew, it stopped
64 22:42 <@matches> I think I have a way to do tests in the Makefile without using cmake
65 22:42 <@matches> You can alter a variable for a target
66 23:20 <@matches> Well, that ended badly
67 23:21 <@matches> With me erasing my own Makefile using make
69 23:24 <@sulix> Better than erasing the source code, I guess.
70 23:34 <@matches> This is why we have git
71 --- Day changed Thu Mar 27 2014
72 00:15 <@sulix> Crazy templated loading and saving!
73 00:15 <@sulix> And now for bed!
75 00:30 <@matches> It doesn't break the test
76 01:22 <@matches> We now have an arbitrary precision float
77 01:22 <@matches> Where in this case arbitrary precision means I have no idea what the precision is but it is very bad
78 01:35 <@matches> Ok I think I made a half precision float
79 01:35 <@matches> They are quite bad
80 01:36 <@matches> Um, I can probably make a graph of it being terrible as my contribution for Friday
81 01:36 <@matches> Assuming I've done it right
82 --- Day changed Wed Apr 02 2014
83 09:41 <@matches> Dammit codejam is coming up again
84 09:41 <@matches> There goes productivity
85 09:42 <@matches> If you are not busy this afternoon we should do work
86 09:42 <@matches> On the project I mean
87 09:45 <@matches> Although it is tempting to just practice codejam problems for the next week
88 12:44 <@matches> If I work on the project it will help when the inevitable "64 bits is not enough" problem comes up!
89 12:45 <@matches> Maybe
90 --- Day changed Sun Apr 06 2014
91 13:40 <@matches> I get the feeling my part of the project could just be "typedef Real to something from boost"
92 14:15 <@matches> I suppose I could talk about FPU and hardware
93 --- Day changed Mon Apr 07 2014
94 21:42 <@matches> I had a horrible horrible thought... implement a FPU in VHDL and then somehow run all our floating point operations on it :P
95 21:42 <@matches> (This is not a good idea at all but it might be fun)
96 21:42 <@matches> You know, for some definition of fun
97 21:44 <@matches> So my lit review will probably be 1) We need higher precision documents because science (Pixels or Perish) 2) This is how FP works in hardware 3) This is how you can get higher precision in software
98 21:45 <@matches> Oh, and 4) Document formats and rendering them (PDF/PostScript etc)
99 21:46 <@matches> That is starting to sound suitably ridiculously broad?
100 21:46 <@matches> Can always cut things out I guess
101 21:47 <@matches> To reflect what we actually end up doing
102 22:05 <@sulix> From what Tim was saying, I don't think "too broad" is a possibility.
103 22:05 <@sulix> We could be talking about Aztec history and it'd probably not be "too broad."
104 22:05 <@sulix> I do need to remember to read "Pixels of Perish," though.
105 22:19 <@matches> I'll have a look into VHDL stuff, there seem to be compilers and simulators for linux
106 22:20 <@matches> For a minute I was afraid I'd have to use the UWA EE VHDL environment
107 22:20 <@matches> Which is like, running a Java program in a Windows XP VM
108 22:20 <@matches> I heard you liked simulations of hardware so I simulated some hardware in your simulated hardware
109 --- Day changed Tue Apr 08 2014
110 10:46 -!- Netsplit arctic.uniirc.com <-> mits.mits.au.uniirc.com, services.uniirc.com, irc.cassa.au.uniirc.com, mussel.ucc.au.uniirc.com quits: @sulix
111 10:46 -!- Netsplit over, joins: @sulix
112 11:31 <@matches> The "Experiment Running" sign seems to have worked
113 11:31 <@matches> Also open source VHDL stuff is less "actually compiles" than I had hoped
114 11:32 <@matches> The two I've tried so far seem to generate absolute monstrosities of C or C++ files and then fail to compile and/or link
115 --- Day changed Wed Apr 09 2014
116 15:27 <@sulix> So I've been hitting my head against my desk for about half an hour trying to work out where some bugs were in some ipdf code I'm writing.
117 15:27 <@sulix> Turns out to have mostly been precision issues due to the lack of precision in your Real data type.
118 15:28 <@sulix> Switching it to double fixes them all.
120 15:32 <@matches> Sorry
121 15:37 <@sulix> Nah: it's good. It means that precision actually is important!
122 15:37 <@sulix> (I was getting a little bit concerned for a while :P)
123 15:40 <@matches> Well
124 15:40 <@matches> I have rediscovered just how awful VHDL is
125 15:41 <@matches> It seems like you can't define anything without typing it three times
126 15:42 <@matches> You define an entity
127 15:42 <@matches> Then you define a component with identical ports
128 15:42 <@matches> Then you tell it to use the entity for that component
129 15:42 <@matches> Then you tell it to map the identical ports to the entity ports
130 15:42 <@matches> adder_0 : adder port map (a=>a, b=>b, cin=>cin, s=>s, cout=>cout);
131 15:44 <@matches> Pretty much all the hardware papers I found talk about using VHDL for things
132 15:46 <@matches> Let's see if I can make any sense of this jop-devel thing that appears to be a JVM implemented in VHDL
133 15:47 <@matches> Uh oh, .bat files
134 15:48 <@matches> This looks ominous
135 15:51 <@sulix> The whole swapping out main thing breaks the makefile in a few ways btw.
136 15:51 <@sulix> Because none of main.cpp's dependencies are listed.
138 15:51 <@sulix> So if you change main.h, nothing gets recompiled.
139 16:02 <@sulix> Okay, just pushed support for click+drag and zoom in and out in the viewer.
140 16:03 <@sulix> (I did change the default precision to single, rather than half, as zoom out was totally broken with half-precision)
141 16:05 <@matches> The half precision implementation is broken itself anyway
143 16:07 <@matches> I can't seem to compile anymore
144 16:07 <@matches> Hang on
145 16:07 <@matches> That would be because `uname -i` reports "unknown"
146 16:12 <@matches> I have made unknown default to x86_64 :S
147 16:13 <@matches> I can see the precision issues! Good work
148 16:14 <@matches> These lecture notes seem useful: http://www.cs.berkeley.edu/~demmel/cs267/lecture21/lecture21.html
149 16:14 <@matches> Lecture notes > Blog post in terms of suitable reference?
150 16:16 <@sulix> It's got fortran in it, it must be suitably acedemic.
151 16:18 <@matches> Haha
152 16:18 <@matches> The VHDL compiler I am experimenting with uses Ada
153 16:18 <@sulix> I'm getting my fill of new languages by trying out some c++11 features.
154 16:20 <@sulix> I think I broak realtofloat by changing things to be single precision by default.
155 16:22 <@sulix> (Hopefully fixed)
156 16:49 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
157 16:49 -!- Netsplit over, joins: @sulix
158 16:54 <@matches> What's with the netsplits
159 16:55 <@matches> You're on mussel? I'm on mantis? They are VMs on the same host?
160 16:55 <@matches> I don't know what this uniirc nonsense is
161 16:56 <@matches> What is "arctic"
162 16:56 -!- matches changed the topic of #ipdf to: Happy Bannana Netsplit
163 17:33 <@matches> Well I compiled the VHDL FPU
164 17:33 <@matches> It seems to work quite well
165 17:33 <@matches> Except for the part where it never actually finishes an operation
166 17:34 <@matches> One of the virtual transistors must have overheated
167 17:43 <@matches> Oops it might help to actually give it a valid operation
168 18:27 <@matches> Alright, so I'm going to experiment with getting our Real operations to run on the virtual FPU
169 18:28 <@matches> Then I can look into both hardware and software ways of getting arbitrary precision.
170 18:30 <@matches> VHDL appears to have text I/O so I could make everything read and write from pipes, but this seems like a terrible idea
171 18:30 <@matches> On the other hand the compiler is using Ada
172 19:21 <@matches> Oh wow
173 19:21 <@matches> To open stdout
174 19:22 <@matches> You pass it a string "STD_OUTPUT" for the filename
175 21:40 <@matches> Ok, I give up trying to link this thing
176 21:40 <@matches> fork() and exec() is seeming more and more elegant
177 22:11 <@matches> Nooo, buffering
178 22:11 <@matches> Ok, screw this
179 --- Day changed Thu Apr 10 2014
180 00:00 <@matches> Note: Actually exec'ing the program instead of just going straight to "exit(EXIT_FAILURE)" is generally important
181 00:00 <@matches> Wall of text commit message incoming
182 00:01 <@matches> Horribly inefficient interface to virtual FPU sort of implemented
183 00:02 <@matches> I would have made it based on binary read/write rather than hex strings but I could not work out how to do that in VHDL
184 00:08 <@matches> I still can't believe what I have just done, it seems ludicrous
185 00:08 <@matches> Using an entire executable and stdio operations
186 00:08 <@matches> To do a floating point addition
187 00:10 <@matches> You should see how much is involved in compiling the VHDL FPU...
188 00:11 <@matches> I'm very pleased that it actually compiled from the version in jop's github without much effort, but still
189 00:11 <@matches> It's like 30 object files
190 00:11 <@matches> This is why it isn't in the code repo
191 00:19 <@matches> Whoops, there are some blatantly wrong comments in vfpu.cpp
192 00:19 <@matches> Oh well
193 --- Day changed Fri Apr 11 2014
194 13:38 <@matches> Warning! I am about to make a ipdf/vfpu repo
195 13:43 <@matches> You know what's really cool
196 13:43 <@matches> You can run it as `vfpu --vcd=output.vcd`
197 13:43 <@matches> And it will create a gtkwave file as it runs showing all the operations
198 13:44 <@matches> You only get the simulated time from the point of view of the device
199 13:44 <@matches> Which is probably useful
200 13:46 <@matches> Or it would be if I wanted to compare different devices
201 13:47 <@matches> To do that I need to actually make different devices...
202 17:12 <@matches> Would you prefer seperate repos for the individual reports or just sub directories in the documents repo?
203 17:12 <@matches> I can't decide
204 17:12 <@matches> On the one hand 5+ repos is getting a bit out of hand
205 17:12 <@matches> On the other hand they are individual work
206 17:24 <@matches> Individual seems best
208 20:31 <@matches> The sqrt op in the fpu appears to have been commented out
209 20:31 <@matches> Also despite having constants for the sizes of things they have just used magic numbers everywhere
210 20:34 <@matches> I need an IRC script to prevent myself from saying stuff unless someone else has said things, or this channel will just be me ranting
211 --- Day changed Mon Apr 14 2014
212 13:42 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
213 13:43 -!- Netsplit over, joins: @sulix
214 14:42 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
215 14:43 -!- Netsplit over, joins: @sulix
216 18:32 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
217 18:32 -!- Netsplit over, joins: @sulix
218 18:34 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
219 18:35 -!- Netsplit over, joins: @sulix
220 18:36 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
221 18:38 -!- Netsplit over, joins: @sulix
222 18:40 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
223 18:41 -!- Netsplit over, joins: @sulix
224 18:41 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
225 18:43 -!- Netsplit over, joins: @sulix
226 18:45 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
227 18:45 -!- Netsplit over, joins: @sulix
228 18:46 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
229 18:47 -!- Netsplit over, joins: @sulix
230 18:54 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
231 18:56 -!- Netsplit over, joins: @sulix
232 18:59 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
233 19:00 -!- Netsplit over, joins: @sulix
234 19:03 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
235 19:04 -!- Netsplit over, joins: @sulix
236 19:07 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
237 19:08 -!- Netsplit over, joins: @sulix
238 19:10 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
239 19:10 -!- Netsplit over, joins: @sulix
240 19:13 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
241 19:13 -!- Netsplit over, joins: @sulix
242 19:15 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
243 19:15 -!- Netsplit over, joins: @sulix
244 19:17 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
245 19:18 -!- Netsplit over, joins: @sulix
246 19:21 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
247 19:22 -!- Netsplit over, joins: @sulix
248 19:23 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
249 19:23 -!- Netsplit over, joins: @sulix
250 19:26 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
251 19:26 -!- Netsplit over, joins: @sulix
252 19:27 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
253 19:29 -!- Netsplit over, joins: @sulix
254 19:30 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
255 19:31 -!- Netsplit over, joins: @sulix
256 19:32 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
258 19:35 -!- ServerMode/#ipdf [+o sulix] by mussel.ucc.au.uniirc.com
259 19:40 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
260 19:42 -!- Netsplit over, joins: @sulix
261 19:44 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
263 19:48 -!- ServerMode/#ipdf [+o sulix] by mussel.ucc.au.uniirc.com
264 19:54 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
265 19:54 -!- Netsplit over, joins: @sulix
266 19:54 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
267 19:54 -!- Netsplit over, joins: @sulix
268 19:55 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
269 19:56 -!- You're now known as 13VAAAAGJ
270 19:56 -!- Netsplit over, joins: @sulix
271 19:57 <@13VAAAAGJ> Why am I now known as 13VAAAAGJ
272 19:57 <@13VAAAAGJ> WHAT THE FUCK IS GOING ON
273 19:57 <@13VAAAAGJ> ARE WE UNDER ATTACK
274 19:57 <@13VAAAAGJ> Oh crap this is #ipdf
275 19:57 <@13VAAAAGJ> Um, disregard this bit of the totally project related conversation
276 19:57 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
277 19:57 <@13VAAAAGJ> I'd edit the commit script to censor the swear words, but effort
278 19:57 <@13VAAAAGJ> Oh dear
279 19:57 -!- Netsplit over, joins: @sulix
280 19:57 <@13VAAAAGJ> Hello
281 19:58 <@13VAAAAGJ> I think mantis might be dying?
282 19:59 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
283 20:00 -!- Netsplit over, joins: @sulix
284 20:01 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
285 20:01 -!- Netsplit over, joins: @sulix
286 20:04 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
287 20:04 -!- Netsplit over, joins: @sulix
288 20:05 <@13VAAAAGJ> Sigh
289 20:07 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
290 20:07 -!- Netsplit over, joins: @sulix
291 20:08 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
292 20:08 -!- Netsplit over, joins: @sulix
293 20:08 -!- Netsplit mantis.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @sulix
294 20:09 -!- Netsplit over, joins: @sulix
295 20:09 -!- Netsplit arctic.uniirc.com <-> mussel.ucc.au.uniirc.com quits: @sulix
296 20:10 -!- Netsplit over, joins: @sulix
297 --- Log closed Mon Apr 14 20:11:29 2014
298 --- Log opened Mon Apr 14 20:18:05 2014
300 20:18 -!- Irssi: #ipdf: Total of 2 nicks [1 ops, 0 halfops, 0 voices, 1 normal]
301 20:18 -!- Irssi: Join to #ipdf was synced in 0 secs
302 22:04 < matches> Ergh
303 22:06 < matches> http://szmoore.net/ghdl.bug
304 22:06 < matches> Pretty
305 22:10 < matches> Looks like we'll be sticking with ASCII input files for the FPU
306 --- Log closed Mon Apr 14 22:29:33 2014
307 --- Log opened Mon Apr 14 22:44:34 2014
309 22:44 -!- Irssi: #ipdf: Total of 1 nicks [1 ops, 0 halfops, 0 voices, 0 normal]
311 22:44 -!- Irssi: Join to #ipdf was synced in 12 secs
312 22:53 <@matches> Someone's VPS was compromised and killing our network
313 22:53 <@matches> For reference
314 22:53 <@matches> It affects this project because
315 23:07 <@matches> Argh
316 23:07 <@matches> So I thought I was being really clever
317 23:07 <@matches> I had files of type "bit" and everything
318 23:07 <@matches> Guess how a "bit" is represented...
319 23:08 <@matches> It's an ASCII "0" or "1"
320 23:09 <@matches> I guess I will just conclude that ASCII strings treated as hex is the only way
321 --- Day changed Wed Apr 16 2014
322 00:10 <@matches> I'm either about to perpetrate horrible things on your nice View class or give up and go to sleep
323 00:28 <@matches> Yeah... the problem is if we want to overlay objects rendered using different precision's over each other
324 00:28 <@matches> It leads towards templates
325 00:29 <@matches> And templates lead towards fear
326 00:29 <@matches> And fear leads towards anger
327 00:29 <@matches> And anger leads to hate
328 00:29 <@matches> I think
329 00:29 <@matches> Saving an image
330 00:29 <@matches> Recompiling the program
331 00:29 <@matches> Overlaying the image
332 00:29 <@matches> Is going to lead to less hate
333 00:30 <@matches> Otherwise pretty much every occurance of "Real" needs to be come a template
334 00:50 < sulix> So there are a couple of problems with doing that.
335 00:51 < sulix> - You'd need some way of storing objects of different precision. Possibly an extra "high-precision bounds" struct.
336 00:51 < sulix> - View also uses real for its internal bounds, which is where many of the problems appear.
337 00:52 < sulix> (I guess you could have a "high-precision viewport" as well.)
338 00:52 < sulix> - The actual upload of graphics data to OpenGL is done as 32-bit floats, no matter the type of real, as that's all OpenGL supports.
339 00:53 < sulix> (At some point we'll have to actually do some view transforms on the CPU rather than just passing all of the bounds and viewport straight to OpenGL)
340 01:02 <@matches> Yeah I don't want to replace Real with templates
341 01:02 <@matches> That'd be horrible
342 01:03 <@matches> By the way, the introduction of SDL_Renderer in SDL2 is confusing
343 01:05 <@matches> I thought it was replacing Surface with Renderer and Texture
344 01:05 <@matches> But Surface seems to still be a thing
345 01:13 < sulix> Yeah: basically Surfaces are in CPU memory, textures in GPU memory.
346 01:13 <@matches> Right
347 01:14 <@matches> And renderers magically put stuff on surfaces/textures
349 01:14 < sulix> So it basically represents the graphics card.
350 01:15 < sulix> But if you want to just use OpenGL yourself, you don't need to create a renderer at all.
351 01:15 <@matches> Ah, but the plot thickens
352 01:15 <@matches> I'm saving the window to a bmp
353 01:16 <@matches> Which requires a renderer
354 01:16 < sulix> Hmm... I'm not sure if that will work.
355 01:16 <@matches> It does
356 01:16 <@matches> It seems rather convoluted though to be honest
357 01:16 < sulix> (In theory, it shouldn't, but it probably actually does)
358 01:16 <@matches> I seem to remember it not being this convoluted in SDL1.2
360 01:16 <@matches> So the procedure is
361 01:17 <@matches> 1. Get a surface associated with the window
362 01:17 <@matches> 2. Allocate buffer of unsigned char for the pixels
363 01:17 < sulix> Yeah, TBH I prefer SDL1.2's rendering API, but I generally just use OpenGL anyway.
364 01:17 <@matches> 3. Get renderer from the window (NB: Don't use SDL_CreateRenderer, only the (undocumented) SDL_GetRenderer works because the Window already has a Renderer because it has been created with an OpenGL flag)...
365 01:18 < sulix> In theory we should probably use glReadPixels.
366 01:18 <@matches> 4. Use the magical renderer power to put pixels into the pixel array
367 01:18 <@matches> 5. Create an RGB surface from the pixel array
368 01:18 <@matches> 6. Now you can call SDL_SaveBMP, congratulations
369 01:18 <@matches> What's this about a glReadPixels
370 01:18 <@matches> I was just about to commit this
371 01:19 < sulix> Well, you can use glReadPixels instead of steps 1,3 and 4.
372 01:19 < sulix> (This is what SDL is doing behind the scenes)
373 01:20 < sulix> Unfortunately, the SDL wiki is still down thanks to heartbleed. :/
375 01:20 <@matches> It was up when I looked at it before
376 01:20 <@matches> Throwing python exceptions when I tried the search function though
377 01:21 < sulix> Ryan said that it was up, but I'm still getting revoked certificate errors.
378 01:22 <@matches> Hmm, I needed to do step 1 in order to know the size of the pixel array in step 2...
379 01:23 < sulix> Screen::ViewportWidth(), Screen::ViewportHeight()
380 01:26 <@matches> Yeah this is looking a lot shorter than what I had
381 01:33 <@matches> It doesn't seem to work though
382 01:35 <@matches> I'm getting a segfault
383 01:35 <@matches> And of course valgrind automatically exits when it gets more than 1000000 errors from the flgrx driver
384 01:38 <@matches> First call to glReadPixels is OK but the bmp is just white, second call segfaults
385 01:39 < sulix> Hmm... what's the call?
386 01:44 <@matches> Segfault was due to me forgetting about the pixels needing 4 bytes for RGBA
387 01:44 <@matches> Still white though
388 01:45 < sulix> (That was going to be my guess)
389 01:45 <@matches> The advantage of getting the SDL_Surface
390 01:45 <@matches> Was that you just pass all the surf->format->format->stuff
391 01:45 <@matches> Everywhere
392 01:45 <@matches> Also makes it rather verbose
393 01:46 < sulix> The white screen might just be an fglrx bug.
394 01:46 <@matches> There we go
395 01:47 <@matches> No, it helps to remember that the pixels need 4 bytes for RGBA
396 01:47 <@matches> I have very selective memory
397 01:47 <@matches> I had to remember it for each line individually
398 01:47 <@matches> Right I guess it is slightly nicer now
399 01:47 <@matches> Although it has a bunch of magical "*4" everywhere
400 01:48 <@matches> I'm going to put this on Stack overflow as an alternative to the answer I was originally following
401 01:48 < sulix> Well, I'm going to attempt to sleep...
402 01:48 <@matches> Thank you for fixing my bug without seeing it :S
403 01:49 < sulix> I have far too much experience breaking glReadPixels...
404 01:53 <@matches> We need an easy way to compare our document rendering the same thing using a different Real and/or view representation
405 01:53 <@matches> Templates would only solve changing the Real and really they probably wouldn't actually solve it
406 01:53 <@matches> They'd just create nightmares
409 01:55 <@matches> Just looking at View::Render
410 01:55 <@matches> Why is there a seperate loop for each type of object...
411 01:56 <@matches> With "continue" statements for the other types in each loop
412 01:56 <@matches> Is this so you can just make one call to glBegin and glEnd...
413 01:56 <@matches> I am suitable scared
414 01:57 <@matches> suitably scared and also suitable scared
415 22:05 <@matches> Ok, trying again. This is the sort of thing a template is supposed to be used for... I just seem to always end up suddenly having to make everything all the way back to Document into a template class :S
416 22:07 <@matches> blergh
417 22:07 <@matches> trying again
418 22:07 <@matches> After getting coffee
419 22:07 <@matches> I think
420 22:30 <@matches> Ok, templates is way too complicated
421 22:30 <@matches> I am going to do the following instead:
422 22:31 <@matches> 1. Allow a "save this region to bmp" argument
423 22:31 <@matches> 2. It reads the specified bmp, saves a new bmp with the current view overlayed in a different colour or something
424 22:32 <@matches> 3. Makefile hacks to recompile the program using a different typedef of Real and then do 1. and 2. for them all
425 22:32 <@matches> 4. Realise I probably should have used templates
426 23:02 <@matches> So, according to my timeline that I haven't looked at since submitting the proposal, I will have done a draft literature review by tomorrow...
428 --- Day changed Thu Apr 17 2014
429 00:26 <@matches> I am not good at OpenGL/SDL
430 00:26 <@matches> I am the master of producing a black screen
431 00:26 <@matches> Also we have FILLED and OUTLINE the wrong way round still
432 14:01 < sulix> Bunch of bugfixes incoming.
433 14:01 < sulix> I'm not proud of what I did to glReadPixels, but it works.
434 14:02 <@matches> Uh oh
435 14:02 <@matches> I just worked it out!
436 14:03 <@matches> The dreaded merge begins
437 14:03 <@matches> I'm tempted to just delete my changes and merge yours but I should probably do a real merge
438 14:04 <@matches> I fixed it (on my machine at least) by: Using GL_BGRA instead of GL_RGBA (should probably detect which one to use from the screen's format)
439 14:04 <@matches> And calling glReadBuffers and glPixelStorei before glReadPixels
440 14:05 < sulix> glPixelStorei is probably nicer than what I'm doing.
441 14:05 < sulix> I tried just giving SDL a negative pitch and a pointer to the bottom-right corner of the buffer, but somehow it didn't like that much.
442 14:07 <@matches> You do actually check the byte order which is good
443 14:08 <@matches> I worked that out but basically just changed it to match my machine :S
444 14:08 < sulix> There are a bunch of endian problems in the load-save code anyway if we want it to run on big-endian machines.
445 14:09 < sulix> (Well, big-endian documents don't load with little-endian code and vice-versa)
446 14:10 <@matches> http://szmoore.net/ipdf/code/src/screen.cpp
447 14:10 <@matches> Is the merge
448 14:10 <@matches> Which one is better?
449 14:11 < sulix> I think yours will still get the bitmaps upside down?
450 14:11 <@matches> Ah yes they do tend to be upside down
451 14:12 < sulix> I'm not sure you should need the front buffer stuff anymore with mine, so if you just keep mine (which is doing the flipping), everything should work.
452 14:12 <@matches> Woah you fixed the FILLED vs OUTLINE
453 14:12 < sulix> I did test overlaying bitmaps and it was fine.
454 14:12 <@matches> Good
455 14:12 <@matches> Does this seem like a better way to compare approaches than using a template and having View<float> View<double> etc?
456 14:13 <@matches> It means recompiling
457 14:13 <@matches> But it means a lot less templates
458 14:13 <@matches> Pretty much the entire thing has to be templatified that way
459 14:13 <@matches> And then there's the issue of what type the document is saving with
460 14:13 < sulix> I think, long term, it'd be worth just extending the view to have a "bounds" and a "high_precision_bounds" or something.
461 14:14 < sulix> But that is kind-of ugly.
462 14:14 <@matches> I think it's probably nicer to just have real.h contain nothing but a typedef
463 14:14 <@matches> And just recompile with a different typedef and overlay the images
464 14:15 < sulix> Yeah, but there are some artefacts that really show better in motion, so it's probably worth supporting that at some point.
465 14:15 <@matches> True
466 14:15 <@matches> But including a video in the pdf will be difficult anyway
467 14:16 < sulix> (Clearly, we should add embedded videos to ipdf)
468 14:16 <@matches> You could also make a movie using ffmpeg
469 14:16 <@matches> Haha
470 14:16 < sulix> (I've got some OpenGL video playback code somewhere...)
471 14:17 <@matches> In fact, if you can output a video you can just overlay two videos
472 14:17 < sulix> Oh... my... god!
473 14:17 <@matches> I don't know if there's a better way to make a video but I'd just be saving a .bmp every frame and then combine them all with fmpeg
474 14:17 < sulix> Nope, that's probably a good way.
475 14:18 < sulix> With some MAGIC it wouldn't even be slow.
476 14:20 <@matches> Whoops your code isn't giving me a second bmp anymore
477 14:20 <@matches> At least, not one that has pixels that aren't white
478 14:20 <@matches> Does my glReadBuffer thing work for you?
479 14:21 < sulix> Hmm...
480 14:21 <@matches> Can fix the upside-downness some other way
481 14:21 < sulix> Did you make clean
482 14:21 < sulix> Because main.cpp/main.h won't recompile otherwise.
483 14:21 < sulix> And I moved the glClear() calls.
484 14:22 <@matches> I did make clean
485 14:22 < sulix> Which is a less hacky way of solving the problem than reading the front buffer.
486 14:23 <@matches> So the problem is that the texture rendered by Screen::RenderBMP isn't in the buffer that glReadPixels reads from?
487 14:23 <@matches> (Even though it gets shown on the screen fine)
488 14:25 < sulix> Hmm... it works fine here with or without glReadBuffers.
489 14:27 < sulix> Try changing it to save the screenshot before calling scr.Present() in main.h
490 14:29 <@matches> That works
491 14:29 <@matches> But why?
492 14:29 < sulix> Because when you call Present(), it allocates a new buffer to render into.
493 14:30 <@matches> Right
494 14:30 < sulix> This is so that you can start rendering the new frame immediately, rather than having to wait for the window manager to actually re-render the screen.
495 14:31 <@matches> I have merged it
496 14:33 < sulix> Excellent.
497 14:33 < sulix> This is much more fun than literature!
499 14:35 <@matches> It's important
500 14:35 <@matches> It's part of the timeline
501 14:35 <@matches> Just you know
502 14:35 <@matches> It comes after the Literature Review
503 14:37 < sulix> Strictly speaking the Lit Review deadline is past, now, isn't it... :/
504 14:38 <@matches> Well technically it was never a deadline for me... :P
506 14:38 < sulix> I'm special.
507 14:39 <@matches> I need to read that last paper on FPUs more
508 14:39 <@matches> It was talking about how there are all these software techniques from the 1960s that can actually be done on the FPU itself
509 14:40 <@matches> I'll probably just work on getting our / jop's FPU working for different sized floats instead
510 14:40 <@matches> In theory that will teach me how they actually work
511 14:40 <@matches> Then I can actually understand the papers
512 14:40 < sulix> But can you write about it in a Lit Review?
513 14:42 <@matches> Maybe not so much the Lit Review, but it means I can write the Background section that explains how they work
514 14:43 <@matches> There's probably some early papers on them that I can reference at the same time if I try hard enough to find them
515 14:44 <@matches> The modern papers all have a lot of assumed knowledge
516 15:59 <@matches> Instead of reading papers I have progressed towards compiling different types and then overlaying them
517 16:02 <@matches> I didn't know you could do such cool things with Makefiles and flags to the compiler
518 16:02 <@matches> If people knew more about this maybe we wouldn't have to have python
519 16:03 <@matches> I will learn to do all my highlevel programming using Make
520 19:48 <@matches> Our Render/Screenshot combo for overlaying bitmaps and $$$ only works the first time it is done
521 19:49 <@matches> Or maybe it's RenderBMP that only works once
523 19:52 <@matches> I swear the format was BGRA earlier and now it's RGBA
524 19:53 <@matches> This can probably get fixed after some Literature Review
525 20:09 <@matches> Anyway you can now set the type of Real with `make DEF=-DREAL=X`
526 20:09 <@matches> Or `make single` and `make double`
527 20:09 <@matches> 0 and 1 respectively
528 20:09 <@matches> I tried a little bit too hard to get it working with actual C strings
529 21:32 <@matches> Minutes now have less "~*" in them
530 21:32 <@matches> Wrong channel
531 21:32 <@matches> Very wrong channel
532 --- Day changed Mon Apr 21 2014
533 12:32 <@matches> Literature Review O'Clock
534 12:36 < sulix> I have just returned from my Easter holiday and have some horrible code for you to scream at while I have lunch. :P
535 12:43 <@matches> Uh oh
536 12:44 <@matches> stb_truetype ?
537 12:44 <@matches> DejaVuSansMono.ttf ?!
538 12:44 <@matches> Is this going to render a font...
539 12:45 <@matches> Woah
540 12:47 <@matches> Looks good
541 12:48 <@matches> I think I will avoid looking at stb_truetype too closely...
542 12:48 <@matches> Too late
543 12:48 <@matches> Why are all the implementations in the .h file
544 12:48 <@matches> And the .cpp file just has some defines...
545 12:48 <@matches> shudder
546 12:53 <@matches> I particularly like #define STBTT_assert(...) do {} while(0)
547 12:53 < sulix> That's a standard way of doing that. SDL_assert does the same thing.
548 12:54 < sulix> It lets you put a semicolon after it or use it in an if() statement without braces, etc, without things breaking horribly.
549 12:56 <@matches> But... your assert doesn't actually assert anything?
550 12:56 < sulix> That's because you generally only want it enabled in debug builds.
551 12:56 <@matches> I guess
552 12:57 < sulix> So typically you'd have it as do {} while(0) in release builds, and something like a function call in debug builds.
553 12:57 <@matches> So what is this "more code" going to do :P
554 12:57 < sulix> Slightly more modern opengl.
555 12:58 <@matches> Yeah, that would be a good idea
556 12:58 < sulix> In particular, uploading the content of a document to a VBO, and just re-rendering from that when needed.
557 12:58 <@matches> Cool
558 12:58 < sulix> Jumps from 9 fps -> 37 fps with 1024^2 objects.
559 12:59 < sulix> I've been without internet, though, so it's pretty broken in many ways, still.
560 12:59 < sulix> Like outline rects don't work.
561 12:59 < sulix> And it still uses a bunch of OpenGL 1.0 stuff.
562 12:59 < sulix> And for some reason the glPrimitiveRestartIndex() function isn't working.
563 13:00 < sulix> Also I need to write shaders at some point, which will be work.
564 13:00 <@matches> I just found the best comment in an IEEE email list
565 13:00 <@matches> We should totally use it
566 13:01 <@matches> "It is too late now to repair the mistakes of the past that are present in millions of installed systems, but it is good to know that careful research before designing hardware can be helpful"
567 13:01 < sulix> Oh man, that's basically the thesis in one sentance.
568 13:02 <@matches> Dibs!
569 13:02 <@matches> Although I doubt anyone will care if we both quote it
570 13:03 < sulix> One thing I did discover reading my OpenGL book is that I think we can get OpenGL to use doubles with some shader hackery.
572 13:03 < sulix> Then again, the book was (partly) written by (one of) the authors of fglrx, so take it with the requisite salt.
573 13:04 <@matches> Ah, so how many goats were required?
574 13:05 < sulix> Just the two, I think.
575 13:06 <@matches> This email message seems sort of vaguely relevant? It's the "infamous double rounding problem"
576 13:06 < sulix> I take it that you worked out how to swap between CPU and GPU transforms with right-click.
577 13:07 <@matches> I read it in the commit message
578 13:08 < sulix> I was quite happy with how much nicer things ran with CPU transforms and doubles.
579 13:08 <@matches> I think I can see a difference zooming in far enough
580 13:08 <@matches> Yeah that's a huge difference
581 13:08 <@matches> I had a tester that was going to automatically make images of the difference when you zoomed in really far
582 13:09 <@matches> But there were issues with the screen shot / bmp rendering
583 13:09 <@matches> Probably the bmp rendering
584 13:09 < sulix> .bmp files suck.
585 13:10 <@matches> They are nice and simple though?
586 13:10 <@matches> At least, the SDL API is nice and simple
587 13:10 < sulix> That's more a triumph of SDL than of the .bmp file format.
588 13:10 <@matches> After looking through gm6 files with a hex editor to try and extract sprites...
589 13:11 <@matches> Well we could always just fwrite the pixel buffer and fread it back
590 13:11 <@matches> If you want no one to be able to use our files without compiling our program :P
591 13:12 < sulix> I'm personally a fan of LodePNG, but I'm too lazy to hack it in at the moment: http://lodev.org/lodepng/
592 13:13 < sulix> I'm probably going to get around to fixing all of the warnings in stb_truetype one day.
593 13:13 <@matches> Haha
594 13:13 <@matches> What about implementing an Oct-tree?
595 13:14 <@matches> So .bmp is just a header and then the raw pixel data? What's disgusting about that? No compression?
596 13:16 <@matches> Wait png is also just a header and then an IDAT section
597 13:17 <@matches> Oh well you can put lodepng in if you want I guess
598 13:17 <@matches> As long as I can open the images in eom
599 13:18 <@matches> Ah so PNG is compressed. I missed that
601 13:19 <@matches> Oh and alpha channel
602 13:19 <@matches> Yeah
603 13:19 <@matches> I remember that now
604 13:20 <@matches> xpaint is the least shitty of the shitty sprite editors in debian, but it only stores RGB because it was designed around bmp and I guess they just hacked on exporting to other formats later
605 13:21 <@matches> If you open things with alpha it gets really confused
606 13:23 < sulix> You can put alpha in .bmp files, but then they sometimes crash KDE.
607 13:26 <@matches> We can just run all our bmp files through image magick
608 13:26 <@matches> convert a.bmp a.png
609 13:26 <@matches> Done
610 13:27 <@matches> Hmm, maybe if I opened the data from the gm6 files as png they would work, I was assuming they were bmps -_-
611 13:27 <@matches> Actually gif
612 13:27 <@matches> Probably
613 13:28 <@matches> Since it has "Save as gif"
614 13:28 <@matches> But that is not relevant to this project
615 13:33 < sulix> Damn, just discovered that the opengl feature I was using for rectangle outlines is not supported by my laptop's hardware.
616 13:34 < sulix> (This explains one of the bugs I've been seeing, but sadly only one)
617 13:47 < sulix> Okay, remaining code pushed.
618 13:47 < sulix> I am sorry.
619 13:53 <@matches> I'm messing around trying to do something evil so I'll merge later
620 13:53 <@matches> But you don't seem to call ToggleGPUTransform when there is a right click
621 13:54 <@matches> Oh wait, I need to make clean
622 13:54 <@matches> Cool
623 13:54 <@matches> (So much for our Makefile magically tracking the header files :S)
624 13:58 < sulix> It does it for every header file except ones included by main.cpp because of the weird main.cpp swapping stuff.
625 13:59 <@matches> That main swapping stuff is amazing
626 13:59 <@matches> It is worth it
627 13:59 <@matches> Except main.cpp ends up eventually including all the header files
628 14:00 <@matches> I think there is an alternative Makefile that just has every single header file in the current directory as a dependency for each .o file
629 14:01 <@matches> But I'm normally used to running make clean anyway
630 14:22 <@matches> Eh I was going to make a video of the difference between GPU and CPU coord transforms but it's not that great
631 14:23 <@matches> Whatever you have done does not look happy
632 14:24 < sulix> It takes like 5 minutes to DebugDumpObjects with a million objects, but it works okay* after that.
633 14:24 <@matches> Yeah ok
634 14:24 < sulix> *well, sort-of.
635 14:25 < sulix> If you just want to show off the precision, getting rid of most of those objects is probably the right thing to do.
636 14:25 <@matches> Interesting things happen when you zoom out really far
637 14:25 <@matches> Do we care about that
638 14:25 < sulix> I'm going to go with "no".
639 14:26 <@matches> Ah, but the rectangles are not all the same size
640 14:26 < sulix> This is a product of rounding to the nearest pixel. They'd be fine with antialiasing.
641 14:29 <@matches> I have pushed a single line contribution
642 14:29 <@matches> I decided not to bother with all the changes I made to make the video prettier
643 14:29 <@matches> Since I can't actually capture my desktop properly anyway
644 14:30 <@matches> I suppose I could make it render bmps and ffmpeg them all but that is way too much effort
646 14:31 <@matches> Literature Reviews
647 14:31 <@matches> They are a thing
648 14:38 <@matches> I probably should not reference an email thread
649 14:53 <@matches> In other news, I think this jvm FPU doesn't actually implement sqrt
650 14:57 <@matches> It always gives 0x0 on sqrt ops
651 14:57 <@matches> I certainly hope there isn't a jvm somewhere that just doesn't do square roots
652 14:58 <@matches> They probably hacked it into some other part of their project instead of putting it in the FPU
653 16:08 <@matches> So fprintf style formatting isn't very clever when it comes to floats vs doubles vs long doubles
654 16:11 <@matches> Eg: %lf will work for either floats or doubles and %f will work for either floats or doubles (by truncating) but %llf is the only thing that works for long doubles and it only works for long doubles
655 16:12 <@matches> s/truncating/casting whatever
656 16:52 <@matches> We can compile with long double
657 16:52 <@matches> I suspect this may break things but whatever
658 16:52 <@matches> Running calculatepi on motsugo to see how much difference it actually makes...
659 17:08 <@matches> Charles Babbage seems like an interesting fellow
660 17:09 <@matches> "e sought to prove the
661 17:09 <@matches> reality of the devil by drawing with his blood a
662 17:09 <@matches> circle on the floor and repeating the Lord’s prayer
663 17:09 <@matches> backward"
664 17:09 <@matches> The things one does for science
665 17:44 <@matches> Reading about Charles Babbage counts as Literature Review...
666 17:44 <@matches> Actually I'm trying to find a paper written *by* him
667 17:44 <@matches> Since apparently he came up with Floating Point representations
668 22:11 <@matches> Found an actual paper talking about floating points on GPUs
669 22:12 <@matches> It says they don't conform to IEEE and also the manufacturers don't like to tell people what they do do
670 22:12 <@matches> So they wrote a version of Paranoia, which was a test program for computers in the 1980s before IEEE, to work out the characteristics of flops on various GPUs
671 22:13 <@matches> This is probably of interest
672 22:14 <@matches> So far one of the best papers I have on algorithms in software is actually one talking about implementing those algorithms in hardware instead
673 22:15 <@matches> I also have a random obituary to Charles Babbage just because
674 22:16 <@matches> Unfortunately UWA directs me to some useless website that just tells me the citation details and doesn't give me a download for any of Babbage's papers
675 22:17 <@matches> He has some books but they are mostly about economics
676 22:18 <@matches> Would have been nice if he'd written something that says "I have invented floating point" but I guess not
677 22:18 <@matches> That's only really of historical interest but it'd be nice
678 --- Day changed Tue Apr 22 2014
679 01:17 <@matches> Urgh looking at the git diff
680 01:17 <@matches> I really have not accomplished much
681 01:30 <@matches> Oooh, motsugo finally got to a number of intervals where long double is better than double
682 11:56 < sulix> So it turns out that I've broken the open-source GL drivers.
683 11:56 < sulix> They don't support one of the features ("primitive restart") we need when in compatibility mode.
684 11:56 < sulix> So I guess now I've got to rewrite everything to use OpenGL 3.
686 14:48 <@matches> I should probably look more carefully at what the OpenGL stuff is doing nowdays
687 14:49 <@matches> Downgrading to OpenGL 3 sounds drastic
688 14:50 <@matches> Wait 3 is the OK one, 4 is the new one
689 14:50 <@matches> 1 is the ancient one
690 14:52 <@matches> I see pointer arithmetic...
691 14:53 <@matches> banana would be furious
692 14:53 <@matches> should be using std::super_ptr_unsegfaultable_arith
693 14:53 <@matches> (I am not a fan of the smart pointers)
694 14:55 <@matches> *indexData = 0xFFFFFFFF; // Primitive restart.
695 14:55 <@matches> I do not know what is going on here :S
696 14:58 <@matches> Ahh I get it now
697 14:58 <@matches> Reading commit messages turns out to be useful
698 15:04 <@matches> git blame for view.cpp reveals that I still own the opening braces on some things
699 15:05 <@matches> That's about it :P
700 15:06 <@matches> I'm going to keep looking at rounding errors and maybe have a better thing than calculating pi
701 15:06 <@matches> As a bench mark
702 15:07 <@matches> I might look at Paranoia
703 15:07 <@matches> Although it was originally in BASIC
704 15:07 <@matches> And ~7000 lines
705 15:07 <@matches> With almost no whitespace characters
706 15:07 <@matches> And totally no indents
707 15:08 <@matches> I sometimes get the feeling people used to be smarter than we are now...
708 15:08 <@matches> Then I remember that those people are ultimately responsible for the tools we use now
709 15:15 < sulix> Speaking of tools we use now, I think I've just got the debug font code to not only randomly corrupt a varible, but also cause valgrind to crash.
710 15:18 < sulix> Okay, we need to fix that makefile at some point.
711 15:38 <@matches> make clean?
712 15:41 <@matches> It's not actually swapping out main.cpp
713 15:41 <@matches> It just doesn't have main.o in the link objects and has main in the $(BIN) target and the testers in the tests/% target
714 15:42 <@matches> I guess that is swapping out main but MAIN is not being changed
715 15:42 <@matches> derp
716 15:45 <@matches> I think I fixed it?
717 15:45 < sulix> Ah. Well, you've got a merge conflict to look forward to.
718 15:45 <@matches> Not if I just NEVER COMMIT IT
719 15:46 <@matches> I'm going to call your fix being right and mine being a horrible hack
720 15:46 <@matches> Ok I pullsd some changes to graphics stuff, is that it or is there more?
721 15:46 < sulix> I haven't fixed it, I've just gone and bu(gg|ff)ered up the font stuff.
722 15:47 <@matches> So it's not meant to render unless something changes? But the debug font stuff is changing so it needs to keep rendering?
723 15:48 <@matches> I see a lot of "Flushing Debug Font arrays"
724 15:48 < sulix> The debug font stuff is separate from the whole view system.
725 15:48 <@matches> That makes sense
726 15:48 < sulix> It's just trying to fill up a buffer of quads with individual characters in it and then draws them when the buffer gets full.
727 15:49 < sulix> One more step on the path of getting rid of all of the OpenGL 1.1 stuff.
728 15:50 <@matches> Haha
729 15:50 <@matches> git stash seems to think I modified graphicsbuffer.h and screen.h
730 15:51 <@matches> git diff seems to think everything is identical
731 15:51 <@matches> I'm just going to git reset the things I didn't actually change...
732 15:52 < sulix> I do that far too often.
733 15:52 <@matches> Oh there is a comment "//test" for testing the Makefile
734 15:53 <@matches> And a comment "This isn't the Screen class?" where you copy/pasted the screen class description for the GraphicsBuffer class
735 15:53 <@matches> I don't think they need to be preserved
736 15:54 < sulix> Yeah, I should have got rid of that screen class bit by now.
737 15:55 <@matches> Have a 2 line commit
738 15:55 < sulix> The best kind.
739 15:55 <@matches> I really need to cut back on the new line creep
740 15:56 <@matches> It's like I can't add a line by itself without putting in extra whitespace
741 15:57 <@matches> Uh oh we've ran out of coffee here
742 15:58 <@matches> I'm slightly scared by how much the graphics code has increased since I last actually understood how it worked...
743 15:59 <@matches> Can you change those *vertexData to *(vertexData++) or is that considered even uglier
744 16:03 < sulix> I thought of that, but given how much pointer arithmetic debugging I was going to do, I wanted it to be really obvious when the pointer was being incremented.
745 16:03 < sulix> The plan is to have a nice AppendFloat() function or similar that will do this eventually.
746 16:05 < sulix> Hmm: "Buffer usage warning: Analysis of buffer object 2 (bound to GL_ARRAY_BUFFER_ARB) usage indicates that the GPU is the primary producer and consumer of data for this buffer object. The usage hint supplied with this buffer object, GL_STATIC_DRAW, is inconsistent with this usage pattern. Try using GL_STREAM_COPY_ARB, GL_STATIC_COPY_ARB or GL_DYNAMIC_COPY_ARB instead.
747 16:06 < sulix> Thanks nVidia.
748 16:06 < sulix> Now if only we were using OpenGL 4.4 which got rid of buffer usage hints entirely.
749 16:29 <@matches> Also I think you forgot the naming scheme :P
750 16:30 <@matches> vertex_data
751 16:31 <@matches> I will go back to pretending to be doing a literature review
752 16:32 <@matches> Instead I will probably plot some graphs
754 16:48 <@matches> If you look up "Handbook of Floating Point Arithmetic" on google (which lots of things like to reference these days)
755 16:48 <@matches> You can download the entire thing
756 16:48 <@matches> I was prepared to pay like $20 for it on amazon
757 16:49 <@matches> Oh amazon doesn't sell actual books anymore though does it?
758 16:49 <@matches> No they do have it
759 16:50 <@matches> For $100...
760 16:50 <@matches> I think I'll just stick with my free pdf thanks
761 16:55 <@matches> "Handbook" being 579 pages...
762 17:04 <@matches> I think if a textbook is citing blog posts we can probably get away with it
763 17:05 <@matches> Oh my god I love this textbook
764 17:05 < sulix> I was watching a conference talk last night when the presenter just said: "this technique is described on this guys blog. Here's a link."
765 17:05 <@matches> "Table 1.1 gives the results obtained by compiling Program 1.1 and running on a Pentium4, using the GNU Compiler Collection and the Linux system"
766 17:06 <@matches> [Complete C code for Program 1.1 follows]
767 17:06 <@matches> None of this "Pseudo code" crap
768 17:07 < sulix> Clearly we should get an extra 10% for every line of pointer arithmetic in our theses.
769 17:07 <@matches> There'
770 17:07 <@matches> s a problem here
771 17:07 <@matches> How do I not make the entire literature review just paraphrased from this one text book
772 17:08 <@matches> Mr Gullible and the Chaotic Bank Society!
773 17:08 <@matches> It has stories!
774 17:08 <@matches> Parents should read this textbook to their children
775 17:09 < sulix> I'm pretty certain I know someone with a data structures and algorithms picture book...
776 17:09 <@matches> That reminds me of a children's book I wrote about multithreading
777 17:10 <@matches> I should probably scan it one day
778 17:10 < sulix> Did it begin "Once upa timeon...
779 17:10 <@matches> I think it did actually
780 17:10 <@matches> Once upon a time there was a process who had a lot of work to do...
781 17:11 <@matches> "I should probably scan it one day" - Why not TODAY!
782 17:11 * matches goes searching
783 17:16 <@matches> It is no where to be found
784 17:16 <@matches> I distinctly remember going to throw it out and deciding not to, but not where it actually went
785 17:27 <@matches> So maybe compiling a bunch of HFPA's examples using our Real type is a good way to make benchmarks
786 17:27 <@matches> I don't know
787 17:27 <@matches> My part of the project seems to move further and further away from the document formats thing
788 17:27 <@matches> Maybe I'll try and compile the GPU Paranoia
790 17:34 <@matches> It seems like whatever Mathematica does is what we should do
791 17:35 <@matches> I wonder if wolfram is open about how Mathematica actually works
792 17:35 <@matches> I don't think they are
793 17:35 <@matches> The CQM lecturer for physics found a bug in Mathematica's number representation once that she showed us
794 17:35 <@matches> Apparently she reported it years ago and it's still there
795 17:36 <@matches> This segued nicely into why we should learn fortran
796 --- Day changed Wed Apr 23 2014
797 14:00 <@matches> IEEE's float encoding is inconvenient
798 14:01 <@matches> The mantissa is like, reversed
799 14:01 <@matches> b_{23-i} * 2^{-i}
800 14:19 <@matches> And the implicit extra 1 except when it isn't is a pain in the ass as well
801 14:19 <@matches> It's like
802 14:19 <@matches> I can't just copy the right bits I have to think
803 14:19 <@matches> What is this
804 14:19 <@matches> Mental effort
805 14:19 <@matches> Blargh
806 14:20 <@matches> I should have just stuck with the 8 bit floats with their 2 sign bits
807 14:20 <@matches> The hydra float
808 15:17 <@matches> Wow you get really different representations with an IEEE encoding of the bits
810 15:17 <@matches> I think I liked just treating the two parts as integers in the order they were stored better
811 15:18 <@matches> I have to call pow(3) and a floating point division for every '1' in the mantissa to get it to agree with IEEE
812 15:19 <@matches> Unless I'm missing some obvious trick
813 15:20 <@matches> But in IEEE the mantissa is not an integer
814 15:37 <@matches> Ooookkk
815 15:38 <@matches> You're precision doesn't disappear as fast when you do it that way
816 15:38 <@matches> Your
817 15:38 <@matches> Whatever
818 --- Day changed Thu Apr 24 2014
819 14:40 -!- You're now known as notmatches
820 14:40 -!- You're now known as matches
821 14:41 <@matches> Ok, less spammy channel time
822 14:41 -!- mode/#ipdf [+b m:matches!*@*] by matches
823 14:41 <@matches> Spam
824 14:41 <@matches> Dammit
825 14:42 -!- mode/#ipdf [-o matches] by matches
826 14:42 < matches> Spam
828 14:43 < matches> ... this will all appear in the next git commit won't it -_-
829 14:52 -!- You're now known as notmatches
830 14:52 -!- You're now known as matches
831 --- Day changed Sat Apr 26 2014
832 00:45 -!- mode/#ipdf [+o matches] by OperServ
833 00:46 <@matches> Incoming "at least it is self consistent in its wrongness" floating point conversion
834 00:48 -!- mode/#ipdf [-o matches] by matches
835 00:48 < matches> Still can't talk
836 00:48 < matches> Excellent
837 --- Day changed Sun Apr 27 2014
838 17:10 < sulix> I finally got ipdf working on my laptop again by porting everything to OpenGL 3.1.
839 17:10 < sulix> I'm so sorry about how ugly the code now is, by the way.
840 18:04 -!- mode/#ipdf [+o matches] by OperServ
842 18:05 < Pommers> You know what a really good feature for a PDF reader would be? The ability to have it completely borderless without menus, so you just see the page!
845 18:06 <@matches> Well we've got that
846 18:06 <@matches> We just can't read pdfs
847 18:07 <@matches> Did you get much done not going to the cleanup?
848 18:07 <@matches> I managed to transfer some dust from one location to another location, and then the wind blew it back in my face
849 18:08 <@matches> Then I looked at Frames' proposal and despaired at how much it is actually complete and has words in it
850 18:08 <@matches> I think he's onto something with his plan of actually writing things
851 18:14 -!- matches changed the topic of #ipdf to: ipdf the pdf doesn't stand for pdf
852 18:14 -!- matches changed the topic of #ipdf to: ipdf the pdf doesn't stand for pdf but the df stands for df
853 18:14 -!- matches changed the topic of #ipdf to: Sort of
854 18:15 <@matches> I need a script to auto mute me...
855 18:15 < sulix> I spent basically all of today on that one commit.
856 18:15 -!- mode/#ipdf [-o matches] by matches
857 18:15 -!- mode/#ipdf [+o matches] by OperServ
858 18:15 < sulix> Whether or not that was more useful than the cleanup remains to be seen.
859 18:16 <@matches> I should look at it I guess :P
860 18:16 < sulix> It compiles*
861 18:16 < sulix> *but has lots of warnings.
862 18:16 <@matches> It can't be worse than my tester that converts floats to floats
863 18:17 < sulix> I'm pretty certain it can be and is worse.
864 18:17 <@matches> However I've realised I kind of actually do need to be able to do that
865 18:17 <@matches> Woah 3k+ lines
866 18:17 <@matches> * 4k+ lines
867 18:18 < sulix> Most of it was autogenerated, but still...
868 18:18 * matches weeps for OpenGL 1
869 18:18 < sulix> You don't realise how much you miss it until it's gone...
870 18:19 <@matches> So what does all this stuff actually do
871 18:19 <@matches> Because I could see rectangles and things just fine without it
872 18:19 < sulix> Exactly the same thing as before but it works on my laptop now.
873 18:19 <@matches> I see a shader program, that's obviously good
874 18:20 <@matches> Actually I see code for a shader but no shader yet
875 18:20 < sulix> The shaders are all #defines, I'm afraid.
876 18:22 <@matches> What was wrong with your laptop?
877 18:23 < sulix> It doesn't support the OpenGL compatibility mode, so --- as I was using the OpenGL 3.1 feature of "primitive restart", I had to code the entire thing in OpenGL 3.1
878 18:23 < sulix> On the bright side, this paves the way for doing things like getting the GPU to use doubles.
880 18:27 < sulix> (also halves)
881 18:27 < sulix> (halfs?)
882 18:27 <@matches> Do GPUs use IEEE floats? I had a reference that seemed to be complaining they didn't
883 18:27 <@matches> Although it was circa 2007
884 18:28 < sulix> I don't think that there's a requirement that they internally use them, but they do read and write them.
885 18:28 < sulix> A bit like how x86 processors use(d) 80 bit reals internally and rounded when reading/writing to memory.
886 18:28 <@matches> Do halves have a different mantissa encoding
887 18:29 <@matches> The examples given on wikipedia seem wrong
888 18:30 < sulix> I think this is the canonical description: http://www.opengl.org/registry/specs/NV/half_float.txt
889 18:33 <@matches> That would be before IEEE specified half (binary16) but its the same summation for the mantissa anyway
890 18:33 <@matches> binary16 is (briefly) in the 2008 revised 754
891 18:34 <@matches> So I think wikipedia is giving a wrong example, but I'm not confident enough to go and change wikipedia
892 18:35 <@matches> Also I just realised my conversion of the mantissa to a Real is horrible
893 18:35 <@matches> Oh well, it can get fixed later
894 18:36 <@matches> Hopefully I can finish my fluid mechanics assignment quick enough to actually work on my progress report / proposal / thing
895 18:36 <@matches> (Hah)
896 18:36 -!- mode/#ipdf [-o matches] by matches
897 18:37 < sulix> Yeah, I'm going to see how much Lit review I can write for tomorrow. :/
898 --- Day changed Mon Apr 28 2014
899 22:06 -!- mode/#ipdf [+o matches] by OperServ
900 22:10 <@matches> Must find motivation to work on Lit Review at 10pm...
901 --- Day changed Tue Apr 29 2014
902 10:06 <@matches> Must find motivation to work on Lit Review at 10am...
903 --- Day changed Wed Apr 30 2014
904 13:04 <@matches> I wanted to rasterise a vector image so I could compare them at the same scale ("These look the same!") and zoomed ("This one looks crappier!)
905 13:04 <@matches> But because vector graphics editor/viewers don't use pixels as units but they do when they export to a bitmap, it's difficult to actually get them to look the same
906 13:04 <@matches> Before scaling
907 13:05 <@matches> I guess a screenshot tool might be the best way
908 13:06 <@matches> I don't know if I need to do this really
909 13:07 <@matches> I guess Mechanical and Chemical engineers marking this will probably benefit from having an example
910 13:07 <@matches> Doesn't everyone know the difference between vector and raster graphics
911 13:07 <@matches> Where was that paper that had tux vector and rasterised
912 13:10 <@matches> Ah, worth2003xr.pdf
913 13:11 <@matches> If it's in a paper I guess it can go in a Lit Review
914 13:11 <@matches> At least until I have something better
915 13:15 <@matches> Kind of ironic that the image I am using was actually scanned first as a bitmap and then converted to vector using Trace Bitmap
916 13:57 <@matches> Ah, it's actually impossible to get it to be exactly the same, because even taking a screenshot on my own screen it will then be different depending on the display of whoever reads the digital pdf
917 13:57 <@matches> Oh well
918 13:58 <@matches> I suppose "It looks shittier" will have to suffice without trying to make them look exactly the same before scaling
919 13:59 <@matches> I have spent WAY too long making this example
920 13:59 <@matches> I could have just gone "See \cite{worth2003xr.pdf}"
921 16:00 <@matches> So it might be worth talking about dpi in pdf viewers and how it SUCKS
922 20:17 -!- mode/#ipdf [-o matches] by matches
923 21:14 -!- mode/#ipdf [+o matches] by OperServ
924 21:46 <@matches> I have perpetrated XML on the codebase
925 21:47 <@matches> I have grand visions of our code supporting SVGs
926 21:47 <@matches> I also have grand visions of actually doing a Literature Review
927 21:50 * sulix git pull's with some trepidation.
928 21:52 < sulix> Never heard of pugixml before but it looks okay.
929 21:55 <@matches> The W3C XML specification is pretty terrifying
930 21:56 <@matches> I will feel more like I've satisfied the "Document Formats" part of the Literature Review if I say some things about it
931 21:57 <@matches> Well SVG in particular
932 21:57 <@matches> SVG defines a "minimum" precision of IEEE binary32
933 21:58 <@matches> But there's a specification for "High Quality" viewers that have to use binary64
934 21:58 <@matches> That's probably the only real thing relevant directly to our problem
935 --- Day changed Thu May 01 2014
936 01:23 <@matches> It's May 1st
937 01:23 <@matches> This means we can no longer say "The Literature Review is due Next Month"
938 01:23 <@matches> IT'S DUE THIS MONTH
939 01:23 * matches freaks out
940 01:23 <@matches> ... but after sleep
941 01:25 <@matches> Page 12 of my Literature Review by the way
942 01:25 <@matches> Is the only page I like
943 16:34 <@matches> The C version of paranoia compiled for me
944 16:34 <@matches> Not terribly exciting (I have an IEEE 754 compatible processor! Amazing)
945 22:17 <@matches> W Kahan's website is a very interesting if slightly difficult read
946 22:24 <@matches> He appears to have written this 80 page pdf in a day
947 22:27 <@matches> It kind of reads like one of those religious propaganda pamphlets
948 22:27 <@matches> "Java is the Work of Satan"
949 22:27 <@matches> "Kernighan-Ritcie C floating-point semantics are the light"
950 22:28 <@matches> But every so often he has a graph or example that makes him seem less crazy
951 22:34 <@matches> "And now Java forbits you to mention or use extra-precise long double arithmetic, though IEEE Standard 754 recommends its use and over 95% of computers on desktops have it built into their hardware. You paid for it, but Java denies you its benefits."
952 22:34 <@matches> Java has long double now right?
953 22:34 <@matches> Although that JOP I was looking at was just 32 bit
954 22:36 <@matches> Ah, java.lang.math.BigDecimal
955 22:37 <@matches> "But be careful with division, because it will throw exceptions if it's like 1/3, then it will be Non-terminating decimal expansion."
956 22:37 <@matches> That sounds horrifying
957 22:40 -!- mode/#ipdf [-o matches] by matches
958 --- Day changed Sat May 03 2014
959 22:01 < sulix> My crazy idea to you would be to research p-adic number representations.
960 22:01 < sulix> It's been a while, but they were mentioned a couple of times in pure maths units and seemed interesting/crazy and I can't remember much about them.
961 22:28 -!- mode/#ipdf [+o matches] by OperServ
962 22:29 <@matches> I will look at them, the "just use a float but with more bits until you run out of memory" doesn't seem a very well thought out approach
963 22:29 <@matches> Maybe someone actually cares enough to research better ways
964 22:29 <@matches> I mean, something's wrong with an idea when you can write "1/3" and have a runtime exception (re: Java BigNumber)
965 22:30 <@matches> I plan to write some nasty things about Java based on Kahan's rants so as to gloss over the fact that it has BigNumber built into it and forstall the inevitable "Why didn't you use Java!"
966 22:31 < sulix> I think we want to truncate to whatever the most accurate you can see at the given zoom level.
967 22:31 <@matches> Yeah
968 22:32 <@matches> It annoys me that XML and HTML specs don't have a PDF version
969 22:32 < sulix> I've been printing to pdf for some of those things.
970 22:33 <@matches> None of the links to different sections would work though
971 22:35 <@matches> Specifications are thrilling
972 22:35 <@matches> "How to read this specification"
973 22:35 <@matches> "This specification should be read like all other specifications"
974 22:35 <@matches> Oh wait
975 22:35 <@matches> It actually gets better
976 22:35 <@matches> Is that...
977 22:35 <@matches> Humour!
978 22:36 <@matches> I thought that wasn't allowed!
979 22:36 <@matches> "First it should be read cover-to-cover, multiple times. Then, it should be read backwards at least once. Then it should be read by picking random sections from the contents list and following all the cross references"
980 22:37 <@matches> As much as I love the idea of reading 626 pages of specification backwards...
981 --- Day changed Tue May 06 2014
982 00:39 <@matches> http://szmoore.net/ipdf/sam/rate-my-litreview.py
983 00:39 <@matches> Tim will be thrilled!
984 00:40 <@matches> I'm probably going to really regret making that
985 00:41 <@matches> The regret is rising to the surface
986 00:41 <@matches> It's an example of an interactive document format
987 00:41 <@matches> The pixels or perish guy would approve
988 00:43 <@matches> So much regret
989 00:43 -!- mode/#ipdf [-o matches] by matches
990 23:37 -!- mode/#ipdf [+o matches] by OperServ
991 23:38 <@matches> So, I have this book called "Computer Graphics" that is pretty amazing
992 23:38 <@matches> Funnily enough it has all the algorithms in it
993 23:38 <@matches> For computer graphics
994 23:38 <@matches> Probably should have been reading it ages ago but I thought we were worrying more about precision
995 23:38 <@matches> I don't even know anymore
996 23:39 <@matches> Have to read 500pg textbook on graphics, 500pg textbook on floating point numbers, 500pg PDF/PostScript standards...
997 23:39 <@matches> Too many pages
998 23:39 <@matches> Anyway it has a section on Octrees
999 23:39 <@matches> But don't worry
1000 23:39 <@matches> It mentions Quad trees
1001 23:39 <@matches> It also actually mentions fractals
1002 23:39 <@matches> As a thing
1003 23:40 <@matches> In other news, extending the document we have at the moment to allow anything other than rectangles will be interesting
1004 23:41 <@matches> Did you do this on purpose? :P
1005 --- Day changed Wed May 07 2014
1006 12:32 <@matches> We need to do something about all the warnings generated by the magic OpenGL 3 stuff
1007 12:32 <@matches> It's making it hard to sport warnings that I actually care about
1008 12:33 <@matches> Like: You forgot to return a value in this function :S
1009 13:52 <@matches> Gah I did it again
1010 13:52 <@matches> auto is dangerous
1011 13:53 <@matches> Possibly because it's buggy
1012 13:53 <@matches> I can't actually see any compiler warnings at all
1013 17:39 <@matches> So you can almost maybe see a difference between beziers calculated using floats and doubles
1014 17:40 <@matches> If you squint
1015 17:40 <@matches> And view them on different monitors
1016 17:41 <@matches> Ah there we go
1017 17:41 <@matches> I successfully broke it
1018 17:42 <@matches> When you round to pixel positions it doesn't make a difference
1019 17:43 <@matches> But on the other hand if you calculate beziers using really big numbers they look wierd :P
1020 17:43 <@matches> That's important
1021 17:43 <@matches> Because if you have an arbitrary infinite document you might be at coordinate positions that are really big
1022 17:43 <@matches> Captain Obvious strikes again
1023 17:45 <@matches> I think I will make a video of a circle moving towards infinity
1024 17:45 <@matches> This probably won't help the literature view much but it's too tempting to resist
1025 17:46 <@matches> I gave up trying to deal with our document format so I currently just generate vector<pair<T, T> > and then map that to a bitmap :P
1026 --- Day changed Thu May 08 2014
1027 10:57 <@matches> So circle was a terrible example
1028 10:57 <@matches> It stays as a circle
1029 10:58 <@matches> Hopefully the first curve I tried wasn't just a bug
1030 12:31 <@matches> I think it was a bug
1031 13:35 <@matches> There are bezier things in ipdf/code/src/tests now
1032 13:57 < sulix> I am very confused.
1033 13:57 < sulix> It generates bitmaps of circles in varying shades of red?
1034 13:58 * sulix decides to actually read the code.
1035 14:03 < sulix> Ah: I see what it's doing now.
1036 14:16 <@matches> Your first comment was pretty accurate :S
1037 14:17 <@matches> I was trying to obtain some amazing animation of a circle that hopelessly collapsed or exploded or something
1038 14:19 < sulix> I'm slightly terrified that it's generating a vector of points.
1039 14:19 < sulix> On the other hand, it looks like the rest of the world is realising what a mistake making OpenGL 3 terrifying was.
1040 14:19 <@matches> Yeah that's a bit lazy
1041 14:19 <@matches> Although, is returning a vector optimisable?
1042 14:20 < sulix> In C++11 it is.
1043 14:20 < sulix> (But not C++98)
1044 14:20 <@matches> C++11 is growing on me
1045 14:20 < sulix> Yeah: I've found the same thing.
1046 14:21 < sulix> I'm a bit scared about what that means, though.
1047 14:21 < sulix> Maybe one day I'll think boost was a good thing, and then I'll truly be lost.
1048 14:21 <@matches> Hahaha
1049 14:21 <@matches> The lambdas remind me of Javascript :S
1050 14:22 < sulix> I think lambdas are one of those things that are good for you in moderation, but poisonous in large quantities.
1051 14:22 <@matches> Probably
1052 14:23 < sulix> I won't be around on Monday, btw, so I'll miss the meeting.
1053 14:23 < sulix> (Also I'm missing out on Codejam, sadly)
1054 14:23 <@matches> No!
1055 14:23 <@matches> (To both of those things)
1056 14:24 <@matches> Oh god that means I need to make double the progress
1057 14:25 < sulix> I spoke to Tim, he said that it might be worth cancelling the meeting altogether.
1058 14:26 <@matches> Haha
1059 14:26 <@matches> I'm not sure if I should take that as a good or bad sign
1060 14:27 <@matches> I'm not going great with the literature review
1061 14:28 < sulix> I rated a bit of it.
1062 14:28 < sulix> (Though there wasn't a submit button, and it took me a few goes to get through the Turing test)
1063 14:28 <@matches> Page 20?
1064 14:28 <@matches> Haha
1065 14:28 <@matches> The turing test defaults to the accepted answer though!
1066 14:28 <@matches> You have to actually change it to get it wrong
1067 14:29 < sulix> Yeah, I thought it was a trick question.
1068 14:29 <@matches> Maybe I'll give Tim a slightly less joke-worthy version of Rate My LitReview when the draft is done
1069 14:29 <@matches> ... when hell has frozen over...
1070 14:29 <@matches> Whichever comes first
1071 14:31 < sulix> You should read James Mickens' USENIX articles: mkdir -p vogl/vogl_build/bin/release64 && cd $_
1072 14:31 < sulix> cmake -DCMAKE_BUILD_TYPE=Release -DBUILD_X64=On ../../..
1073 14:31 < sulix> make -j 10
1074 14:31 < sulix> (Also I should paste the right thing in)
1075 14:31 < sulix> http://research.microsoft.com/en-us/people/mickens/
1076 14:32 <@matches> Ah yes, I will read those at some point
1077 14:32 * matches downloads the first one, telling himself he will not read it until working on the lit review
1078 14:33 * matches reads it anyway
1079 14:33 * sulix feels guilty about not doing lit review at the moment, too.
1080 14:40 <@matches> Have you heard of /read Computer Graphics by Hearn and Baker?
1081 14:40 <@matches> I have the ancient "In the dawn before OpenGL 1.0" version
1082 14:41 <@matches> There is a section mentioning "GL"
1083 14:41 <@matches> I don't think they use GPU anywhere, they call them "Display Rasterisers" or something
1084 14:42 <@matches> Display Processor
1085 14:43 <@matches> A Display Processing Unit is a Display Processor for a Random Scan (vector) display system
1086 14:43 <@matches> Input devices are interesting too
1087 14:44 <@matches> "Input Dials"
1088 14:44 <@matches> A box of variable resistors basically
1089 14:45 <@matches> Working with computers in the 70s/80s would have been interesting although probably just as horrible as it is today
1090 14:45 <@matches> But at least it would have been less "webby"
1091 14:46 <@matches> Where "How do I centre a div" is a deep philosophical question
1092 14:48 <@matches> I should probably obtain / read a newer edition of this if I want to put it in my lit review?
1093 14:48 <@matches> Although it is referrring to the same papers we had about Bressenheim
1094 15:14 <@matches> Oh my god they mention precision!
1095 --- Log closed Thu May 08 17:04:49 2014
1096 --- Log opened Thu May 08 18:56:39 2014
1098 18:56 -!- Irssi: #ipdf: Total of 2 nicks [0 ops, 0 halfops, 0 voices, 2 normal]
1099 18:56 -!- Irssi: Join to #ipdf was synced in 3 secs
1100 19:55 < matches> Bresenham made a tutorial on rasterisation
1101 19:56 -!- mode/#ipdf [+o matches] by OperServ
1102 19:56 <@matches> Bresenham has a nice tutorial about rasterising
1103 19:56 <@matches> Ties in with the "Since mankind came down from the trees" angle
1104 19:57 <@matches> "Needlepoint or counted cross-stitch, such as that popularised by the image on a box of Whitman Sampler chocolates..."
1105 19:57 <@matches> I'm not sure how to relate the rasterisation stuff to the precision stuff
1106 19:58 <@matches> You have to make a pretty big rounding error to end up in the wrong pixel
1107 19:59 <@matches> Hearn and Baker mention rounding errors as being one reason why you don't want to use floats and then round them when rasterising things
1108 19:59 <@matches> The (bigger) reason being that floating point operations are expensive
1109 20:00 <@matches> They say pixels could move from the line in the DDA algorithm for long lines... I wonder how long they mean... 1e38 pixels? :P
1110 20:02 <@matches> I want to actually make a drawing of things that look different due to rounding errors dammit!
1111 20:03 <@matches> None of this "here is a sentence or two of handwaving"
1112 20:04 <@matches> Maybe rounding errors were a problem on computers with 8 bit floats and terrible resolution displays :P
1113 20:04 <@matches> Except the worse your resolution the less you'd notice rounding errors
1114 20:04 <@matches> Well the more you'd notice them, the less they'd happen?
1115 20:04 <@matches> I don't know
1116 20:05 -!- mode/#ipdf [-o matches] by matches
1117 20:30 < sulix> These were recommended: http://www.amazon.com/Forman-S.-Acton/e/B001IYTXGY/?_encoding=UTF8&tag=mollrock-20&linkCode=ur2&qid=1358713701&camp=1789&sr=8-3&creative=390957
1118 20:30 < sulix> (By a blog post, which turned out not to be about floating point precision, but does mention it: http://mollyrocket.com/casey/stream_0009.html )
1119 21:21 < matches> I'll look at them at some point
1120 21:22 -!- mode/#ipdf [+o matches] by OperServ
1121 21:22 <@matches> I'll look at them at some point
1122 21:22 <@matches> I sort of got bogged down explaining how lines are drawn :S
1123 21:22 <@matches> Graphics is complicated dammit
1124 21:22 <@matches> I'll probably get more done if I just write about everything no matter how irrelevant
1125 21:23 <@matches> Sooner or later I'll actually write something important and the rest can be appendicised
1126 21:23 <@matches> Hopefully
1127 21:25 <@matches> Oh dear, I have located some primitive form of blog
1128 21:26 <@matches> Well, it talks about things, so I'll reference it...
1129 21:26 <@matches> "The good-looking textured light-sourced bouncy fun smart and stretchy page"
1130 21:26 <@matches> It's generally best to not go up a directory from the page you are looking at
1131 --- Day changed Sun May 18 2014
1132 15:47 <@matches> So things are due soonish
1134 20:09 <@matches> Are you alive, because I don't think I'm alive
1135 20:10 -!- Irssi: #ipdf: Total of 2 nicks [1 ops, 0 halfops, 0 voices, 1 normal]
1136 20:11 <@matches> I'm feeling pretty guilty about all that work I didn't do last week...
1137 20:47 * sulix agrees entirely.
1138 22:11 <@matches> I give up
1139 22:11 <@matches> I will just have to face disappointment O'clock tomorrow
1140 22:22 < sulix> I still have to put more work in before I feel I even deserve disappointment, I think... :/
1141 22:23 < sulix> I made the mistake of deleting the irrelevant bits from my lit review, and now there's almost nothing left!
1142 --- Day changed Mon May 19 2014
1143 08:03 < sulix> At the risk of causing terror: is your final Lit review due on Friday too?
1144 08:05 * sulix has discovered that his is, and is currently trying to work out how to type in the foetal position.
1145 09:59 <@matches> Yeah it's due on Friday
1146 09:59 <@matches> Today is not off to a good start
1147 10:00 <@matches> I missed an 8am lab but I can't blame the 2 buses I missed because I actually forgot about it
1148 10:00 <@matches> Then I missed the 2 buses
1149 10:00 < sulix> Ouch.
1150 10:00 <@matches> I think missing my tutorial this afternoon is a bad idea but on the other hand... lit review
1151 10:02 < sulix> I'm frantically writing irrelevant stuff now.
1152 10:03 < sulix> I'm a little concerned about my entire section on "Document Formats" basically only referencing some specs and "Pixels or Perish".
1153 10:03 <@matches> That's what I do
1154 10:03 < sulix> Whereas I have like 20 references for graphics papers that are almost totally irrelevant.
1155 10:04 <@matches> I have a reference for shading on a vector display
1156 10:04 <@matches> That's pretty irrelevant
1157 10:04 < sulix> I've started referencing data structures that are "kinda similar I guess" to the Quadtree, as well. :/
1158 10:05 <@matches> I ham fisted it in with the paper about embedding 3d figures in documents
1159 10:05 <@matches> Like "in the 70s even though people had algorithms for things they still drew the diagrams by hand!"
1160 10:05 < sulix> Peasants!
1161 10:06 <@matches> My section on number representations is pretty shite
1162 10:06 <@matches> I still haven't finished floats
1163 10:06 < sulix> Mine is currently under construction, as it were.
1164 10:06 <@matches> And the whole point is supposed to be that floats are not good enough
1165 10:07 < sulix> I've got a terrible thing about floats, but have started talking about arb. precision floats now.
1166 10:07 <@matches> All I know about the alternatives are that they are basically still floats
1167 10:07 <@matches> But with more bits...
1168 10:07 < sulix> I also have some maths I made up that is maybe wrong.
1169 10:07 <@matches> Will you be here in time for doom O'clock?
1170 10:08 <@matches> *cough* I meant eleven
1171 10:08 < sulix> We can but hope.
1172 10:08 < sulix> I'm sort-of counting on the meeting not happening until 11:20 or so. :/
1173 10:09 <@matches> I'll show him Rate My Lit Review
1174 10:09 <@matches> That'll distract attention from the lack of Lit Review
1175 10:09 <@matches> I'm also assuming no one is ever going to read these IRC logs
1176 10:09 < sulix> We can but hope.
1177 10:10 < sulix> I'm feeling dirty because I've just referenced the java standard library.
1178 10:10 < sulix> Out damnéd
1179 10:10 < sulix> java.math.BigDecimal.
1180 10:15 <@matches> I was going to reference that at some point
1181 10:16 <@matches> I feel like you've made more progress than I have been
1182 10:17 <@matches> Possibly the decision to sleep last night was a poor one
1183 10:26 < sulix> Having done the opposite: it wasn't.
1184 10:30 <@matches> Argh I just got the icedove reminder for the meeting
1185 10:32 <@matches> So I washed out my Guild (TM) Keep Cup (TM) this morning and now my tea tastes like detergent
1186 16:37 <@matches> We didn't get the random asking about photoshop in here
1187 16:37 <@matches> I feel left out
1188 22:49 <@matches> sjy in #ucc mentioned Knuth's Metafontbook in regards to rendering fonts at infinite precision
1189 22:49 <@matches> Yet another thing to get around to reading :S
1190 22:49 <@matches> Before Friday :S
1191 22:50 <@matches> Wednesday I guess if we meet Tim's deadline for a draft :P
1192 23:08 <@matches> I guess I was meaning to reference the TeX book at *some* point
1193 23:08 <@matches> So many things
1194 23:08 <@matches> Oh dear it's 11pm already
1195 23:08 <@matches> :-(
1196 23:09 <@matches> Ok, so
1197 23:09 <@matches> Metafont seems to basically be symbolic ways to define fonts
1198 23:10 <@matches> Also Beziers
1199 23:10 <@matches> All the beziers
1200 23:10 <@matches> Good
1201 23:11 <@matches> I was on the right track when I said Beziers were really important
1202 23:11 <@matches> Knuth is so detailed
1203 23:12 <@matches> Now I feel guilty about my cartesian line
1204 23:12 <@matches> That I didn't reference Rene Descartes La Geometrie 1637
1205 23:12 <@matches> I'm a monster
1206 23:15 <@matches> I think instead of thinking of things as either interpreted or dom-y like Hayes does, the best way is to think about "What is in the document (eg: DOM)" and "How it is drawn" and you can either have your format care about both things or just one but obviously the most useful formats allow for both
1207 23:16 <@matches> Actually scratch that
1208 23:16 <@matches> I don't know anymore
1209 23:16 <@matches> It's too philosophical
1210 23:16 * matches goes back to floating point nubers
1211 23:18 <@matches> So I wonder how much of PostScript is a shameless rip off of METAFONT
1212 23:18 <@matches> It's not like you have to give credit when you are trying to sell something
1213 --- Day changed Tue May 20 2014
1214 09:29 <@matches> So it's quite easy to do a
1215 09:29 <@matches> "fractal" in SVG/Javascript
1216 09:29 <@matches> Where easy does not necessarily mean it isn't horrifying
1217 09:30 <@matches> Aaand there goes all my RAM
1218 21:43 <@matches> Casually slipping in a footnote directing the reader to rabbitgame.net in my lit review...
1219 21:43 <@matches> So my webby documents section is probably my least shitly written, or maybe that's just because it has pointless pretty pictures in it
1220 21:44 <@matches> It is basically
1221 21:44 <@matches> "Here are the w3c standards"
1222 21:44 <@matches> "Here are some pretty pictures made with SVGs, can't you just see the DOM leaching out of them?"
1224 21:45 <@matches> I'm not sure how well I am treading the line between actually reviewing literature and just giving examples of things
1225 --- Day changed Wed May 21 2014
1226 12:09 <@matches> PDF is a mess of a "standard"
1227 12:09 <@matches> As are all useful things I suppose
1228 12:09 <@matches> As far as I can work out
1229 12:09 <@matches> It is not a DOM but a graph
1230 12:10 <@matches> However, it is also PostScript-y
1231 12:10 <@matches> But they deal with "interactivity"
1232 12:10 <@matches> By including XHTML
1233 12:10 <@matches> And having an "action dictionary" which is literally just a string of javascript
1234 12:11 <@matches> I just
1235 12:11 <@matches> Can't even begin to understand how it all works
1236 12:12 <@matches> But yeah, not really "Crippled Postscript" so much as "Everything including the kitchen sink except for a few bits of Postscript"
1237 12:13 <@matches> So the Postscript part of it is no longer turing complete, but I don't think you can pretend something in which you can stick arbitrary Javascript isn't turing complete :S
1238 12:13 <@matches> Oh and even though they have XHTML-ish stuff their Javascript API is totally different to W3Cs
1239 12:13 <@matches> Hooray
1240 12:14 <@matches> I suppose the fact that nothing except Adobe products seem to actually use Javascript/XHTML stuff is telling us something about this approach
1241 12:15 <@matches> I reckon the ideal standard
1242 12:15 <@matches> Would probably be the DOM but with the "we actually care about efficiency" parts of PDF
1243 12:18 <@matches> The interactivity of web pages combined with the actually professional looking type setting of PDF
1244 12:18 <@matches> Or just plain text files
1245 12:19 <@matches> Plain text files are an underapreciated Document Format
1246 12:23 <@matches> Ah, I think it sort of makes sense now
1247 12:23 <@matches> PDF uses what is essentially PostScript to construct this graph thing
1248 12:24 <@matches> And the graph thing can have elements in it that are just "Make this part of the graph the equivelant DOM from this XHTML"
1249 12:25 <@matches> And it can also have elements that are "Execute this Javascript to dynamically change this graph"
1250 12:25 <@matches> But the normal elements are just like PostScript as it would be sent to a printer to show the thing statically
1251 12:26 <@matches> So when it's rendered it is interpreting the Postscripty bits and when its being interacted with it is updating the Postscripty bits
1252 12:26 <@matches> I *think*
1253 12:26 <@matches> This is different from the webby standards which don't really specify how things are actually drawn
1254 12:27 <@matches> No wait it's not
1255 12:27 <@matches> Argh I don't know
1256 12:27 <@matches> You can't classify this shit
1257 12:27 <@matches> Document Goes In -> Pixels Come Out
1258 12:28 * matches despairs
1259 12:49 < sulix> You will find (slightly) less despair if you relegate javascript to the footnote where it belongs. :P
1260 13:07 < sulix> Hmm... the HTML 2 spec looks like it almost got properly IETF standardised. Might just reference that.
1261 13:12 < sulix> Oh, they obsoleted it and replaced it with a "Just look at w3c" standard...
1262 13:17 <@matches> But PDF isn't just flattened PostScript
1263 13:17 <@matches> It is like, everything
1264 13:17 <@matches> All merged into one horrifying standard
1265 13:18 <@matches> Oh well
1266 13:18 <@matches> I made my shape example in PostScript by removing the alpha
1267 13:18 <@matches> I'm not sure whether there's any point in including it as a figure
1268 13:18 <@matches> Most of the PostScript file is taken up by the header
1269 13:19 < sulix> Holy balls, I just looked up the CSS spec. There are like 200 of them.
1270 13:19 <@matches> Yeah I just used CSS2
1271 13:19 <@matches> The others are like
1272 13:19 <@matches> Colours
1273 13:19 <@matches> Or something
1274 13:20 < sulix> That's what I'm using, too.
1275 13:20 < sulix> The tired and tested "what gets top result on google" method of paper selection.
1276 13:20 <@matches> Closer examination reveals that most of the PostScript header is defining commands to be shorter :P
1277 13:20 <@matches> Amazing
1278 13:20 <@matches> Cairo probably needs to get referenced somewhere
1279 13:21 <@matches> If only so I have a way out of my Javascript in PDF section by saying that Cairo doesn't support it
1280 13:21 <@matches> I desperately need to escape the Javascript
1281 13:25 < sulix> I think the secret is to use the phrase "rendering model" wherever possible.
1282 13:51 <@matches> Dammit
1283 13:51 <@matches> So I have that wierd shape in both SVG and PostScript now
1284 13:51 <@matches> The SVG version fits beautifully and is wonderfully concise and you can see how SVG works
1285 13:51 <@matches> The PostScript version is just like, BLARGH WALL OF TEXT
1286 13:51 <@matches> ALSO WE DON'T HAVE ALPHA
1287 13:52 <@matches> So I'm not sure whether to cut just the PostScript one or both of them now :S
1288 13:53 <@matches> PDF looks distinctly not like it is just PostScript the more I think about it
1289 13:53 <@matches> It's like "We are using the same model as PostScript in that commands go in and pixels come out"
1290 13:54 <@matches> By that logic SVG is also the same
1291 13:54 <@matches> I think what I should do is just make an appendix
1292 13:54 <@matches> "A Shape in 20 Document Formats"
1293 13:56 <@matches> SVG really is the most concise compared to PS and PDF
1294 14:53 <@matches> Right I can simplify the god awful mess of PS a bit
1295 14:54 <@matches> I'm hoping I can just say "Here is the PS reference and here is some PostScript as you can see it is interpreted-ish"
1296 14:54 <@matches> Cairo appeared to draw each element backwards and reverse it after drawing it
1297 14:54 <@matches> It is stupid
1298 14:57 <@matches> Like, why bother doing definitions like m == moveto etc
1299 14:57 <@matches> If you're just going to stick pointless crap in
1300 14:57 <@matches> My document is half the size without using single letter definitions
1301 15:31 < sulix> Welp. The wrath of Tim is upon us...
1302 15:41 <@matches> I'm choosing to latch onto the "quite good" rather than "some way to go"
1303 15:45 <@matches> It sort of sounds like "Well at least you gave me a pdf file" :P
1304 15:48 < sulix> One day, all anyone will use are ipdf files...
1305 15:50 <@matches> Right, TeX is very different from PostScript I think
1306 15:50 <@matches> At least, pure tex
1307 15:50 < sulix> Also, holy mackerel, I might have just found a paper on precision in document formats...
1309 15:50 < sulix> It even quotes Kahan
1310 15:50 <@matches> :OOO
1311 15:50 < sulix> https://www.tug.org/TUGboat/tb28-3/tb90beebe.pdf
1312 15:50 <@matches> What is it
1313 15:50 <@matches> Emergency rewrite of entire lit review
1314 15:51 < sulix> It's a bit TeX specific, but still.
1315 15:51 <@matches> That's alright
1316 15:51 <@matches> It ties in amazingly with my decision to hamfist TeX and Metafont into the lit review
1317 15:52 <@matches> Although I'm not sure it is wise because it means I have to talk about fonts and things
1318 15:53 <@matches> I wonder if "Fonts are just bezier curves" is sufficient
1319 15:53 <@matches> They are always treated seperately to curved paths
1320 15:53 <@matches> Which is understandable because it's a bit inconvenient if you want text in a document to have to define the paths for each glyph
1321 15:54 <@matches> Anyway I'm glad my assertion that Beziers are the only curves we care about is proving true
1322 16:33 <@matches> Are you in a position to retrieve this "envelope"
1323 16:57 < sulix> Not tonight: I'm going to pick it up tomorrow morning.
1324 16:58 < sulix> And hopefully replace it with a sparkling, glorious review of literature.
1326 17:00 <@matches> I cannot concentrate now
1327 17:00 <@matches> Because I haven't read the comments, I could be doing everything wrong!
1328 17:02 <@matches> Admitedly I'm technically "working" right now
1329 17:27 < sulix> My "Document Format Taxonomy" is almost complete... Just need to add SVG.
1330 17:28 < sulix> (And close my eyes and assert that Microsoft Word documents are not actually documents or something)
1331 17:28 <@matches> I am jealous
1332 17:29 <@matches> I just added PostScript it's not particularly well written
1333 17:29 < sulix> (I don't have any pretty pictures or code, though)
1334 17:29 < sulix> I've discovered that, despite having totally different numbers for "implementation limits", the PostScript and PDF specs are (a) talking about the same data types and (b) lying.
1335 17:32 <@matches> Bahaha
1336 17:32 < sulix> Do you know where the SVG spec mentions precision?
1337 17:33 <@matches> Ah, I regret not noting the page number
1338 17:33 <@matches> But a text search should find it
1339 17:33 <@matches> It specifically says things
1340 17:33 <@matches> I am interested in whether or not Javascript is subject to the same requirements
1341 17:34 < sulix> All I've found is "must be correct within 1px at 1:1 zoom", and "It is suggested that viewers attempt to keep a high degree of accuracy when zooming".
1342 17:35 <@matches> There's something that is about IEEE floats
1343 17:35 < sulix> Aaah... and a "High-Quality Viewer" must support at least double precision on coordinate system transforms.
1344 17:35 < sulix> But "IEEE" does not show up in a search of the spec.
1345 17:36 <@matches> Ah right
1346 17:36 <@matches> My brain just inserts IEEE whenever I hear "single" or "double" now
1347 17:36 <@matches> "An IEEE Double Episode of MasterChef!"
1348 17:36 <@matches> (Which would probably be infinitely more exciting)
1349 17:37 < sulix> (Or would it be NaNly more exciting...? :P)
1350 17:38 <@matches> Speaking of "where things are" are we meant to reference page numbers in standards?
1351 17:38 <@matches> I guess I'll find out when I read Tim's comments
1352 17:39 <@matches> Excellent my lab finished 20 minutes early
1353 17:39 <@matches> And also 40 minutes later than the other demonstrators :S
1354 17:40 <@matches> Do you want me to pick up your comments and scan them and email them to you? :P
1355 17:40 < sulix> That'd be great.
1356 17:40 < sulix> Also probably depressing.
1357 17:41 < sulix> But great.
1358 17:41 <@matches> Alright, ETA Transperth + Scanner is probably broken O'clock
1359 17:42 < sulix> I'll savour the blissful ignorance.
1360 19:50 <@matches> I don't think scanning is worth it, I'll just spam the feedback into this channel
1361 19:50 <@matches> First up, David's Lit Review
1362 19:50 <@matches> There is either "Gool" or "Cool" or possibly "Good" written and underlined on the first page
1363 19:51 <@matches> The opening paragraph is "A little overdramatic?"
1364 19:51 <@matches> (Since it's a question, I'd like to voice a "No" opinion here)
1365 19:51 <@matches> The DOM in a footnote is not defined
1366 19:52 <@matches> Page 2
1367 19:52 <@matches> There is a tick
1368 19:52 <@matches> A question mark in regards to the hyphenated bit in the rendering paragraph
1369 19:53 <@matches> Say "avoid" instead of lack
1370 19:53 <@matches> Add what the "basic primitives" actually are
1371 19:53 <@matches> There appears to be an issue with hyphenated phrases the hyphens are circled
1372 19:53 <@matches> Another tick!
1373 19:54 <@matches> Oh, you have a $2^64 - 1$\footnote{} which is unfortunate because it looks like $2^64 - 1^2$
1374 19:54 <@matches> That footnote (probably others?) would work in the paragraph without being footnote
1375 19:54 <@matches> Fullstops go after \cite{}
1376 19:55 <@matches> A tick (in regards to the quadtree diagram)
1377 19:55 <@matches> The concluding comment
1378 19:56 <@matches> "OK, Much to do (underline) There doesn't seem to be much scholarly references used. You have enough, but you seem to cite them in the context of their contributions to standards as opposed to how they addressed a research question or open problem"
1379 19:57 <@matches> And (not even our references lists are safe!)
1380 19:57 < sulix> Oh dear.
1381 19:57 <@matches> Where referencing web pages, include the date retrieved
1382 19:57 <@matches> That's it
1383 19:57 <@matches> I shall move on to my own for completeness although you might not need to care
1384 19:58 < sulix> Phew, that's not quite as horrible as it could have been, I guess.
1385 19:58 <@matches> I also have "Good"
1386 19:58 <@matches> There are some "I didn't read this bit but it had words that seemed vaguely relevant" ticks in Chapter 1 (Introduction) and 2 (Proposal)
1387 19:59 <@matches> Sorry Tim if you read this
1388 19:59 <@matches> But when I mark lab reports for Physics that's usually where I put the ticks :P
1389 19:59 < sulix> (The secret comes out)
1390 20:00 <@matches> (In my defence I did spend two hours marking the reports this morning and I am paid for none of them, so...)
1391 20:01 <@matches> (It's the bits that I scribble all over that are where the marking gets done)
1392 20:01 <@matches> (I think I've covered myself in case the lawyers of any of my students read this channel now, so I will resume my story...)
1393 20:02 <@matches> Attention is called to the many glaring instances of [?] and "Refer to Section ?"
1395 20:02 <@matches> I should probably define a vector image before comparing it to a raster image
1396 20:02 <@matches> Incidentally my Fox looks amazing
1397 20:02 <@matches> On printed paper
1398 20:03 <@matches> (Tim didn't say that, that's just my modest opinion)
1399 20:03 <@matches> Ahem.
1400 20:03 < sulix> Can you see the difference between the vector and bitmap versions easily?
1401 20:04 <@matches> At the original scale there is, alas, a very slight fuzziness
1402 20:04 <@matches> But I reckon the markers will be old and blind
1403 20:04 <@matches> Hmm, I should either be more careful about what I say here or stop logging this channel...
1404 20:04 * sulix hopes they don't read that.
1405 20:04 <@matches> Sorry markers
1406 20:04 <@matches> I worship your power
1407 20:04 <@matches> Please do not smite me
1408 20:05 <@matches> The scaled up version is interesting
1409 20:06 <@matches> It looks a bit like your circle with the blocky non-anti-aliased bit but actually anti-aliased by the pdf viewer
1410 20:06 < sulix> I guess the scaling would be done by the printer's postscript RIP.
1411 20:06 <@matches> Yeah I guess
1412 20:07 < sulix> (Side note: I find the whole idea of Postscript interpreters being called RIPs somewhat fitting)
1413 20:07 <@matches> The PDF decides not to antialias it and converts it to Postscript and then the postscript interpreter adds its own antialiasing?
1414 20:07 <@matches> I don't know
1415 20:07 * sulix joins the "SVG is the least broken format" club.
1416 20:07 <@matches> It's very tempting to descend into footnote madness with this lit review
1417 20:07 <@matches> "By the way, this very document is an example of this thing!"
1418 20:07 <@matches> Etc
1419 20:08 <@matches> Moving on
1420 20:08 <@matches> The point of talking about vector displays at all is questioned (at least I think that's what the "Why?" refers to here)
1421 20:08 <@matches> Or it could be "Why is there yet another ?? in this paragraph" I guess
1422 20:09 <@matches> But probably the former
1423 20:09 <@matches> I do not have space to include Bresenham's algorithm
1424 20:09 <@matches> Oh boy, he's going to love what I did with the SVG and Postscript images...
1425 20:09 <@matches> But I am glad I do not have to actually explain Bresenham's algorithm because it's actually annoyingly detailed
1426 20:11 < sulix> All sane descriptions of Bresenham's algorithm end up being cascades of "By symmetry" anyway.
1427 20:11 <@matches> I need to actually find a reference that applied Wu/Bresenham directly to a non-straight line
1428 20:11 <@matches> You said Bresenham adapted his algorithm to circles but I don't think I'll bother unless someone adapted them to beziers
1429 20:12 <@matches> Bresenham's paper on rasterisation techniques basically says "Compute some points close enough together and then just connect them with straight lines"
1430 20:12 <@matches> But I think things might have advanced since the 1980s
1431 20:12 < sulix> Well, we can compute points that are closer together and draw more lines, I guess.
1432 20:13 <@matches> Next, Tim wants an example of a spline
1433 20:13 <@matches> (Oh boy have I got that covered)
1434 20:13 <@matches> My mathematics terminology on Beziers is not really great
1435 20:14 <@matches> Well it's right but confusing maybe
1436 20:14 <@matches> Or I just need to say "t is a trajectory parameter"
1437 20:14 <@matches> Haha
1438 20:15 <@matches> He found one of my "????" that is actually just me typing question marks and not a broken reference
1439 20:15 <@matches> The *entire* section on shading and compositing has a big question mark
1440 20:15 <@matches> Oh dear
1441 20:15 <@matches> I just finished writing the compositing bit
1442 20:16 <@matches> I hope the question mark means "Why isn't this written" and not "Why is this in here"
1443 20:16 <@matches> Because it is quite useful for an excuse to say PostScript can't do alpha
1444 20:17 <@matches> I need to refer to the IM (I really don't think that's a thing) and DOM when citing Hayes
1445 20:17 <@matches> "I don't think Turing Completeness is essential" (Big cross through the Crippled Interpreted Model)
1446 20:17 <@matches> Fair enough
1447 20:17 <@matches> A tick appears
1448 20:18 <@matches> Predictably in the web based documents part
1449 20:18 <@matches> I need to explain why Ipython is cool if I want to talk about it
1450 20:18 <@matches> My entire section on Precision as defined in the various formats is ?
1451 20:20 <@matches> My still to be completed/started section on Graphics APIs, GPUs and Arbitrary Precision is three question marks and "How's all this going"
1452 20:20 <@matches> The progress report gets a single tick
1453 20:20 <@matches> And the references have similar issues
1454 20:20 <@matches> Well
1455 20:21 <@matches> I'll take a few minutes to quiver in terror
1456 20:21 <@matches> But I think if I can just find a way to not sleep and still maintain productivity, I might be able to pull this off
1457 20:22 <@matches> Interestingly he didn't call me out for just talking about standards
1458 20:22 <@matches> But now I realise that's because I didn't have all the crap I've just written on standards in there
1459 20:23 < sulix> It's going to be a long night, but I think we'll manage it.
1460 20:23 <@matches> Mine will be too long but I don't care
1461 20:24 <@matches> I'll ask for an extension to prepare a condensed version if I must :P
1462 20:28 <@matches> It's kind of funny I've been spending more time making my vector image in SVG and PostScript nice than actually writing about either of those standards
1463 20:50 <@matches> Argh the idea of making my koch snowflake example for PS just got in my head
1464 20:50 <@matches> Which would be brilliant I guess if the topic was still "Fractal Document Formats"
1465 20:50 <@matches> It probably would be useful if I could demonstrate precision issues..
1467 20:50 <@matches> MUST WRITE
1468 20:50 <@matches> WORDS
1469 20:50 <@matches> NOT PICTURES
1470 20:51 <@matches> But still it would make the PostScript and SVG sections consistent with each other...
1472 20:51 <@matches> Must control urge to put pointless pictures in
1473 20:51 <@matches> No matter how much it seems like a good idea
1474 20:51 <@matches> And not pointless
1475 20:52 <@matches> Help I'm losing this battle
1476 20:53 <@matches> It is probably actually a better way of making a Koch curve than the hideous Javascript parsing of strings version
1477 22:50 <@matches> By the way, you can totally have "pre layout" stages in PostScript since you can define your own operators
1478 22:50 <@matches> Or do I misunderstand your sentence
1479 22:50 <@matches> Oh well it sounds smart anyway
1480 22:51 <@matches> In fact it's a lot more concise than my DOM-y section
1481 22:52 <@matches> I should sign my Lit Review as Captain Obvious
1482 22:57 < sulix> My current version does have a "PostScript programs typically embody documents which have been type-
1483 22:57 < sulix> set, though as a turing-complete language, some layout can be performed
1484 22:57 < sulix> " sentence.
1485 22:57 < sulix> by the document.
1486 23:01 * sulix is still a little bit concerned about how he should reference things for their solutions to open problems rather than their contributions to standards.
1487 23:03 <@matches> I think I am managing to do it
1488 23:03 <@matches> I will commit something at some point
1489 23:03 < sulix> I'm hoping that rewriting most of the rendering section with painful discussions of algorithms will do it.
1490 23:03 <@matches> An example is Porter and Duff Compositing
1491 23:04 <@matches> Because PostScript doesn't have alpha and I am really hoping that's just because Adobe had moved on to PDF by the time alpha was a thing
1492 23:04 <@matches> And not because they thought alpha was dumb :P
1493 23:05 <@matches> So I can relate Porter and Duff's model to the standards that do use it and say how it solves a problem that the standards that don't use it still have
1494 23:05 <@matches> And then I can sit back in satisfaction
1495 23:05 <@matches> And realise this says fuck all about precision
1496 23:06 <@matches> But at least by talking about it, I have eliminated it from the set of things we need to worry about when talking about precision :P
1497 23:06 < sulix> I've got a section which basically goes through all of the different document formats and looks at what their specs say about precision now.
1498 23:06 <@matches> Yeah I have that, but it was dot-pointed
1499 23:06 <@matches> I thought that would be OK actually but it has a question mark here... :S
1500 23:06 < sulix> Basically most of them say "implementation-defined" anyway.
1501 23:06 <@matches> Oh right because I was saying random stuff about how Postscript *used* to not have IEEE
1502 23:07 <@matches> Yeah it is odd that the standards don't actually reference IEEE
1503 23:07 <@matches> You'd think, since it's a standard...
1504 23:07 <@matches> Instead they just say "single" or "double" or "it might be single if you're lucky but we don't care really"
1505 23:08 <@matches> I assume "single" is widely accepted to mean IEEE single
1506 23:08 < sulix> From my reading of the Postscript spec, it says basically "We've put IEEE here, but ask your printer manufacturer because they could be using anything for all we care."
1507 23:08 <@matches> Ah I will check that more carefully
1508 23:08 <@matches> But it sounds about right
1509 23:08 < sulix> They give "typical limits" for their data types, but specifically do not specify what they are to be implemented as.
1510 23:08 <@matches> I don't think I have the time to look at what PostScript did historically before IEEE-754 although it would be kind of interesting to know
1511 23:09 <@matches> PostScript also does a bunch of silly maths because of units
1512 23:09 < sulix> The idea being that each postscript interpreter could do whatever they liked.
1513 23:09 <@matches> Cool
1514 23:09 <@matches> I should know this already :S
1515 23:09 <@matches> I just included a single character as a figure
1516 23:10 <@matches> But I want to actually work out how to do it in LaTeX by setting the size of the font appropriately
1517 23:10 < sulix> The PDF spec says pretty much the same thing, but notes that Adobe's implementation uses "Mostly IEEE singles" but "used to use 16.16 fixed point" and "still uses it for some things"
1518 23:10 <@matches> I did see that
1519 23:10 < sulix> TeX using 14.16 fixed point.
1520 23:10 < sulix> DVI uses "up-to 32bit" signed integers.
1521 23:11 <@matches> So basically no one actually uses IEEE for anything :P
1522 23:11 <@matches> Good work
1523 23:11 <@matches> I shall panic a bit and then try and actually do that work myself
1524 23:11 < sulix> SVG uses "implementation defined" or "double-precision floating point" "for coordinate transforms" if you want to be certified "High Quality"
1525 23:12 <@matches> I saw that one
1526 23:12 <@matches> But I'm skeptical about how this plays with Javascript
1527 23:12 <@matches> Not for High Quality even, just in general
1528 23:12 < sulix> Javascript numbers are always IEEE 754 doubles.
1529 23:13 <@matches> Ah thanks
1530 23:13 < sulix> (Even their integers are IEEE 754 doubles, which just happen to be integers)
1531 23:13 <@matches> Yes I have heard this before
1532 23:13 <@matches> From you probably :P
1533 23:14 < sulix> I don't have a source for that, and I'm not going to read the ECMAscript spec to find one, though.
1534 23:16 <@matches> Oh right, Javascript is actually ECMAscript
1535 23:16 <@matches> I forgot that
1536 23:17 <@matches> Dammit I am struggling to stay awake here
1537 23:17 <@matches> I'm not sure whether it's healthier to try to not sleep and give Tim a draft tomorrow and demand he read it in enough time to make last minute changes
1538 23:18 <@matches> Or sleep and then be more coherant tomorrow
1539 23:18 <@matches> I guess I'll try and finish a couple more sections
1540 23:19 < sulix> I'm going to try to finish this tonight.
1541 23:20 * sulix has another assignment due Friday that needs significant work.
1542 --- Day changed Thu May 22 2014
1543 00:47 <@matches> So X just managed to totally shit itself
1544 00:47 <@matches> Time to see when I last pressed Ctrl-S
1545 00:48 <@matches> Oh good (I typically press it once per sentence)
1546 00:48 <@matches> I hope it wasn't one of my SVGs that broke everything
1547 00:49 <@matches> Making all my figures in SVG
1548 00:49 <@matches> Lovingly hand written
1549 00:49 <@matches> I'm not sure that was a good idea
1550 01:43 <@matches> Heh, converting SVG to PS in Inkscape appears to introduce rounding errors of up to 0.1 of whatever unit PS is using
1551 01:44 <@matches> 1.25*56
1552 01:45 <@matches> Disregard I appear to have mistaken this terminal for a calculator
1553 01:45 <@matches> They do look similar at this hour in the morning
1554 01:49 < sulix> I'm seriously doubting whether starting to write stuff about how all of the rasterizing algorithms work was a good idea.
1555 01:49 <@matches> Haha
1556 01:49 <@matches> I don't really have much rasterising as such
1557 01:49 <@matches> Straight lines
1558 01:50 <@matches> Beziers aren't really "rasterising" so much as defining curves
1559 01:50 <@matches> Then I mention fonts and how they are all bezier-ish
1560 01:50 < sulix> I've just spent about an hour trying to prove that bresenham's algorithm can be made to not depend on any coordinates outside the screen.
1561 01:50 <@matches> Then I have a bit on shading that I can't do
1563 01:50 <@matches> That's probably not important
1564 01:50 <@matches> To quote Tim in my thesis regards to actually explaining algorithms
1565 01:50 < sulix> It turns out that if you clip the endpoints (even correctly), you adjust the slope of the lines slightly.
1566 01:51 <@matches> "No you don't have space. [...] details should be left to where they are useful"
1567 01:51 <@matches> Having said that absolutely none of this is "useful"
1568 01:51 < sulix> To fix it you have to initialize bresenham's "accumulated error" properly.
1569 01:51 <@matches> That sounds sort of relevant
1570 01:52 <@matches> Well I just copied some SVG into PostScript and discovered that the coordinates are all reflected :S
1571 01:52 < sulix> My choices for referencing this is either a textbook I don't have that "apparently mentions this" or a blog post which rants about it a bit.
1572 01:52 <@matches> Bahaha
1573 01:53 <@matches> That sounds like work for the final lit review rather than this one
1574 01:53 < sulix> Still need to work out how to actually work in how my references solved problems rather than contributed to standards or whatever.
1575 02:05 <@matches> minipage for latex is great by the way
1576 02:06 <@matches> Like <div> but not awful
1577 02:40 <@matches> Oh year
1578 02:40 <@matches> The best spline curve example ever
1579 02:41 <@matches> Fuck it took about 3 hours to do that
1580 02:41 <@matches> Dear god
1581 02:41 <@matches> Ergh remind me to go censor the log before Friday 1am
1582 02:50 <@matches> I pushed my references changes by the way
1583 02:51 < sulix> Yay merging!
1584 02:51 <@matches> Oh I forgot the actual pdfs but whatever
1585 02:51 <@matches> I am jealous of your actual concise and to the point lit review
1586 02:51 <@matches> Mine suddenly exploded into figures
1587 02:51 <@matches> I should stop
1588 02:51 <@matches> I think I will delete the Shading section
1589 02:52 <@matches> No wait it would be a gap to take it out now
1590 02:52 <@matches> Argh
1591 02:53 <@matches> I will just have to resist the urge to put a diagram in showing how shading works
1592 02:53 <@matches> All these diagrams will probably kill me
1593 02:53 < sulix> I'm thinking of scrapping chunks of the Rendering section of mine just so I don't have to finish them and can go to sleep.
1594 02:54 <@matches> Tim did seem strongly in favour of covering the rendering stuff
1595 02:54 <@matches> At least referencing the papers and giving the definitions of things if not actually how to render them
1596 02:55 <@matches> Anyway I think Beziers at least are important
1597 02:55 <@matches> I'm discovering a few interesting things about SVG
1598 02:56 <@matches> The path definitions are basically exactly the same as postscript's commands except less stack-y
1599 02:56 <@matches> But it has relative commands as well
1600 02:56 <@matches> Which is interesting because if you have a really really long curve defined with relative commands
1601 02:56 <@matches> Maybe, just maybe, it will actually cause a precision issue
1602 02:57 <@matches> I doubt it though
1603 03:00 <@matches> Well good luck, I am going to sleep
1604 03:01 < sulix> Thanks. I will see what I can say about Béziers.
1605 03:38 * sulix collapses.
1606 08:07 <@matches> Nice work
1607 08:07 <@matches> Mine is too detailed I think
1608 08:07 <@matches> It's horrible because now I'm committed to following through on that level of detail everywhere
1609 08:09 <@matches> Removing detail feels like murder
1610 08:12 <@matches> Would you be offended if I cited your lit review as a "more concise" overview for the bored reader? :P
1611 14:28 < sulix> So apparently the entire internet is talking about Bézier curves today.
1612 14:28 < sulix> This would have been really useful, say, yesterday.
1613 14:30 < sulix> Also this page looks amazing... http://pomax.github.io/bezierinfo/
1614 14:39 <@matches> Haha
1615 14:39 <@matches> I think I've got the Beziers covered
1616 14:39 <@matches> If you could just hop over to ratemylitreview and check me on that...
1618 14:42 < sulix> Ratemylitreview has broken some of the equations...
1619 14:46 <@matches> If I had time I would include a "rate ratemylitreview" field
1620 14:46 <@matches> I sent an email
1621 14:46 <@matches> Now to fear the wrath
1622 14:47 <@matches> Half time
1623 14:48 <@matches> Haha I'm somewhat regretting choosing such condescending ratings
1624 14:49 < sulix> I got terrified seeing that email before I realised it was from you.
1625 14:49 <@matches> Bahaha
1626 14:49 <@matches> Well I have to give him something
1627 14:50 < sulix> You should clearly make the ratings be all amazing, like: "Good, Great, Amazing, Truly Spiffing, Superlative and \"Everything in creation has been leading up to this page of my Lit review\""
1628 14:51 <@matches> You can POST your own ratings but expecting that might be a bit much
1629 14:51 <@matches> They are emailed to me as text and they are also stored in the database as text
1630 14:51 <@matches> Not the most objective of systems :S
1631 14:51 <@matches> I thought the ratings covered all the bases though
1632 14:52 < sulix> You could always do what the Shakespeare proramming lanugage does. Positive adjectives +1, negative adjectives -1.
1633 20:18 <@matches> I haven't said half of what I thought I should about floats
1634 20:18 <@matches> Tim has been scarily silent :P
1635 20:19 <@matches> I guess I will assume that means everything is FINE
1636 20:19 <@matches> I finally got my Table of Contents and things to not take up 6 pages
1637 20:20 <@matches> I have to resist the urge to add some snarky comments to my section on LaTeX
1638 20:22 <@matches> About how in theory you don't have to worry about where things go but in practice you spend hours doing horrible things like arbitrarily adding vertical space to force something into position because the anchor position doesn't take into account your line spacing and thus isn't where you expect and the next element overlaps things as a result
1639 20:22 <@matches> Blargh
1640 20:22 < sulix> Just stumbled upon a mention of numerical precision causing issues with Wu's algorithm in a book.
1641 20:22 <@matches> Cite it
1642 20:23 <@matches> Also give it to me to cite if I can stay awake long enough to read it
1643 20:23 <@matches> You have contributed some good last minute references
1644 20:23 <@matches> I seem to have contributed Javascript
1645 20:23 <@matches> I should be ashamed
1646 20:24 < sulix> It's in Abrash's Black Book, end of Chapter 42 ("Wu'ed in Haste; Fried, Stewed at leasure"(
1647 20:24 <@matches> Haha
1648 20:24 <@matches> My venerable graphics book doesn't even have Wu in it :S
1649 20:25 <@matches> It also doesn't really properly reference people's papers, or I guess you don't need to in a textbook?
1650 20:26 <@matches> It gives De Casteljau's method without crediting him for example. It does credit Bezier with things though.
1651 20:26 <@matches> But they clearly didn't have to worry about putting \cite{} after every single statement
1652 20:27 <@matches> I'm not sure how well my "I couldn't find a reference, have a look at my website" footnotes are going to go down
1653 20:27 < sulix> Casey from the Jeff & Casey show did a video blog on Bézier curves and interpolation this morning.
1654 20:28 <@matches> I really should start reading blogs more, but I was in no position to be watching blogs this morning
1655 20:28 < sulix> He claims that Bézier did not invent the Bézier curve, so why the hell are they named after him.
1656 20:28 <@matches> Had I known
1657 20:28 <@matches> Oh he didn't
1658 20:28 <@matches> I know that
1659 20:28 <@matches> That's in Metafont
1660 20:28 < sulix> I should read Metafont.
1661 20:28 <@matches> They were named after him because he was the first guy that said "These look useful for things"
1662 20:28 <@matches> However
1663 20:28 <@matches> I can't find the paper in a peer reviewed journal
1664 20:29 <@matches> I think it was for Industry (TM)
1665 20:29 <@matches> And also in French
1666 20:29 <@matches> I just cited a paper he did write in English about his experiences with Computer Aided Design
1667 20:29 <@matches> He doesn't define anything, just shows pictures of nice bezier curvey car bodies
1668 20:30 < sulix> I saw a bunch of french papers by him, but couldn't be bothered trying to work out which ones were useful.
1669 20:30 <@matches> De Casteljau's paper is also hard to find
1670 20:30 <@matches> The order of events was
1671 20:30 <@matches> 1912 - Bernshtein comes up with the basis polynomials for some sort of mathematical fitting
1672 20:31 <@matches> 1959 - De Casteljau decides to approximate the curves using his algorithm
1673 20:31 <@matches> (De Casteljau is only an approximation by the way, but it would converge to the true bezier curve)
1674 20:31 <@matches> 196? - Bezier does stuff
1675 20:31 <@matches> 1983/4 - Knuth decides to use them for fonts
1676 20:32 <@matches> Somehow they ended up in PostScript around that time as well
1677 20:32 <@matches> Now we're stuck with them
1678 20:34 <@matches> Hmm I hope my Tensor Equation is right there
1679 20:34 < sulix> Here's that Wu thing, btw: http://www.jagregory.com/abrash-black-book/#notes-on-wu-antialiasing
1680 20:36 <@matches> I'm struggling to make my floating point section sound sane
1681 20:36 <@matches> Too many little details
1682 20:37 <@matches> Like that you can choose different bases
1683 20:37 <@matches> What even is a number anyway
1684 20:38 <@matches> I think the proper way to approach it is talking about a number represented by digits and some numbers take infinitely many digits etc
1685 20:39 <@matches> Then computers can only fit X binary digits in their registers
1686 20:39 <@matches> A floating point is basically where you have a fixed point mantissa and then shift the location of the fixed point
1687 20:39 <@matches> Let's try and ignore the implicit leading one...
1688 20:40 <@matches> It sort of all falls apart when trying to fit IEEE in there
1689 20:40 <@matches> This thing is too big as well :(
1690 20:41 <@matches> Page limits are stupid
1691 20:42 <@matches> I can't remember what it even is but it's definitely less than what I have
1692 21:06 <@matches> Ah I see, the aliasing of Wu's line isn't perfect
1693 21:06 <@matches> I think Wu admits that
1694 21:06 <@matches> Hmm, that is interesting
1695 21:11 <@matches> Argh that blog is all like "We should use web based documents instead of PDF"
1696 21:11 <@matches> Pixels or Perish detected
1697 21:11 <@matches> To be fair it does actually look nice
1698 21:14 <@matches> The table of contents in the black book are quite amusing
1699 21:14 <@matches> No! I finished writing about graphics stuff I need to do floating stuff
1700 21:15 < sulix> It's a brilliant book, but possibly for another day.
1701 21:16 <@matches> Dammit I guess I do need to produce more figures
1702 21:16 <@matches> Sigh
1703 21:17 <@matches> A picture is worth a thousand words and all that
1704 21:17 <@matches> And therefore takes at least as long as writing a thousand words to make
1705 21:27 <@matches> The more I look at SVG files the more convinced I am that they are actually the write way to do things
1706 21:27 <@matches> right
1707 21:28 <@matches> Despite all that philosophical guff, you can do the same things as postscript in similar ways, but you also have a DOM that isn't terrifying like whatever PDF supposely does
1708 21:28 <@matches> I guess PDF would be more efficient though
1709 21:29 < sulix> That's pretty much the conclusion I've come to.
1710 21:31 <@matches> Unfortunately now I know more about SVG I keep hand editing my figures
1711 21:32 <@matches> I don't need to use gnuplot's terrible data point markers anymore!
1712 21:32 <@matches> I am free!
1713 21:40 <@matches> I particularly like
1714 21:40 <@matches> That I can make the points have alpha now
1715 21:40 <@matches> So if you plot overlapping points it is no longer impossible to see them
1716 21:40 <@matches> Of course we are restricted by the zoom in the pdf viewer...
1717 21:41 <@matches> This project is too meta
1718 21:41 <@matches> It is doing my head in
1719 --- Day changed Fri May 23 2014
1720 11:22 < sulix> Welp. Submitting this version: http://davidgow.net/stuff/LitReviewDavid.pdf
1721 11:23 < sulix> (The introduction has only got more over the top, I'm afraid)
1722 11:48 < sulix> Do you know what is happening/isn't happening RE: Revised project proposals?
1723 12:03 < sulix> Well: Literature review is submitted.
1724 12:03 < sulix> (In person to the coordinator, which is a little bit scary)
1725 --- Day changed Sun May 25 2014
1726 15:42 <@matches> No meeting tomorrow by the way
1727 15:43 < sulix> Ah. Cool. I have like 300 things to do.
1728 15:43 < sulix> Just pushed fixes for all of the compile warnings, btw.
1729 15:43 <@matches> Cool
1730 15:44 <@matches> I want to keep editing my Lit Review :S
1731 15:44 < sulix> I had thought that Float() always returns a "float", but it sometimes returns a double.
1733 15:44 <@matches> Whoops
1734 15:44 <@matches> Well a double is technically still a float...
1735 15:44 < sulix> (Also, it turns out OpenGL actualy breaks the C++ spec, and is therefore impossible to use without hacks if you have -Werror enabled)
1736 15:45 <@matches> Sigh
1737 15:45 < sulix> It was warning that I was losing precision from float x = Float(blah);
1738 15:45 <@matches> One of the things I want to put in my lit review is a snarky paragraph about how no one actually obeys standards anyway
1739 15:46 < sulix> There are points where you get function pointers as void* pointers, but C++ needs to work on systems where code and data are stored in different bits of memory with different size pointers.
1740 15:46 <@matches> On the other hand no matter how much better I make the lit review no one will read it because I'm being assessed on a conference paper not a dissertation
1742 15:46 < sulix> So casting any data pointer to a function pointer is apparently illegal.
1743 15:47 <@matches> That's annoying
1744 15:47 < sulix> Fortunately, gcc doesn't complain if you start the line that does it with "__extension__", so that's what we do.
1745 15:47 <@matches> Haha
1746 15:54 <@matches> Ok I am still about 3 days behind on sleep but I guess I should do work
1747 15:54 <@matches> Bye
1748 --- Day changed Tue May 27 2014
1749 12:49 <@matches> We missed the Computable Document Format (CDF) by Mathematica by the way
1750 12:49 <@matches> Wolfram would be offended
1751 12:50 * matches hopes none of the Mathematica fanatics read the lit review
1752 12:50 <@matches> Or this channel
1753 12:54 <@matches> I did have both Mathematica and IPython as a dot point and they got the mighty question mark of confusion over them
1754 12:54 <@matches> Maybe I'll add them in later
1755 12:55 <@matches> Frames does not seem to agree with my assertion that practically anything can be considered a document format :P
1756 12:55 <@matches> Plain text!
1757 12:57 <@matches> So I was looking through last year's Mech/Chem final year conference
1758 12:57 <@matches> Naturally there is nothing remotely like this project in there
1760 --- Log opened Tue Jun 10 13:59:54 2014
1762 13:59 -!- Irssi: #ipdf: Total of 2 nicks [1 ops, 0 halfops, 0 voices, 1 normal]
1763 13:59 -!- Irssi: Join to #ipdf was synced in 3 secs
1764 16:29 -!- mode/#ipdf [+o matches] by OperServ
1765 --- Day changed Wed Jun 11 2014
1766 13:13 <@matches> JVB just asked me if it was possible to have a self updating pdf
1767 13:13 <@matches> I suppose he didn't ask whether it would be easy
1768 13:15 <@matches> As in, one that would download itself if there was a newer version
1769 13:16 <@matches> Technically with the crippled PostScript it would be "No" but with all that stuff in the standard about DRM and "Action Objects" and Javascript and stuff...
1770 13:18 <@matches> Hmm, a webserver running git and a cronjob and requiring authentication may solve his problem
1771 13:18 <@matches> With pdf.js or just relying on browsers to have pdf plugins
1772 13:18 <@matches> Wait why am I solving JVB's problem
1773 --- Day changed Fri Jun 13 2014
1774 16:31 <@matches> So exams have finished
1775 16:31 <@matches> Well my exams have finished
1776 16:32 <@matches> Now I can go back to panicking about the project again
1777 16:37 <@sulix> Excellent: it's your turn to do something impressive for the meeting on Monday, then. :P
1778 --- Day changed Mon Jun 16 2014
1779 14:49 <@matches> Some of this OpenGL stuff doesn't quite add up
1780 14:49 <@matches> Like m_cahed_display.UnBind(); m_cached_display.Blit();
1782 14:50 <@matches> I see
1783 14:50 <@matches> There is a difference
1784 14:51 * matches -> OpenGL documentation
1785 14:53 <@matches> Right I see
1786 15:03 <@matches> Well, this makes a lot more sense than tpg's "Adventures in VEMS" in #ucc at least
1787 15:18 <@sulix> Well that's an achievement, I guess.
1788 15:18 <@sulix> Also http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2007/n2431.pdf
1789 15:18 <@sulix> It's the spec for "nullptr"...
1790 15:20 <@matches> "Pointers are pretty cool. Pointers that are NULL are pretty cool. We address these issues by proposing nullptr" ?
1791 15:21 <@sulix> Haha: Pretty much.
1792 15:22 <@matches> Why not just reserve the word "null"
1793 15:22 <@matches> I could get behind it then
1794 15:23 <@matches> It is not a null reference
1795 15:23 <@matches> But if you ever have a null reference something is horrifyingly wrong
1796 15:23 <@matches> Bah humbug
1797 15:24 <@sulix> They have a page about that where they said "everyone was naming their variables \"null\"".
1798 15:24 * sulix realises he just escaped those quote. Oh dear.
1799 15:24 <@matches> A valid point, which should have lead to "Everyone are morons"
1800 15:24 <@matches> Fun fact
1801 15:25 <@matches> Matlab (like fortran actually) will let you use key words as variable names.
1802 15:25 <@sulix> "It's a feature!"
1803 15:25 <@matches> Question on CITS2401 exam: "Why is 'ii' preferred over 'i' and 'jj' over 'j' for loop iteration?"
1804 15:26 <@matches> Both i and j are the complex number sqrt(-1)
1805 15:27 <@matches> It seems like naming it 'nullptr' instead of 'null' missed an opportunity to stop people naming their variables really confusing things :P
1806 15:30 <@matches> Well I am infinitely wiser about nullptr now but I don't think it will stop me just using NULL
1807 16:15 <@matches> This is slowly starting to make sense
1808 16:35 <@matches> I'm having a bit of difficulty because it's all so optimised for the GPU
1809 16:35 <@matches> I think what we need is an ObjectRenderer class or similar
1810 16:36 <@matches> Have one for each type of object and implement RenderUsingGPU and RenderUsingCPU
1811 16:36 <@matches> I will see what monstrosity I can come up with
1812 16:36 <@matches> But a lot of that code for the circles and rectangles is exactly the same just copied and pasted with different variable names
1813 16:37 <@matches> I think there is an official software engineering term for that sort of thing...
1814 16:46 <@matches> Alright, breaking everything
1815 16:46 <@matches> This will probably take a while
1816 16:47 * matches dreads merges
1817 17:02 <@sulix> Yeah, I was going to basically end up with an ObjectRenderer class, too.
1818 17:02 <@sulix> So I'll leave it to you for now and procrasinate from doing my maths in some other way.
1819 17:07 <@matches> Cool :)
1820 17:52 <@matches> Now I'm not very strict about the whole RAII thing myself, but there seem to be some cases where things can probably just go in the constructor?
1821 17:52 <@matches> I guess we can fight about those later
1822 17:54 <@matches> I guess I can see us wanting to set m_render_inited to false and have it automatically recreate everything, maybe, but I can't see a case where we will want to recompile the shaders, so I am going to put the shader initialisation in the constructors
1823 17:54 <@matches> I have gone mad with power
1824 17:54 <@matches> Maaad
1825 20:30 <@matches> This ObjectRenderer is like a two headed hydra
1826 20:31 <@matches> I thought about using just the ibo and vbo and getting the indices out of it, but that's a bit terrible
1827 20:31 <@matches> So it will have an ibo for the GPU and an array for the CPU
1828 20:35 * sulix thinks this makes sense.
1829 21:47 <@matches> Good news! I have things on the screen. Bad news. They are still on the GPU and they are wrong.
1830 21:49 <@matches> More coffee required...
1831 21:58 <@matches> Not sure if code wrong or fglrx having an existential crisis
1832 22:04 <@matches> Unrelated to the segfaults of doom, but I just realised that there's all this interest in using GPUs for scientific computation, and everything seems to point to them having vastly inferior floating point computation to a CPU, which is a bit disconcerting
1833 22:05 <@matches> I imagine the OpenCL libraries etc actually use the IEEE floating point capabilities? I hope?
1834 22:05 * matches files under "beyond the scope of project"
1835 22:06 <@sulix> If you commit code I can test it on non-fglrx if that'll help.
1836 22:06 <@matches> I'd rather not commit it just yet, I am 90% sure it is something I have done wrong and don't want to shame myself by committing broken code
1837 22:06 <@matches> It'd be like vomiting all over the codebase
1838 22:07 * sulix is pretty sure half of his commits broke something.
1839 22:29 <@matches> Ok I committed stuff that should in theory work with the test pattern again
1840 22:30 <@matches> Using new and improved ObjectRenderer classes
1841 22:30 <@matches> For some definition of improved
1842 22:30 <@matches> It doesn't actually use the CPU yet
1843 22:30 <@matches> It'll get there
1844 22:33 <@matches> You can go and add your MandleBrotSetRenderer now though :P
1845 22:36 <@sulix> Cool.
1846 22:37 <@sulix> The circles do trigger that intel driver bug again, though...
1847 22:38 <@matches> Ah damn
1848 22:38 <@matches> Would making CIRCLE_FILLED the first ObjectType fix that?
1849 22:38 <@sulix> It does.
1850 22:38 <@sulix> (I just checked)
1851 22:38 <@sulix> Yay intel!
1852 22:39 <@matches> Haha
1853 22:49 <@matches> Hey, does our rendering approach completely break compositing?
1854 22:49 <@matches> Render all circles
1855 22:49 <@matches> Then render all rectangles...
1856 22:49 <@matches> There's no depth...
1857 22:49 <@sulix> Um, maybe.
1859 22:49 <@matches> I was just worrying about how to make compositing work using the CPU
1860 22:49 <@matches> and then I realised it won't work using the GPU as it is either
1861 22:50 <@sulix> If we're not doing alpha-blending, we could just add a z-coordinate and use the z-buffer.
1862 22:50 <@sulix> Otherwise, they need to be sorted if they intersect.
1863 22:50 <@matches> Let's not worry about this for now...
1864 22:50 <@matches> Even though I spent a rather significant part of my lit review explaining how awesome compositing is
1865 22:51 <@matches> I guess it doesn't matter, clearly the lit review was not intended for anyone to actually read
1866 23:42 <@matches> I suspect RenderPixels doesn't work
1867 23:45 <@sulix> This may be the case.
1868 --- Day changed Tue Jun 17 2014
1869 01:05 <@matches> Behold the magnificence of CPU based rendering
1870 01:06 <@matches> In which we pass a massive pixel array to the ObjectRenderer's and just save it to a BMP
1871 01:06 <@matches> Because OpenGL
1872 01:06 <@matches> I'll probably fix it later
1873 01:11 <@matches> Whoops seg faults ahoy
1874 01:12 <@matches> Wait I swear it wasn't segfaulting before I commited that
1875 01:19 <@matches> It segfaults sometimes depending on where you have the view. If you just have one circle and zoom in you can see the amazingly less jagged looking lines.
1876 01:19 <@matches> That, is a certified result
1877 10:13 <@sulix> I got the CPU rendered version to show up on the screen.
1878 10:14 <@sulix> RenderPixels was pretty broken.
1879 10:38 <@matches> Nice
1880 10:39 <@sulix> I think the segfaults are it trying to draw outside of the display.
1881 10:39 <@matches> Yeah
1882 10:39 <@sulix> I haven't managed to zoom in too much yet.
1883 10:39 <@matches> My clipping is probably wrongish
1884 10:40 <@matches> fonttex_frag.glsl? Fear
1885 10:40 <@sulix> Yeah, the "basictex_frag" file did a bunch of pretty font-specific things.
1887 10:43 <@sulix> The CPU renderer also shows some random lines when there are no outlines on the screen.
1888 10:43 <@matches> Is it just me or does it look slightly nicer on the CPU when zoomed out a bit
1889 10:44 <@matches> Well it looks different
1890 10:44 <@matches> I can't quite work out whether it looks nicer or worse :P
1891 10:44 <@sulix> Okay, it doesn't segfault if the rectangle outlines are disabled.
1893 10:45 <@matches> Yeah it is setting top/bottom/left/right outside the pixels
1894 10:45 <@sulix> I think there are just some slightly different edge-case rules when zoomed out.
1895 10:47 <@matches> With my skills at writing renderers I should be working on the fglrx drivers
1897 10:47 <@sulix> Things vanish if you zoom in too much, too.
1898 10:48 <@matches> Blame it on floats
1899 10:49 <@matches> The CPU renderer is using a width of 1 pixel less than it should be
1900 10:49 <@sulix> Why are the floats sinking? :P
1901 10:49 <@matches> That's unrelated
1902 10:49 <@matches> It's probably an integer rounding issue somewhere
1903 10:49 <@matches> When the view gets big something goes to zero
1904 10:50 <@matches> The CPU renderer transforms everything to pixel positions and then renders using those
1905 11:02 <@matches> The GPU renderer occassionally leaves off the corners of things
1906 11:02 <@matches> I never noticed before
1907 11:02 <@matches> Anyway I think the segfaults are fixed
1908 11:03 <@matches> Looking at the mysterious disappearances next
1909 11:04 <@matches> It's integer overflow!
1910 11:04 <@matches> Facepalm
1911 11:05 <@sulix> Is it int64_t time?
1912 11:05 <@sulix> Or magic huge bigint time?
1913 11:05 <@matches> No I don't think that's necessary
1914 11:06 <@matches> If the transformed width is smaller or larger than the width of the screen you can surely just ignore it
1915 11:06 <@matches> Well, depending on what you are drawing
1916 11:08 <@matches> Well it does matter unless you do some clever maths
1917 11:08 <@matches> The clever maths was your part of the project...
1918 11:09 <@matches> So I will just make them int64_t for now
1919 11:50 <@matches> Pushe
1921 11:50 <@matches> I should probably just make all the coorinates into Real instead of int64_t actually
1922 12:03 <@sulix> Wow: this is pretty awesome.
1923 12:03 <@sulix> I still get the artefacts, but I have to zoom in a lot more.
1924 12:03 <@matches> Haha
1925 12:04 <@sulix> Also, the CPU rendering artefacts and the GPU rendering artefacts are pretty similar on my machine, so it looks like it's just precision causing them.
1926 12:04 <@matches> That is nice
1927 15:20 <@matches> I'm trying to do a performance graph. Gnuplot doesn't like it much
1928 15:20 <@matches> Well actually it's python's crappy interface to gnuplot that doesn't like it
1929 15:23 <@matches> ... tempted to implement the performance graph in ipdf...
1930 15:23 <@matches> Just keep adding circles to the document
1931 15:24 <@matches> So meta
1932 15:25 <@matches> We need a cool performance graph-y thing
1933 15:26 <@matches> I think doing it in OpenGL is going to be the least shitty way actually
1934 15:26 <@matches> This is a wheel that's probably been invented but it was invented wrong
1935 17:11 <@matches> I've got a performance graph sort of working
1936 17:11 <@matches> It almost looks like we are doing real science!
1937 17:13 <@matches> I think we'll need to put some effort into our data analysis though because it's extremely noisy
1938 17:13 <@matches> Smoothing averages or something
1939 17:13 <@matches> Smoothing averages are the best
1940 17:14 <@matches> They make any data look amazing
1941 17:15 <@matches> Anyway, CPU rendering is only worse than GPU rendering when you force re-rendering
1942 17:16 <@matches> So well done with the amazingly efficient cached frame buffer
1943 17:18 <@matches> We can probably make it draw every frame both on the CPU and GPU to compare them in real time
1944 17:18 <@matches> The possibilities are limitless!
1945 17:19 <@matches> Graphs graphs graphs!
1946 17:20 <@matches> Also I did end up using Gnuplot and python (sorry) but I made it slightly less shitty
1947 17:27 <@matches> I have pushed things
1948 17:27 <@matches> Also we probably don't need all three of those ways to measure performance
1949 21:00 <@sulix> I got the graphs working on my laptop: very nice.
1950 21:00 <@sulix> I can see how more GPU time is used with GPU rendering and more CPU time with CPU rendering quite well, actually.
1951 21:59 <@matches> So the objects all being in one structure of arrays is sort of inconvenient because the size of objects has to be constant
1952 21:59 <@matches> Also the size of Real has to be constant
1953 22:03 <@matches> I guess we could have union {Rect rect, Bezier bezier} and do rectangles and Beziers in the same thing but that is slightly terrifying
1954 22:04 <@matches> But there's still the problem of Real because as soon as Real becomes arbitrary precision it will start allocating memory and not be fwrite/fread'able
1956 22:05 <@matches> You'll still want the struct of arrays because that will make view reparenting much easier
1957 22:05 <@matches> Gah
1958 22:06 <@sulix> I'm tempted to just split it by type: Have an array of Rects, an array of Béziers, an array of Circles, etc.
1959 22:06 <@matches> Yeah but you need to store an "indexes" array as well
1960 22:06 <@matches> But that might be the least terrible
1961 22:07 <@matches> Yes that should work
1962 22:07 <@matches> Our fwrite/fread is still doomed though
1963 22:07 <@matches> When shit gets Real
1964 22:07 <@sulix> Yeah, you'd need some indices one way or another.
1965 22:08 <@sulix> And yeah, we'll need a massive "ConvertWhateverRealIsStoredInTheFileToWhateverRealIsDefined" function or something.
1966 22:08 <@matches> That will totally be its name
1967 22:08 <@sulix> (Some old compilers had function name limits... I wonder if modern gcc has)
1968 22:09 <@sulix> (There is but one way to find out!)
1969 22:09 <@matches> We don't need an array of circles/ellipses since we get those for free with Rect. Unless you want circular arcs as well as bezier curves
1970 22:10 <@matches> Which you probably will? So you can approximate beziers
1971 22:10 <@matches> Anyway I'll worry about adding Beziers first and once we've worked that out others should hopefully be easier
1972 22:11 <@matches> I was sort of thinking it would be good to be able to define groups of objects as a special object type
1973 22:11 <@matches> Then you can make paths out of your beziers
1974 22:11 <@matches> But they won't be fixed size
1975 22:12 <@matches> Anyway I will see if you have magically solved the problem for me in the morning :P
1976 22:14 <@sulix> I wouldn't count on it...
1977 --- Day changed Wed Jun 18 2014
1978 16:50 <@matches> Behold, the glory of Beziers and Bresenham!
1979 16:50 <@matches> Breseniers I shall call them
1980 16:54 <@matches> I'm getting a very strong vibe of "Reinventing graphics technologies from the 1990s" at the moment
1981 16:55 <@matches> I should probably look at floating point some more
1982 16:55 <@matches> And reinvent numerical computation technologies from the 1980s
1983 16:56 <@matches> We need a GPU renderer for the Beziers, and we need to fix save/load to not assume everything is a fixed size, and ...
1984 16:56 <@matches> Argh
1985 16:56 <@matches> All the things
1986 17:00 <@matches> Maybe I'll multithread the CPU rendering too, just to make things more difficult :P
1987 22:48 <@sulix> So I got some béziers rendering on the GPU.
1988 22:48 <@sulix> They're not the ones in the document, and they're not in the right spots, though.
1989 23:12 <@sulix> First 3 béziers are rendering properly on the GPU, then we get corrupted memory or something.
1990 23:14 <@sulix> Ah: I see, I need some way of uploading which bézier IDs are being used.
1991 23:41 <@sulix> Okay, Béziers now render on the GPU as well.
1992 23:43 <@sulix> The code is a little bit ugly, and for that I am sorry, but blame the fact that the GL feature that makes this nice is only about a year old and so nothing supports it.
1993 23:43 <@sulix> So we're basically uploading all of the raw document data into a huge texture.
1994 23:47 <@sulix> (If you ask me, the GPU ones also look slightly nicer, though that's probably a bug)
1995 --- Day changed Thu Jun 19 2014
1996 13:55 <@matches> From those papers I was under the impression just rendering a bezier on a GPU was an impressive feat, so well done :P
1997 13:57 <@sulix> (It's basically just a direct port of your CPU implementation, tbh)
1998 13:59 <@matches> When I zoom out the GPU ones look nicer, there's probably something wrong with my Bresenham implementation
1999 13:59 <@matches> Although it was mostly shamelessly copied from "Computer Graphics"
2000 13:59 <@matches> Except they were like "Here it is for 0 < m < 1, the rest can be done by symmetry"
2001 14:02 <@matches> The lines definitely have gaps on the CPU, and they also seem too thick as you zoom out
2002 14:02 <@matches> Having said that, I don't think the GPU starts at x0,y0
2003 14:03 <@matches> Anyway I'm going to implement a Rational number type and see if anything exciting happens (that's how you do research right?)
2004 14:06 <@matches> Fortunately I sort of did most of this for one of the codejam problems (where it turned out I didn't really need Rational numbers but I thought I did)
2005 14:20 <@sulix> Fixed the missing bit off the end.
2006 16:43 <@matches> I suspect Rationals are either not a very good idea, or there is a bug in one of my fundamental operations
2007 16:43 <@matches> +, -, *, / are hard
2008 16:56 <@sulix> Is it at least a pretty bug?
2009 16:59 <@matches> Um...
2011 17:00 <@matches> It seems buggy for anything other than the {0,0,1,1} starting view
2012 17:14 <@matches> I suspect it's the expf in the mouse wheel scrolling
2013 17:14 <@matches> Since you know, exp is definitely not a rational number...
2014 17:15 <@matches> Hmm, but it shouldn't matter because it will just convert to the nearest rational number
2015 17:16 <@matches> ie: p = (int64_t)(whatever*1e10) q = (int64_t)1e10
2018 17:28 <@matches> Head -> Desk
2019 17:28 <@matches> Rational & operator*=(const Rational & r) {this->operator=(*this*r); return *this;}
2020 17:28 <@matches> Rational & operator/=(const Rational & r) {this->operator=(*this*r); return *this;}
2021 17:29 <@matches> Rational & operator-=(const Rational & r) {this->operator=(*this+r); return *this;}
2022 17:29 <@matches> I think the worst part is that I actually said "It is probably a bug in my +,-,*,/
2023 17:29 <@matches> And it still took me this long to notice
2024 17:30 <@matches> The second worst thing is I've made that sort of mistake like 1000 times before
2025 17:30 <@matches> The third worst thing is I am recalling that article where the guy says "At least plus and times are sort of the same thing"
2026 17:30 <@sulix> Feel the power of copy and paste flowing through you.
2027 17:32 <@matches> Well let's not celebrate just yet, the view still goes to shit. Just slightly slower :P
2028 17:33 -!- matches changed the topic of #ipdf to: NaNpdf
2029 17:33 <@matches> Our document supports a view of {-inf,-inf,nan,nan} thus making it truly infinite precision
2030 17:34 <@sulix> I had that happen a lot when I was writing the original zoom code.
2031 18:19 <@matches> So I suspect that Rationals are just a really shitty number representation :P
2032 18:20 <@matches> Specifically, you get integer overflows really really fast
2033 18:20 <@matches> And if you are going to have a Rational that's BigInt / BigInt you may as well just have a BigFloat
2034 18:21 <@matches> The ancients were probably right.
2035 18:21 <@matches> When they decided not to use rationals.
2036 18:23 <@matches> I guess floats are rationals technically, I mean the representation using P/Q
2037 18:23 <@matches> I kind of wish I'd done some pure maths here...
2038 18:23 <@matches> Or paid more attention in second year
2039 18:28 <@matches> At least I sort of have conclusive evidence that rationals suck. As opposed to "it should be obvious to anyone with half a brain"
2040 19:00 <@sulix> Floats are not rationals.
2041 19:00 <@sulix> Not exactly.
2042 19:01 <@sulix> Something which can be stored in a finite amount of space as a rational cannot always be stored in a finite amount of space as a float, but not vice-versa.
2043 19:01 <@sulix> e.g.: 1/3
2044 19:04 <@sulix> Basically floats = rationals where the denominator must be a power of two.
2045 19:05 <@sulix> (Of course, these are all the same in the limit, but the limit of a cauchy sequence of rationals gives the reals, so the point is kinda moot, there, anyway)
2046 19:18 <@matches> Yeah floats are a subset of the rationals I think I meant
2047 --- Day changed Sat Jun 21 2014
2048 16:22 <@matches> So I got a new smartphone today... I'm having fun trying to zoom in on things
2049 16:24 <@matches> Actually the zooming isn't annoying so much as the needing to pan around to view all the text in a pdf
2050 16:25 <@matches> This is what I get for buying the cheapest phone with the lowest resolution they had :P
2051 16:26 <@matches> I didn't want to get one more expensive than my old one because I'm still convinced I will get around to fixing it one of these days
2052 16:26 <@matches> Or months. Or years. But eventually.
2053 --- Day changed Mon Jun 23 2014
2054 11:51 <@matches> The meeting happened at ~11:30
2055 12:49 <@sulix> Oh damn. I thought there wasn't going to be one today.
2056 12:49 <@sulix> What did I miss?
2057 13:04 <@matches> Basically I told him what we'd done
2058 13:05 <@matches> Which was rendering on CPU or GPU, and Beziers (but only degree 2)
2059 13:06 <@matches> And I implemented Rationals but they are terrible, so he sent me some code from an ACM competition about approximating things with Rationals
2060 13:13 <@sulix> Some javascript guy is trying to convince people that the only numeric data type should be some custom 64-bit decimal thing.
2061 13:13 <@sulix> A lot of people have started quoting Kahan at him.
2062 13:20 <@matches> Haha
2063 --- Log closed Tue Jun 24 16:07:29 2014
2064 --- Log opened Tue Jun 24 16:17:39 2014
2066 16:17 -!- Irssi: #ipdf: Total of 2 nicks [1 ops, 0 halfops, 0 voices, 1 normal]
2067 16:17 -!- Irssi: Join to #ipdf was synced in 15 secs
2068 --- Day changed Wed Jun 25 2014
2069 21:41 < matches> So... if we need to put a HTTP server in IPDF I've got us covered...
2070 21:44 <@sulix> You terrify me.
2071 21:45 < matches> I still have to work out Websockets
2072 21:45 < matches> They are silly
2073 21:46 <@sulix> I take it they aren't compatible with regular sockets.
2074 21:46 < matches> Well you use a TCP socket, but they aren't just plain text
2075 21:46 < matches> There's a protocol on top of it
2076 21:46 < matches> It's wired
2077 21:47 < matches> *wierd
2078 21:47 < matches> There are frames and things
2079 21:47 < matches> And copious amounts of base64 encoding the SHA1 hash of a string prepended to a magic string
2080 21:48 < matches> Apparently this makes it more secure
2081 21:48 * sulix looks doubtful.
2082 21:48 <@sulix> I could understand it being more secure if they actually encrypted things.
2083 21:48 < matches> They have wss:// as well which is just the same thing through ssl
2084 21:48 < matches> Yes it doesn't make sense
2085 21:49 <@sulix> But plain "boring" TCP has been used since (I am assured) the beginning of time and there are still one or two things that use it which are not currently being hacked into.
2086 21:49 < matches> Haha
2087 21:49 < matches> I'd like websockets so I can control my robot via my phone through its web browser...
2088 21:50 < matches> I can probably do that without websockets but it will also be useful for humphrey the rabbit
2089 21:50 <@sulix> Surely with plain old TCP you could control your robot via telnet, which is more awesome and just as user friendly.
2091 21:51 < matches> It will have a webcam image though
2092 21:51 * sulix mutters "ASCII art" under his breath.
2093 21:52 < matches> I figure if I implement everything in C++ it will be less "Webby"
2094 21:52 < matches> And more segfaultastic
2095 21:52 * sulix has a new favourite word.
2096 --- Day changed Thu Jun 26 2014
2097 14:58 < matches> I think the WebSocket RFC put this "Framing" business in just to confuse people
2098 14:58 < matches> My RFC Compliant* WebSocket server just sends the entire message in a single frame
2099 14:59 < matches> Doesn't TCP already do framing or something
2100 15:00 < matches> I mean, they specifically allow for frames with up to 2^63 bytes
2101 15:00 < matches> Sorry, bits not bytes
2102 15:02 < matches> So server -> client is pretty much TCP but you stick some stupid frame guff on the message
2103 15:02 < matches> client -> server involves... "masking"
2104 15:02 < matches> There are several long paragraphs on it
2105 15:10 < matches> I'm going to pretend this is related to precision
2106 --- Day changed Wed Jul 02 2014
2107 17:30 < matches> I put the code repository on github
2108 17:32 < matches> This is where a more hip and cool person would say something like "#yolo" but I prefer to wonder how I got 26K of lines and you have 65K...
2109 17:32 < matches> "How many lines of code does it take to draw a rectangle?" 91000
2110 --- Day changed Fri Jul 04 2014
2111 12:59 < matches> Found my rational bug
2112 13:00 < matches> ...
2113 13:00 < matches> Copy paste strikes again
2114 13:00 < matches> My operator/ looked suspiciously similar to operator+ -_-
2115 13:00 < matches> -____-
2116 13:02 < matches> Not sure how it got like that, it was definitely working at one point
2117 13:02 * matches should probably use git properly
2118 13:12 < matches> Pushed it
2119 13:21 < matches> So yeah, it should work fine until you zoom
2120 13:21 < matches> Then crazy things happen
2121 16:44 < matches> I think all the references you added at the last minute break my litreview
2122 16:45 < matches> (he says several months later :P_
2123 21:00 <@sulix> Question: in Rational::Simplify(), why is (-P)/(-Q) becoming (-P)/(Q) rather than P/Q?
2124 21:01 <@sulix> Shouldn't P = -P here? http://git.ucc.asn.au/?p=ipdf/code.git;a=blob;f=src/rational.h;h=41cce09078b4e96eb17a6bb6e28478ea64637b57;hb=HEAD#l74
2125 21:02 <@sulix> Also zoom can be improved (but not totally fixed) by reducing precision when overflow would be possible.
2126 21:19 < matches> Yeah, that's the idea of using the Stern Barcot tree Tim was talking about
2127 21:20 < matches> Which I still need to read about
2128 21:20 < matches> Also yes P should be -P because I am just that amazing at maths
2129 21:22 < matches> In other news assembly happened
2130 21:22 < matches> http://git.ucc.asn.au/?p=ipdf/code.git;a=blob;f=src/tests/add_digits.s
2131 21:22 < matches> Which no doubt has terrible bugs in it :P
2132 21:23 < matches> It was a bit confusing how passing and returning arguments with x64 was completely and utterly different from x86 and almost every tutorial is from at least ten years ago talking about x86
2133 21:24 < matches> x64 is actually much easier though
2134 21:25 < matches> At least, I think they are different, I didn't have to mess with the stack which seemed to be a thing with x86 and also that motorolla language we used in first year
2135 21:25 < matches> Just got all the arguments in a bunch of registers
2136 21:31 <@sulix> I think I just worked out why you have 16 "inc" instructions, and am loving it!
2137 21:32 <@sulix> inc doesn't change flags!
2138 21:32 < matches> I wonder if it would be faster to just store the carry flag in a register though
2139 21:33 < matches> Actually probably not because there'd be branches involved
2140 21:33 < matches> Hmm
2141 21:33 <@sulix> Traditionally, mucking about with flags is slow.
2142 21:33 < matches> Yep cool
2143 21:34 < matches> I've got the AMD64 ABI
2144 21:34 < matches> Chapter 2 is a good one
2145 21:34 < matches> "Software Installation"
2146 21:35 < matches> "This document does not specify how software must be installed on an AMD64 architecture machine"
2147 21:35 < matches> End chapter
2148 21:35 <@sulix> Just a thought, can use use the lea instruction instead of all of those incs.
2149 21:37 <@sulix> Like lea rsi, [rsi+8h]?
2150 21:37 * sulix ponders.
2151 21:38 < matches> Maybe, my assembly is very rusty and was based on those ELEC1301 labs in the first place...
2152 21:38 < matches> I don't even remember if they had lea
2153 21:39 <@sulix> (Also I just got a message from KDE: "could not launch a process. perhaps you have reached the limit of running processes")
2155 21:40 < matches> When did you last reboot?
2156 21:40 <@sulix> This morning, actually.
2157 21:40 < matches> Integer overflow in PID?
2158 21:40 <@sulix> It looks like it was an out-of-memory problem.
2159 21:40 <@sulix> On a related note, the intel instruction set documentation is a 3 and a half thousand page pdf.
2160 21:41 <@sulix> and firefox has a PDF viewer written in javascript.
2161 21:41 <@sulix> ...that runs by default.
2162 21:44 < matches> Hahaha
2163 21:44 < matches> Yes, pdf.js is apparently the best thing since sliced bread...
2164 21:45 < matches> I'm having some difficulty with lea, possibly the syntax
2165 21:47 < matches> "lea (%rsi+8), %rsi" is wrong
2166 21:48 <@sulix> Yeah, gcc's assembler uses whacked syntax.
2167 21:48 < matches> I thought it was just () instead of [], add % to all the registers, and swap destination and source
2168 21:48 <@sulix> http://git.ucc.asn.au/?p=ipdf/code.git;a=blob;f=src/tests/add_digits.s;hb=HEAD
2169 21:49 <@sulix> They replaced the arithmetic operators with strange things, too.
2170 21:49 < matches> Riight
2171 21:49 < matches> Thanks
2172 21:49 < matches> So 8(,%rdi,1) means [rdi+8]
2173 21:49 < matches> That is dumb
2174 21:49 <@sulix> I'm not sure if it's worth using nasm or the ".syntax intel" directive.
2175 21:51 < matches> Hopefully there won't be too much more assembly to write...
2176 21:51 <@sulix> Technichally lea (and other memory addressing instructions) take the form "[register+register*(1,2 or 4)+(number)]" in intel and "number(%register,%register,(1,2 or 4))"
2177 21:51 <@sulix> in gcc
2179 21:51 < matches> Well once I know the syntax I can cope with it
2180 21:51 < matches> But all the documentation being written in a different syntax doesn't help :P
2181 21:52 <@sulix> Yeah, there are a lot of people who hate gcc syntax, which was an (unsuccessful) attempt to make assembly for different machines more consistent)
2182 21:52 < matches> I decided to go with it because I vaguely remembered your commander keen patches looked like they were in gcc syntax :P
2183 21:53 <@sulix> Hah: they were in "ckpatch" syntax.
2184 21:53 <@sulix> And were just the final output bytes.
2185 21:53 < matches> The memory can play strange tricks...
2186 21:54 <@sulix> I used the "debug.com" utility that came with dos to assemble them, which let you write assembly to arbitrary memory.
2187 21:54 <@sulix> and then read the contents of that memory to see what it assembled down to.
2188 21:55 < matches> So, I need subtraction and multiplication but not division to implement arbitrary rationals
2189 21:55 < matches> I'll try sort those out tomorrow
2190 21:56 < matches> I'm pretty sure I'm doing better than at least one arbitrary integer representation
2191 21:56 < matches> Which had a int32_t word and int64_t "double" word
2192 21:56 < matches> So they could tell if they had a carry by doing the additions using double words
2193 21:57 < matches> The code in C was longer than add_digits is in assembly
2194 21:57 < matches> Although to be fair it might actually work on things that aren't amd64
2195 21:57 <@sulix> Well, the good news is that there's an "sbb" instruction like "adc"
2196 21:57 < matches> Excellent
2197 21:58 <@sulix> The bad news is that multiplication is harder.
2198 21:58 < matches> Yeah, at least I don't need division though
2199 21:59 < matches> Although converting it to a string will be annoying
2200 21:59 <@sulix> IIRC, the IMUL instruction gives an output twice as long as the inputs, to prevent overflow.
2201 22:00 <@sulix> Converting to a string won't be so bad if you get division done.
2202 22:00 <@sulix> And you can always just implement multiplication and division by slow recursive addition/subtraction.
2203 22:03 < matches> There's a bunch of papers on number representations that use things that aren't floats or rationals from a quick look
2204 22:04 < matches> I guess I just need to implement as many things as possible and compare them all
2205 22:04 < matches> Seems easy enough...
2206 22:04 < matches> (Famous last words)
2207 --- Day changed Sat Jul 05 2014
2208 11:09 < matches> Well... subtraction took me two seconds
2209 11:09 < matches> Should probably have just done it last night
2210 11:09 < matches> I got distracted
2211 11:09 < matches> By someone actually using my websocket library
2212 11:09 < matches> Does this mean I'm officially an open source developer now
2213 11:09 < matches> I didn't even give it a license :S
2214 11:11 < matches> Hmm
2215 11:12 < matches> Multiplication...
2216 11:12 < matches> It is tempting to just use a for loop and call the += operator...
2217 11:40 < matches> I'll do that for now I think
2218 11:41 < matches> Once += actually works
2219 11:41 < matches> Turns out I will definitely need division also
2220 11:42 < matches> I was looking at the rationals going "There are no divisions here!"
2221 11:42 < matches> But there are
2222 11:42 < matches> In the gcd
2223 11:42 < matches> Mind you maybe I don't need to call gcd if I have arbitrary integers :P
2224 11:42 < matches> Just use all the memory
2225 12:39 < matches> Ok for loop for multiplication is not easy
2226 12:39 < matches> Bah
2227 12:39 < matches> Having to actually learn assembly properly
2228 12:40 < matches> I mean I could use a for loop but it would have to use an Arbitrary integer and take a very long time...
2229 12:47 <@sulix> There's no need to have a for loop index > 64bit because there will never be enough memory for an integer that big.
2230 12:47 < matches> True
2231 12:54 < matches> Putting "_asm" in filenames so that our Makefile can do the .s files is kind of irritating
2232 12:54 < matches> Would naming them .cpp work? That would be even more irritating though!
2233 12:54 < matches> Oh well it compiles
2234 12:58 < matches> I bet it turns out 128 bit integers would have been more than sufficient for any zoom range people could feasibly get to by scrolling
2235 13:15 <@sulix> I'm told 256-bit integers can represent the visible universe to the width of a proton or something.
2236 13:45 < matches> I might put a maximum length in once this works
2237 13:45 < matches> So I'm not doing a Java
2238 13:46 < matches> Where 1/3 becomes "Runtime Exception!"
2239 13:46 < matches> That's with BigDouble not BigInt but whatever
2240 14:00 < matches> I must admit, assembly is pretty nice for this sort of thing
2241 14:00 < matches> At least, the inner operations
2242 14:00 < matches> Not for managing the memory
2243 15:29 < matches> Alright. We have an Arbint class that is equivelant to an int64_t when it only has one digit.
2244 15:30 < matches> I'm not sure how to test that extra digits function correctly
2245 15:30 < matches> Other than just looking at a few cases and going "Yep that looks right" :S
2246 16:40 < matches> Well... somewhere along the way I lost kcachegrind and now I seem to have to run dist-upgrade to get it back
2247 16:41 < matches> Currently 10% of the way through upgrading tex-live...
2248 16:41 < matches> The mystery of just how shitty the Arbints are will have to wait
2249 16:42 < matches> Well it's clear they are shitty but I suspect there is a bug in division
2250 16:42 < matches> Maybe I should actually read my code
2251 16:42 < matches> Blargh
2252 16:47 < matches> Ok constructing a rational takes infinite time
2253 16:51 < matches> Ah, 1 % 1 == 1
2254 16:51 < matches> That is definitely not correct :P
2255 17:56 < matches> Oh my goodness I just rendered a frame
2256 17:56 < matches> On the GPU...
2257 18:00 <@sulix> Hmm... {add,sub}_digits_asm.s are missing from git.
2258 18:05 < matches> Really
2259 18:05 < matches> Whoops
2260 18:07 < matches> There you go
2261 18:07 < matches> Addition, subtraction, multiplication should be fine, I need to actually do division properly
2262 18:08 < matches> Also probably pow at some point
2263 18:08 < matches> Since pow still just casts things to doubles
2264 18:22 <@sulix> It works!
2265 18:22 <@sulix> (Eventually)
2266 18:24 < matches> Haha
2267 18:24 < matches> Of course it's still in the single digit range
2268 18:24 < matches> Or did you manage to wait long enough to zoom a bit
2269 18:25 < matches> I'm sure I can make this division better if I use some kind of binary searchish type algorithgm
2270 18:25 < matches> I could look up the proper algorithm too I guess
2271 18:25 <@sulix> Yeah, I was thinking binary search.
2272 18:26 <@sulix> You'd want a decent bitshift implementation.
2273 20:32 < matches> Yeah it is more difficult than I thought
2274 20:33 < matches> I'm going to need "shift_digits" assembly functions
2275 20:33 < matches> C style bit shifting has too many wierd edge cases
2276 20:36 < matches> 0x7000000000000000 >> 3 == 0x0e00000000000000 ?
2277 20:36 < matches> There are more bits set in the second number than the first!
2278 20:38 < matches> And unlike multiplication where it fills an extra register for you, you just lose the extra bits with shr and shl :S
2279 20:39 < matches> Except for a carry
2280 20:39 < matches> So I'd have to do a loop one bit shift at a time
2281 20:39 < matches> Blergh
2282 20:40 < matches> I shall resume my epic battle with primary school level mathematics tomorrow
2283 --- Day changed Sun Jul 06 2014
2284 00:32 <@sulix> long division of an arbint by a uint64 worksish.
2285 09:47 <@sulix> long division of an arbint by a uint64 now actually works and is less obviously badly designed
2286 10:32 < matches> Thanks...
2287 10:33 < matches> In other news I seem to have broken X again
2288 10:34 <@sulix> Oh no!
2289 10:35 < matches> We need an ASCI renderer...
2290 10:36 <@sulix> That can not possibly end badly.
2291 10:40 < matches> shouldn't be too hard...
2292 10:40 < matches> Lets see if reinstalling fglrx works
2293 10:41 < matches> I need to add that to a boot script :S
2294 10:42 < matches> uhoh installation failed
2295 10:48 <@sulix> Hmm... enabling optimization makes Arbint segfault.
2296 10:49 < matches> You should probably make quadtrees
2297 10:50 < matches> Although I admit you're better at Arbints than me :P
2298 10:50 <@sulix> Yeah... I should.
2299 10:50 <@sulix> But segfaults...
2300 10:51 < matches> Yeah somewhere in std vector I had some as well
2301 10:51 < matches> I think I might change to a custom dynamic array
2302 10:52 < matches> Then fast copying is easier
2303 10:52 <@sulix> Yeah, it's segfaulting in std::swap somewhere, I think.
2304 10:52 < matches> I'll fix it when I have X
2305 10:53 < matches> Or an ascii renderer...
2306 10:54 < matches> Probably more realistic at this stage
2307 10:55 <@sulix> I'm pretty certain there's a unicode character for "face of disapproval due to fglrx bugginess".
2308 10:56 < matches> Yeah but when you try and render it you get a seg fault
2309 11:24 < matches> Woo, xvesa
2310 11:26 < matches> I see what you mean about fglrx being hard to remove
2311 11:58 <@sulix> So: quick profile. 98% of frame rendering time is spent in Arbint::Division.
2312 11:59 < matches> Yeah
2313 12:00 < matches> I *was* trying to install kcachegrind you know
2314 12:00 <@sulix> The vast majority of that time is spent in malloc and free.
2315 12:01 < matches> Well copy constructing an Arbint requires copy constructing the vector
2316 12:01 < matches> I'll investigate
2317 12:01 < matches> Now I have X again
2318 12:01 < matches> Go make a quad tree
2319 12:02 <@sulix> Also, should add_digits clear the carry flag at the beginning?
2320 12:02 < matches> Hmm
2321 12:02 < matches> Probably
2322 12:02 < matches> Yeah
2323 12:09 < matches> I suspect having a copy-on-write Arbint won't help as much as having a non-terrible division algorithm so I'll try fix that first
2324 12:09 < matches> But copy-on-write Arbint shouldn't be too hard
2325 12:09 < matches> Oh unless the original changes
2326 12:10 < matches> Yeah it would be hard
2327 12:10 < matches> I'm sure I did something like it before
2328 12:10 < matches> Back when I actually made my own data structures instead of just using std::vector
2329 12:15 <@sulix> My custom vector class: http://davidgow.net/stuff/DynamicArray.h
2330 12:15 <@sulix> Probably won't compile without a bunch of other files, though.
2331 12:16 < matches> I think I have several, in different versions of neurosis
2332 12:16 < matches> Which was never under git
2333 12:16 < matches> Some of them were prone to segfaults though
2334 12:17 < matches> So, will you have some kind of quad-tree-y thing by monday? :P
2335 12:17 <@sulix> Mine works, but doesn't call constructors or destructors, so it's only really useful for ints/floats/pointers or structs thereof.
2336 12:18 <@sulix> I have a "struct QuadTree".
2337 12:18 <@sulix> (You can't do anything with it, but the basic idea is there)
2338 12:20 < matches> My shifting operators might not have been as fith as I thought last night
2339 12:20 < matches> Since 7 and e do actually have the same number of bits
2340 12:21 < matches> I'm not sure where that came from
2341 12:21 < matches> Ah dammit
2342 12:21 < matches> Cannot find -lGL
2343 12:22 <@sulix> Try just removing that from the Makefile
2344 12:22 <@sulix> I think when I GL-3'd everything I made it load libGL at runtime.
2345 12:23 < matches> Well it will compile but I get glx errors
2346 12:23 < matches> I get similar errors running glxinfo though...
2347 12:24 < matches> Odd
2348 12:26 <@sulix> I think you still have some of fglrx breaking things.
2349 12:26 <@sulix> It replaces the system libGL.so
2350 12:26 < matches> I installed mesa's stuff
2351 12:27 <@sulix> You need to also get the version of libGL that runs as part of the X server, IIRC.
2352 12:28 < matches> I'll try some apt-get --reinstalls
2353 12:29 < matches> I think most of my problems were that debian was going "You don't have a graphics driver, here, have one" and then things were breaking and reinstalling fglrx would give me graphics but leave me with a partially installed debian driver as well :S
2354 12:30 < matches> Well ok
2355 12:30 < matches> most of my problems were fglrx
2356 12:30 <@sulix> I remain sceptical that installing fglrx by hand will ever cause anything that is not pain.
2357 12:30 < matches> Well it worked better than the open source drivers for a very long time
2358 12:30 < matches> It all ended in tears and segfaults though
2359 12:32 < matches> Also it worked better than the non-free fglrx package in the debian repos which always ended in tears and segfaults
2360 12:33 <@sulix> I use the non-free nvidia package, which works fine, even if it is horribly outdated.
2361 12:34 <@sulix> I'm scared of the sheer amount of ripping up the rendering code getting quadtrees working will take.
2362 12:39 < matches> Haha
2363 12:40 < matches> It might be easier to do the CPU rendering in quad trees first?
2364 12:40 < matches> I'm not quite sure how you'll do it
2365 12:41 < matches> CPU rendering at least gives you direct control over the rendering target as a pixel buffer
2366 12:41 < matches> GPU rendering is a bit shady (ho ho)
2368 12:42 <@sulix> That was terrible, yet accurate!
2369 12:42 < matches> With the CPU rendering we're already passing the objects list, view, and target bounds
2370 12:42 < matches> So maybe you can adapt View or something to be quad-tree-ish
2371 12:43 < matches> Good luck
2372 12:44 <@sulix> Yeah, the current plan is to adapt the rendering stuff to accept a "first object id" and "last object id", and then only render those.
2373 12:47 < matches> Hmm, you could call RenderUsingCPU with a different target pixel buffer for each part of your quad tree or something
2374 12:48 < matches> Oh well, hopefully I can adapt whatever you do into RenderUsingASCII :P
2375 12:49 < matches> I don't need GLX working to try and fix division but I'd really like to have GLX working
2376 12:49 < matches> Sigh
2377 13:12 < matches> I'm making progress, instead of reporting an invalid operation glxinfo segfaults now
2378 13:13 < matches> So I have the wrong libraries or something... how to get the right ones...
2379 13:32 <@sulix> Woo. I got it compiling again!
2380 13:32 <@sulix> And the rendering is only half-wrong.
2381 13:33 < matches> Nice
2382 13:33 < matches> Manually deleting anything on my system with fglrx in it...
2383 13:34 <@sulix> Someone really needs to write an "fglrx removal guide".
2384 13:34 < matches> I'm not sure why reinstalling the radeon driver and mesa isn't actually giving me a libGL
2385 13:34 < matches> At least
2386 13:34 < matches> I suspect it is just leaving the existing broken one
2387 13:36 <@sulix> So it turns out that *2 I don't understand is important.
2389 14:01 < matches> I have glxgears
2390 14:01 < matches> An epic achievement
2391 14:01 < matches> I had to go back to fglrx but at least this time its the debian packaged version
2392 14:02 < matches> The radeon driver seems to work better for stuff that isn't glx
2393 14:03 < matches> Which is pretty cool but I kind of like glx actually working
2394 14:11 < matches> Ok... time to fix division
2395 14:11 < matches> Finally
2396 14:11 < matches> No wait time for coffee
2397 14:11 < matches> Then time to fix division
2398 14:24 < matches> Did you test the div_digits function at all?
2399 14:25 < matches> # We want to point to the end of the buffer (LSB)
2400 14:25 < matches> Aren't I doing it the other way around...
2401 14:25 < matches> I'll test it
2402 14:28 <@sulix> With long division you do things in the opposite direction to addition.
2403 14:29 <@sulix> I tested it with some simple things and it worked.
2404 14:34 < matches> Hmm
2405 16:35 < matches> Well the good news is that division is faster now
2406 16:35 <@sulix> Cool!
2407 16:36 < matches> The bad news is that I can't see anything
2408 16:36 < matches> Even when GPU rendering
2409 16:36 < matches> Also segfaults
2410 16:36 <@sulix> I've just pushed some initial quadtree-enabling breakage.
2411 16:36 < matches> I didn't work out how to apply the div_digits to a multi digit number either
2412 16:36 <@sulix> Yeah, I thought about that, then went to bed.
2413 16:36 < matches> I might convert the current algorithm which is all based on bit shifting
2414 16:36 < matches> To assembly
2415 16:37 < matches> But maybe not
2416 16:37 < matches> Actually being able to see something render would be good even if it were slow
2417 16:37 < matches> Then I can think about making it faster
2418 16:38 <@sulix> The old one had all of the beziers connected for no readily apparent reason.
2419 16:38 < matches> I have a tester called "Arbitrary Rationals" (arbrationals.cpp) which sounds like some kind of political party
2420 16:39 < matches> A rather indecisive political party
2421 16:39 <@sulix> That's amazing.
2422 16:48 < matches> Ah, I might be having trouble with the sign of things, hm
2423 16:49 < matches> Rational 15625/288230376151727369 (0.000000) is not close enough at representing 1.000000 (1.000000 vs 0.001000)
2424 16:49 < matches> That doesn't look like a sign problem...
2425 16:52 <@sulix> I admit, 15625/288230376151727369 is not really similar to 1.
2426 16:54 < matches> So, somewhere along the way Arbint(1e6) became 15625 and Arbint(1e6) also became 288230376151727369 ?
2427 16:54 < matches> Uninitialised things?
2428 16:54 < matches> Constructing from int64_t seems fine
2429 17:00 < matches> That amazing moment when it works perfectly and then you realise that's because you compiled with reals as doubles and not rationals
2430 17:00 <@sulix> Ha: I've done that a few times.
2431 17:19 < matches> Maybe I shouldn't have pulled your latest commit
2432 17:19 < matches> Now I have two problems :(
2433 17:22 < matches> Meeting tomorrow
2434 17:22 < matches> "So how much progress have you made?"
2435 17:22 < matches> "Segfaults are up 100% to a record high!"
2436 17:26 < matches> Oh ok
2437 17:26 < matches> I'm not sure if this is new
2438 17:26 < matches> But if there is nothing in the document the GPU rendering doesn't like it
2439 17:27 <@sulix> Hmmm... there might be some issues with that.
2441 17:28 < matches> Rectangle renders
2442 17:28 < matches> With arbitrary rationals
2443 17:28 < matches> In the default view
2445 17:31 < matches> I'm going to commit this even though it's totally broken
2446 17:32 <@sulix> Things being broken hasn't stopped me from committing them.
2447 17:35 < matches> I love this though
2448 17:35 < matches> When you activate CPU rendering
2449 17:36 < matches> You get like a billion errors from Rational<Arbint>::CheckAccuracy
2450 17:36 < matches> And then the Bezier appears!
2451 17:36 < matches> Magically!
2452 17:36 < matches> I guess I shouldn't rule out a bug in CheckAccuracy :P
2453 17:36 < matches> Although zooming will definitely cause a segfault
2454 17:38 <@sulix> I tried running the really slow version under valgrind, but that turned out to be a bad idea.
2455 17:39 < matches> I've pushed the slightly less really slow version
2456 17:39 < matches> That is still really slow
2457 17:39 < matches> I can't install kcachegrind
2458 17:39 < matches> It is really annoying
2459 17:40 < matches> There aren't any valgrind analysis things that don't require the entirity of KDE are there?
2460 17:40 < matches> Or should I just learn how to read the raw output files
2461 17:40 <@sulix> Cool: it works.
2462 17:41 <@sulix> (I dunno, I just use KDE :P)
2463 17:41 <@sulix> You can try "perf" or "oprofile".
2464 17:41 < matches> Well I can't even install KDE :S
2465 17:41 <@sulix> They're terminal-based.
2466 17:41 <@sulix> Well, perf is, at least. There's a gtk+ oprofile gui.
2467 17:41 <@sulix> They both measure the entire system by default though.
2468 17:43 <@sulix> Rational::Simplify is ~54% of time (w/ GPU rendering) with the new build, according to callgrind.
2469 17:50 < matches> I'm a little confused by some of the outputs of Rational::CheckAccuracy given that it does actually seem to render the bezier properly on the CPU
2470 17:50 < matches> I think the segfault might be related to my (ab)use of things like memcpy on std::vector::data()
2471 17:52 < matches> Also I think I've fixed my dependency hell
2472 17:52 < matches> Well
2473 17:53 < matches> Depending on how this turns out I will end up with the entirity of KDE hopefully including kcachegrind, or I will end up with no X again
2474 17:53 < matches> And much swearing will ensue
2475 17:54 <@sulix> Yeah, it segfaults immediately if you enable optimization level 2 or higher.
2477 17:56 < matches> I'll probably get rid of std::vector then
2478 17:57 <@sulix> The problem I'm seeing is in the vector<Bezier>::push_back() bit, so who knows what is going on.
2479 17:58 < matches> Oh that's odd
2480 17:58 <@sulix> Okay, don't run it in valgrind.
2481 17:58 <@sulix> I'm getting about 200 "uninitialized values" a second in the nVidia driver.
2482 17:58 < matches> But I just installed kcachegrind!
2483 17:59 <@sulix> (Oh wait, that was with REAL=double)
2484 17:59 < matches> Haha
2485 18:00 < matches> Wait, vector<Bezier>::push_back causes problems even with REAL=double :S
2486 18:00 <@sulix> Nah, the nVidia driver seems to.
2487 18:01 <@sulix> I've now compiled again with REAL=5 and I'm getting a number of invalid reads off the end of the vector<digit_t> in GetBit()
2488 18:02 < matches> Division
2490 18:02 < matches> oooh
2491 18:02 < matches> > -> >=
2492 18:02 < matches> ...
2493 18:03 <@sulix> Well, my first attempt to fix it just broke rendering.
2494 18:04 <@sulix> Ah, yep, I see it.
2495 18:05 < matches> I think I need to do less memcpy/memseting and more std::vector::operator= -_-
2496 18:05 < matches> I mean, they implemented it for a reason
2497 18:05 <@sulix> So, with that change it no longer segfaults!
2498 18:05 <@sulix> (It corrupts memory and throws an exception instead)
2499 18:05 < matches> Woah
2500 18:06 < matches> I changed Arbint operator= to use std::vector operator=
2501 18:06 < matches> And now I get a filled circle on top of my bezier...
2502 18:07 < matches> Ok, I will rewrite Arbint to just rely solely on std::vector operations
2503 18:08 < matches> That might slow it down but it's impossible to test that the maths works properly with all these memory errors
2504 18:09 <@sulix> Should you be using m_sign more often?
2505 18:09 <@sulix> It seems to be "forgotten" in a number of places...
2506 18:10 <@sulix> Hmm... everything looks like it's working at the moment, actually.
2507 18:10 < matches> What else did you change?
2508 18:10 < matches> Try zooming
2509 18:11 < matches> That will break it
2510 18:11 <@sulix> Zooming seemed to work, actually.
2511 18:11 < matches> Are you sure you compiled with REAL=5
2512 18:11 < matches> The sign is a bit wierd
2513 18:12 <@sulix> It works for a little bit, then the view bounds suddenly become huge.
2514 18:12 < matches> Haha
2515 18:12 <@sulix> For no reason, the y coordinate jumps to 10^8.
2516 18:13 <@sulix> Also after you've zoomed, sometimes translating goes in the wrong direction or is otherwise broken.
2517 18:15 < matches> That might be related to m_sign...
2518 18:15 < matches> The joys of implementing your own number representation
2519 18:16 <@sulix> Trying to "fix" the sign fixed zoom, but broke translation even more.
2520 18:18 <@sulix> Also: best representation of 1 ever: 9223372036854775807/9223372036854775807
2521 18:18 < matches> Hahaha
2523 18:18 < matches> I think I commented out a Simplify() somewhere :P
2525 18:18 < matches> I didn't
2526 18:18 < matches> Hmm
2527 18:19 < matches> Well I am going to push some minor changes
2528 18:20 <@sulix> Also: T g = gcd(T(llabs(P)),T(llabs(Q)));
2529 18:20 <@sulix> Is that "llabs" truncating everything?
2530 18:20 < matches> ...
2531 18:20 < matches> Probably
2533 19:05 <@sulix> Well, my attempt to write a templated abs() function that worked for Arbint may have worked, but it crashed X before I could find out.
2534 19:05 <@sulix> I do not want to know how that is possible.
2535 19:05 < matches> Heh, I was trying to do that before dinner
2536 19:06 < matches> I just got segfaults
2538 19:08 < matches> I see
2539 19:08 <@sulix> I suspect the best thing for me to do is to just stash all of my changes and break some more quadtrees.
2540 19:08 < matches> Yeah
2541 19:08 <@sulix> But first I'll have to get X working again.
2542 19:08 < matches> The problem with overriding llabs is that the constructor for Arbint calls llabs...
2543 19:08 < matches> And Arbint has a constructor from int64_t...
2544 19:09 < matches> So as soon as you implement llabs using Arbint it will call itself for eternity
2545 19:09 <@sulix> I made a new function called "real_abs" (because "i_went_to_the_gym_now_look_at_my_abs" was too long) and just used that.
2546 19:09 < matches> Bahaha
2547 19:10 <@sulix> Then X started using 100% of my CPU and nothing responds to anything.
2549 19:10 <@sulix> The mouse cursor still moves, but everything else is a black screen.
2550 19:11 <@sulix> Excellent, "sudo pkill -9 Xorg" has made everything work again.
2551 19:12 <@sulix> (And even fixed an unrelated problem, which is good but worrying)
2552 19:17 <@sulix> Note to self: if something crashes X when you run it once, don't run it again after fixing X.
2553 19:18 < matches> Haha
2554 19:18 <@sulix> Now to try it for a third time.
2555 19:18 < matches> I've got an abs working
2556 19:18 <@sulix> Ah, excellent.
2557 19:19 <@sulix> Does it make things better? Worse? The same but slower?
2558 19:19 <@sulix> Crash X?
2559 19:20 < matches> It doesn't crash X but I think I have an infinite loop in gcd now
2560 19:21 < matches> I should probably work out why X/X for really big X is not simplifying to 1/1
2561 19:43 < matches> So it turns out it's a pretty good idea to have a tester that actually tests lots of different numbers with every single operation
2562 19:43 < matches> As opposed to
2563 19:43 < matches> "1/2 == 0.5" yep looks good she'll be right
2564 19:43 < matches> One could say
2565 19:43 < matches> There are *signs* of serious problems
2566 19:44 * matches puts on sunglasses
2567 19:44 < matches> No wait I got the order wrong
2568 19:44 < matches> Oh well I wasn't cool enough to watch that show anyway
2569 20:18 < matches> So now I have a tester that actually tests things
2570 20:18 < matches> I just need to blindly fumble around with m_sign until all the tests are passed!
2571 20:18 < matches> Right?!
2572 20:26 < matches> Heh, it's amazing how many errors fixing the sign in Division prevents
2573 20:27 < matches> Well amazing until you realise that every single Rational calls a division at least once
2574 20:44 <@sulix> Hmmm... there seem to be a lot of cases where we get things becoming 1 (randomnumber/randomnumber) instead of what they should be, according to CheckAccuracy.
2575 20:46 < matches> Yeah
2576 20:46 < matches> It should work nicer nowish
2577 20:46 < matches> At least, the tester works most of the time...
2578 20:46 < matches> It still randomly fails sometimes
2579 20:46 <@sulix> Yeah, it's working much better.
2580 20:46 < matches> Now 12% of the time is in std::vector::size()
2581 20:46 <@sulix> All of the things I've seen have been to do with subtraction in ScaleAroundPoint()
2583 20:47 < matches> Wait I haven't pushed the 12% in std::vector::size yet
2584 20:47 < matches> tl;dr I was allocating one extra digit every single addition operation and then deleting it if it wasn't used
2585 20:48 < matches> Because I am just that good :P
2586 20:48 < matches> I suspect the Arbint is going to break when it actually does need to resize itself
2587 20:49 < matches> But for now I will settle for "At least as good as an int64_t"
2588 20:49 < matches> * But slower
2589 20:49 <@sulix> I'm curious as to why digit_t is int64_t and not uint64_t.
2590 20:49 < matches> Oh it is uint64_t now
2591 20:49 < matches> It was int64_t mainly because geany doesn't syntax highlight uint64_t
2593 20:50 < matches> Then I realised being int64_t instead of uint64_t broke the shift operators
2594 20:50 < matches> Because they try to preserve the sign
2595 20:50 <@sulix> Shift on signed ints is undefined behaviour, IIRC.
2596 20:51 < matches> Yeah I was considering writing the shift in assembly but its easier to handle the underflow/overflow into the next digit in C
2597 20:52 < matches> Zooming is still incredibly slow
2598 20:54 < matches> In fact it is infinitely slow
2599 20:54 <@sulix> I once waited about 10 minutes and did get another frame.
2600 20:54 < matches> In fact the Arbint appears to want to grow an infinite number of 0L digits
2601 20:55 <@sulix> Ah ha! I have realised why CheckAccuracy is only throwing errors on "-".
2602 20:55 <@sulix> The CheckAccuracy calls for everything else have been commented out,
2603 20:55 < matches> They should all be commented out so I can use tests/realops.test instead
2604 20:57 < matches> 23 / 12000 operations failed... doesn't seem too bad...
2605 20:58 <@sulix> Keep in mind that == isn't precise as it casts to doubles.
2606 20:58 < matches> You get more like 40 out of Rational<int64_t>
2607 20:58 < matches> Yeah they are "equal" if it is within 1e-1 :S
2608 20:58 < matches> I'm searching for the more totally catastrophic failures like the wrong sign first :P
2609 21:20 < matches> Ok tomorrow I might just try using boost arbitrary integers or something
2610 21:28 <@sulix> All of the failures I'm getting are with "/="
2611 21:28 <@sulix> Except one, which was only just outside the error bounds, so probably the doubles' fault.
2612 21:30 < matches> I was restricting the test values somewhat though
2613 21:30 < matches> (With rand()%100 + 1)
2614 21:33 <@sulix> You do know that the Rational<Arbint> -> double conversion is totally broken if either P or Q have > 1 digit.
2615 21:33 < matches> Yeah I noticed that
2616 21:37 <@sulix> Oh man, re-enabling the operator double on Arbint totally breaks things.
2617 21:40 < matches> Yeah don't touch it
2618 21:40 < matches> Basically touching anything will totally break something else
2620 21:40 < matches> I should probably sleep some more before attempting to fix things
2621 21:40 <@sulix> The other bug I just found is that Arbint didn't have the unary "-" operator, so writing "-P" would silently cast to int64_t.
2622 21:41 < matches> Ah ok
2623 21:41 < matches> I thought unary - was equivelant to "0 - x" but I suppose that doesn't make sense
2624 21:41 <@sulix> Nah, it has to be written explicitly.
2625 21:44 <@sulix> Okay, I think I've got realops.test down to 0 errors with Arbint.
2626 21:45 < matches> Well you're doing better than me
2627 21:46 <@sulix> I'm running it with a huge number of tests now.
2628 21:47 <@sulix> 85.95% of my entire computer's processing time is in IPDF::Rational<IPDF::Arbint>::operator/=
2629 21:48 < matches> Can you push what you have?
2630 21:48 <@sulix> Yeah, I will in a second.
2631 21:48 <@sulix> It's basically a lot of hacks.
2632 21:49 <@sulix> CPU rendering is broken, though.
2633 21:49 < matches> There were cases where it would grow an infinite number of digits; did you fix those?
2634 21:49 <@sulix> Probably not: it still gets very, very slow zooming.
2635 21:51 < matches> Yeah there's something wrongish it shouldn't need extra digits so fast
2636 21:51 <@sulix> Pushed now.
2637 21:52 <@sulix> Now to look at the diff and see what I actually changed.
2639 21:54 < matches> Simplify call in Rational operator= is good
2640 21:56 < matches> I think I fixed some more bugs but I also introduced an infinite loop so...
2641 21:57 < matches> How is the quad tree bit going :P
2642 21:57 <@sulix> Yeah, about that...
2643 21:58 < matches> Remember Arbints don't have to work for quad trees to work :P
2644 21:58 < matches> Or do they
2645 21:58 <@sulix> In theory, no.
2646 21:58 <@sulix> In practise, it probably depends on how one defines "success".
2647 21:59 < matches> I think we need to design a test pattern where you actually want to zoom forever
2648 21:59 <@sulix> Highly theoretically, I'm pretty certain the QuadTree basically ends up isomorphic to the Arbint.
2649 21:59 < matches> Seems legit
2650 21:59 < matches> Can you prove it?
2651 21:59 < matches> That would be a useful result
2653 22:00 <@sulix> *hopefully
2654 22:00 < matches> I'm afraid most of my "theory" is based solely on intuition at this point :S
2655 22:00 < matches> Well the division algorithm came from wikipedia but division isn't as intuitive as the other operations
2656 22:01 < matches> Actually if I can fix all these other stupid bugs I'll try my binary searchish division idea
2657 22:01 <@sulix> Yeah, I spent a while racking my brain for long division, and then only managing to work it out for dividing by a single digit.
2658 22:02 <@sulix> That's really, really easy in asm, as well, which is nice, but not quite as useful as I'd hoped.
2659 22:02 < matches> It's pretty useful; we can still have a "if div.m_digits.size() use the assembly function"
2660 22:02 < matches> * size() == 1 that is
2661 22:02 <@sulix> Wikipedia mentioned doing newton-raphson for division, which is a terrifying thought.
2662 22:02 < matches> Haha
2663 22:03 < matches> So I've got Arbint::Shrink() calls pretty much after every operation now for no apparent reason
2664 22:04 <@sulix> Does it help?
2665 22:05 < matches> On the 551st test Shrink has been called 4 times in a row and then something after that causes an infinite loop
2666 22:05 < matches> I can't work out what it is
2667 22:05 < matches> I have a GrowDigit that spams Debug messages now
2668 22:05 < matches> So its not that
2669 22:05 < matches> It's silent
2670 22:05 < matches> Let's see what callgrind has to say
2671 22:06 < matches> A few minutes should be enough for it to be obvious right
2672 22:07 < matches> I think the division loop runs forever
2673 22:07 < matches> Bugger
2674 22:08 < matches> Wait that doesn't make sense there's only 1 digit
2675 22:08 < matches> Blargh
2676 22:17 <@sulix> So using the asm version when possible makes realops.test 20x faster, but there were a couple of failed tests.
2677 22:18 <@sulix> I'm not sure if they're new or not, because I was able to run many more tests.
2678 22:20 <@sulix> Also, I think I've found your infinite loop.
2679 22:23 <@sulix> Okay, maybe I've just obfuscated the code some more.
2680 22:24 < matches> Infinite loop is due to subtraction bug
2681 22:24 < matches> I'm pretty sure
2682 22:24 <@sulix> Yeah, pushing a "fix"
2683 22:26 <@sulix> Pushed, but think I might have found a bug in the fix.
2685 22:27 < matches> (0,0) - +(16292953875657448384,1) = -(16292953875657448384,2) seems wrong
2686 22:28 <@sulix> Pish! Nothing could be more beautiful.
2687 22:28 < matches> It's pretty hard to debug with such huge numbers :P
2688 22:31 <@sulix> Try this...
2689 22:31 * sulix pushes.
2690 22:32 < matches> What does that do...
2692 22:33 < matches> You negate the entire thing and add one
2693 22:33 < matches> Seems legit
2694 22:33 <@sulix> So "two's complement" negatives mean that -A == flip all bits of A and add 1.
2695 22:33 < matches> Yeah
2696 22:34 <@sulix> There seems to be a bug in the asm div_digits though.
2697 22:35 < matches> I'm taking that out until I know the main algorithm works :P
2698 22:35 < matches> Silly I spent so much time wondering why division was broken without checking subtraction
2699 22:36 < matches> After I spent all that time thinking I was so clever for doing division using bit shifts and subtractions
2700 22:37 <@sulix> It still takes forever.
2701 22:37 <@sulix> (Zooming in, that is)
2702 22:37 < matches> Nah there's still a bug
2703 22:37 < matches> You're running with an older version of realops test
2704 22:37 < matches> Hang on
2705 22:37 < matches> I will push things
2706 22:38 < matches> Trying to avoid a merge...
2707 22:38 <@sulix> And I will get a drink.
2708 22:44 < matches> So if you run enough tests you eventually get a bunch of SubBasic spam
2709 22:45 < matches> Which is actually occuring due to "<" operators
2710 22:45 <@sulix> Yeah: I'm going to add a bunch of hacks to "<" at some point.
2711 22:45 < matches> Which are in the Division loop ( >= is implemented in terms of <)
2712 22:45 < matches> Don't add the hacks until subtraction works!
2713 22:46 < matches> I don't want to optimise it yet
2714 22:46 < matches> Otherwise all the optimisations hide the bugs
2715 22:47 < matches> Time for a tester that just does subtractions I think
2716 22:48 <@sulix> It might be worth adding a bunch of SDL_assert_paranoid()'s everywhere that check that (a - b)+b == a and similar on every - call.
2717 22:49 < matches> Most of the tested numbers can be represented in a single digit which is causing problems
2718 22:50 < matches> With finding the bugs in the multi-digit algorithm
2719 22:52 < matches> For small enough multi digit Arbints testing against doubles might work?
2720 22:52 <@sulix> Maybe.
2721 22:53 <@sulix> You'd lose enough precision that it's not worth it for anything > 2 digits, I'd say.
2722 22:53 < matches> Yeah, 2 digits should be enough to make sure the algorithms actually work though
2723 22:54 < matches> 2 digits in one operand and 1 in the other (which will resize itself to have a leading zero if necessary)
2724 22:55 < matches> At some point I am going to have to use a library for Arbints and compare ours to it
2725 22:55 < matches> But I kind of want ours to work properly before bringing in an external one
2726 22:56 < matches> When I try floats I might just bring in the external library :P
2727 22:56 <@sulix> Two questions:
2728 22:57 <@sulix> 1. What is "borrow" supposed to be? Is it the "carry" or the "result has changed sign" or something else?
2729 22:57 <@sulix> 2. What do you think of replacing Sub() with Add() + Negate()?
2730 22:59 < matches> borrow's like carry (I'm using subtract with borrow in asm) it should get subtracted from the next digit, and if at the end it is still set then it means your result is less than zero
2731 22:59 < matches> and 2 sounds like an excellent idea
2732 23:00 <@sulix> Okay.
2733 23:00 <@sulix> 'cause there's a separate "carry" flag and "sign" flag, so I wasn't sure.
2734 23:01 < matches> subtract actually working would be slightly faster than add and then negate I think
2735 23:01 <@sulix> I'm not 100% sure if all the code is using m_sign as the sign or if some of it is trying to use 2's complement, which isn't really possible with arbints.
2736 23:03 < matches> When I was testing the subtraction code I was using int64_t instead of uint64_t and it seemed to work :S
2737 23:04 < matches> I should check whether sbb does unsigned or signed subtraction
2738 23:04 <@sulix> Both, I think.
2739 23:04 <@sulix> 2's complement makes them the same, which is why it's so nice.
2740 23:04 <@sulix> But it requires a fixed number of bits to work.
2741 23:09 < matches> Whoops now the bit shift tester runs forever as well
2742 23:13 < matches> Oh no it doesn't
2743 23:13 < matches> It just exceeds the limit on GrowDigits
2744 23:14 < matches> Phew
2745 23:14 < matches> Bit shifting works at least
2746 23:14 < matches> Or at least it is consistently broken
2747 23:14 < matches> (a >> X) << X == a
2748 23:14 < matches> But == could be broken
2749 23:14 < matches> NOTHING IS SAFE
2750 23:14 < matches> When you are writing your own number represntation :(
2751 23:15 <@sulix> Idea: 2's complement subtraction will make -ve numbers have an "infinite" number of 1s at the front, so would that be infinite looping, trying to continually add digits with 1s.
2752 23:15 < matches> Operator overloading is amazingly powerful and yet it seems like the best way to make seemingly sane code do batshit insane things
2753 23:15 <@sulix> I think the answer is "no: we're always trapping it and converting it" but am not sure.
2754 23:16 < matches> Our subtraction doesn't take forever
2755 23:16 < matches> The issue is when the result would be less than zero
2756 23:16 < matches> Hang on let me make another tester
2757 23:17 < matches> Well, I'll put it in the arbint tester instead of the real tester
2758 23:18 < matches> Since it's pretty clear the bugs are in arbint not rational<arbint> itself
2759 23:18 < matches> Although there might be bugs in rational<T> as well
2760 23:18 < matches> But we'll squish those when we get to them
2761 23:18 < matches> Also quad trees are they a thing yet?
2762 23:20 <@sulix> "The quadtrees aren't in the code, they were in your heart all along."
2763 23:21 < matches> Haha
2764 23:21 < matches> Tell that to Tim :P
2765 23:21 < matches> Ok (45,10) - (128,0) = -(83,18446744073709551605)
2766 23:21 < matches> But I'm pretty sure it should be -(83,9)
2767 23:21 < matches> + even
2768 23:22 < matches> (The least significant digit is first)
2769 23:23 < matches> Actually the plot thickens
2770 23:23 < matches> +45,10 - +128,0 = +18446744073709551533,9
2771 23:23 < matches> +45,10 - +128 = -83,18446744073709551605
2772 23:23 <@sulix> What is 0 + -1?
2773 23:24 < matches> +0 + -1 = -1
2774 23:24 <@sulix> I am pleased to hear that.
2775 23:25 < matches> +0 - +1 = -1
2776 23:25 <@sulix> -1 + 2?
2777 23:25 < matches> Now you're getting tricky...
2778 23:25 <@sulix> (As a maths major, I should know this)
2779 23:25 < matches> -1 - +2 = +1
2780 23:25 < matches> They're all single digit though
2781 23:26 <@sulix> Do you mean -1 + +2 = +1?
2782 23:26 <@sulix> Otherwise I'm slightly concerned.
2783 23:26 < matches> Uh yeah
2784 23:26 < matches> Sorry I have to change both the Arbint c(a + b) and the Debug
2785 23:26 < matches> Before copy pasting
2786 23:27 < matches> So I guess I should calculate if 45 + 10*2^64 - 128 == 18446744073709551533 + 9*2^64
2787 23:31 < matches> Expletive grenade
2788 23:31 < matches> I started Mathematica
2789 23:31 < matches> fglrx segfaulted
2790 23:31 < matches> startx
2791 23:32 < matches> -bash: /usr/bin/startx: No such file or directory
2792 23:32 < matches> Also now I am getting zenity int3 error messages all over the bottom line of my terminal
2793 23:33 < matches> Whyyy
2794 23:33 <@sulix> That's terrible.
2795 23:33 < matches> Why does this computer persecute me so
2796 23:34 < matches> I gave it a good life
2797 23:34 <@sulix> You should ask frames how he runs mathematica on the command line.
2798 23:34 < matches> I fed it lots of AC power
2799 23:34 < matches> I replaced that keyboard that I broke
2800 23:34 < matches> I only spilled coffee on it once
2801 23:34 < matches> I *tried* to use the drivers that weren't fglrx
2802 23:35 < matches> I give up, see you tomorrow. Try not to totally rewrite all the Arbint code :P
2803 23:35 < matches> I think the subtraction might actually work except for the resizing bit
2804 23:35 <@sulix> Nah, I'm calling it a night here too.
2805 23:36 <@sulix> I tried replacing the carry bit with the sign bit and all went to hell there, so I'm inclined to agree with you.
2806 23:37 < matches> Well using my calculator I get something with a 184 ish at the start and too many digits to display
2807 23:37 < matches> Hence the mathematica disaster
2808 23:40 < matches> Oh well, I got startx back by installing "xinit" so that's something
2809 23:41 < matches> I guess apt decided to remove it as part of apt-get autoremove in its infinite wisdom
2810 23:41 <@sulix> I have decided to never call apt-get autoremove.
2811 23:42 <@sulix> If at any point there are too many old packages for me to get things done, I'll just reinstall the OS entirely.
2812 --- Day changed Mon Jul 07 2014
2813 08:40 < matches> I try to avoid it, but sometimes apt won't let me do anything because it would break some dependency
2814 08:40 < matches> Although usually even after I run autoremove it complains
2815 08:41 < matches> Had to use aptitude to fix my "You can'd have kde-runtime!" issues
2816 08:45 < matches> Anyway wolfram alpha says that example +45,10 - +128 == +18446744073709551533,9
2817 08:45 < matches> is correct
2818 08:45 < matches> Checking the algorithm using 8 bits it should work too
2819 08:46 < matches> So it works as long as you supply the leading zeros...
2820 08:46 < matches> (On that particular case...)
2821 08:48 < matches> I hope it's not interpreting Arbint(128L) as Arbint(128u, ...)
2822 08:48 < matches> Nope sanity seems to prevail there
2825 08:50 < matches> First bug of the day: The subtracted argument needs to be resized
2826 08:50 < matches> As I am doing (45,10) - (128)
2828 08:59 < matches> I passed the realops test
2829 09:00 < matches> And it's also a lot faster now!
2830 09:00 < matches> Probably because the subtraction bug meant it was spending a long time in division to get the wrong result
2831 09:01 < matches> Yeah it should have been obvious division can't have been that slow for numbers starting at 64 bits since it does at most as many iterations as the number of bits
2832 09:01 < matches> Oh well
2833 09:02 < matches> Should probably not celebrate just yet, now the bezier looks really wierd again
2835 09:04 < matches> Actually compiling with Real == Rational<Arbint> might be a better test
2836 09:04 < matches> But it still passes
2837 09:05 < matches> In particular it passes with Arbints where int64_t fails horribly
2838 09:05 < matches> So something is right
2839 09:05 < matches> Just not all of it...
2840 09:08 < matches> The GPU rendering seems to workish
2841 09:09 < matches> We really quickly end up with Arbints of size 30 digits or so :S
2842 09:09 < matches> Which is big enough to give an overflow in doubles I think
2843 09:10 < matches> Rational::ToDouble can probably be written as double(integer division) + Rational(remainder, quotient).ToDouble()
2844 09:10 < matches> (With some things to stop it recursing infinitely)
2845 09:20 < matches> Or even double(integer division) + double(remainder)/double(quotient)
2846 09:20 < matches> I don't think any recursion is needed
2847 09:22 < matches> Oh, there's a "pow" call that is probably broken
2848 09:24 <@sulix> Yeah, I feel the growing quickly is just a fact of life.
2849 09:25 <@sulix> Perhaps changing the zooming to avoid coprimes or something.
2850 09:27 <@sulix> I'm pretty certain that about 3 seconds of zooming is enough to make us need precision and range to match measuring the size of the visible universe in Planck lengths, too.
2851 09:28 <@sulix> "New biggest Arbint of size 173"
2852 09:28 <@sulix> and counting...
2853 09:34 < matches> haha
2854 09:34 < matches> There are more issues I spotted with add and subtract
2855 09:34 < matches> Basically you have to add the leading zeros - all of them
2856 09:35 < matches> My patch only simulates adding one leading zero
2857 09:35 < matches> Actually maybe I don't need to add them all, just enough until the borrow becomes zero
2858 09:36 < matches> But whatever
2859 09:36 < matches> If you're left with a borrow you can't assume that subtracting it from the next digit won't cause a borrow as well
2860 09:37 < matches> Realops is not a good enough tester
2861 09:42 <@sulix> Surely if there's borrow bit, subracting it from zero will always give another...?
2862 09:42 < matches> You only need as many as there are digits in your number
2863 09:42 < matches> At most
2864 09:42 < matches> If you still have a borrow at the end
2865 09:42 < matches> That means the result is negative
2866 09:43 <@sulix> Yeah, that's what I thought: you need to cap the number of digits you add.
2867 09:44 < matches> The problem was the algorithm before didn't finish all the borrows
2868 09:44 < matches> eg: 100 - 2 would have become "-8"
2869 09:44 < matches> Because 2 only has one digit
2870 09:44 <@sulix> My powers of mathematics tell me that that is not quite correct.
2871 09:45 < matches> What, that our algorithm was wrong?
2872 09:45 <@sulix> No, that 100-2 is not -8
2873 09:45 < matches> Yes
2874 09:45 < matches> Yes that's the problem :P
2875 09:46 < matches> So basically the two numbers need to be the same size but since we have a constant argument I'm sticking in a "borrow_digits" vector if there is a borrow
2876 09:46 < matches> Addition also needs it for carry
2877 09:46 < matches> Damn
2878 09:47 < matches> I'd hoped that might fix the beziers not looking like beziers on the CPU bug
2879 09:50 <@sulix> I've pushed code to clear the carry/borrow flag before adding/subtracting.
2880 09:50 <@sulix> I don't think any bug is triggered, but there was the potential for off-by-one errors.
2881 09:52 < matches> Argh no
2882 09:52 < matches> Merge
2883 09:53 < matches> Yeah I'm pushing a fix for bugs that I don't think we've seen (yet)
2884 09:54 <@sulix> Out of curiosity, I tried removing the Simplify() calls from Rational.
2885 09:54 <@sulix> Relatedly, I've coined the term "NaN soup"
2886 09:54 < matches> Oh, clearing the carry/borrow flag might actually fix a bug with the code I just committed
2887 09:54 < matches> No wait it won't
2888 09:55 < matches> Eh it's still kind of dangerous to assume it isn't set to start with
2889 09:55 <@sulix> I think the flag was probably already cleared by the code the compiler generated, but that might not be the case if it's optimized.
2890 09:55 < matches> One of my early attempts had inline adc instructions in a for loop
2891 09:56 < matches> Which failed miserably because the for loop was clearing the carry flag :P
2892 09:57 <@sulix> For some reason "*=" segfaults if optimization is enabled, but I don't know why.
2893 09:57 < matches> Yuk
2894 09:58 <@sulix> Somehow we end up with "mul" as a null reference.
2895 09:58 < matches> Rataional::*= or Arbint?
2896 09:58 <@sulix> Arbint
2897 09:58 <@sulix> (So, also rational)
2898 09:58 <@sulix> Valgrind isn't picking anything up.
2899 09:58 < matches> null references are the worst
2900 09:59 < matches> Since I'm pretty sure the point of a reference is that it is never null...
2901 09:59 < matches> I'm sure this doesn'
2902 09:59 < matches> t stop people from deliberately creating null references
2903 09:59 < matches> And also checking that they are not null
2904 10:02 <@sulix> Tried compiling with clang. Nope.
2905 10:02 <@sulix> Apparently not even #include <iostream> compiles with clang.
2906 10:02 < matches> It might be worth implementing a non assembly version of those functions
2907 10:03 <@sulix> Yeah, that's always a good idea.
2908 10:03 <@sulix> The assembly ones are more fun, though.
2909 10:16 < matches> Hmm I should probably head towards the meeting
2910 16:23 < matches> I have pushed a Gmpint wrapper thing
2911 16:23 < matches> It mimics Arbint as closely as possible.
2912 16:24 < matches> Well by that I mean I just wrapped all the calls to mpz_add() etc in operators
2913 16:24 < matches> It doesn't have all of Arbint's more esoteric functions
2914 16:24 < matches> If we need bit shifts I think I can add them
2915 16:27 < matches> So, writing stuff...
2916 16:28 < matches> Running arbint_vs_gmpint.test in valgrind takes a fair while
2917 16:28 < matches> We will have the definitive answer of just how terrible our multiplication is compared to the Professional (TM) implementation
2918 16:29 < matches> I do have some good news, I think our string conversion might be better
2919 16:29 < matches> Well if you ignore the division
2920 16:31 < matches> *= is about 11 times slower
2921 16:32 < matches> Most of that is in std::vector::resize()
2922 16:32 < matches> But even just mul_digits is greater than the entire Gmpint::operator*=
2923 16:32 < matches> The plot thickens as I try division
2924 16:32 < matches> Which is a good opportunity to get food because it will probably take an hour
2925 16:34 < matches> Note: Don't disable Debug messages if you want time to get food :(
2926 16:35 < matches> And division is just under 100* slower
2928 16:36 < matches> 1000*
2929 16:37 < matches> So... if we change Rational<Arbint> to Rational<Gmpint>...
2930 16:41 < matches> We still get {nan,nan,nan,nan} eventually but it is generally less shoddy
2931 16:50 <@sulix> Cool: looking at this now.
2932 16:50 <@sulix> I'm digging through the gmp source code as well.
2933 16:51 <@sulix> They have a similar "divide by single int64_t" optimization.
2934 16:51 <@sulix> Also their code is 90% preprocessor macros.
2935 16:51 < matches> Welp
2936 16:51 < matches> At least we have a readable Arbint
2937 16:52 < matches> I don't know what use a readable but slow implementation is
2938 16:52 <@sulix> Ah: I see what they're doing: this is quite clever, particularly from a pure maths point of view.
2939 16:52 <@sulix> They've got a natural number implementation, and have then built their integer representation around that.
2940 16:53 <@sulix> The natural numbler implementation is just a huge set of directories with different assembly implementation.
2941 16:53 <@sulix> So there's an "x86_64" directory, and in that there's a bunch of assembly + a bunch of directories with optimized versions for different individual processor models.
2942 16:55 <@sulix> Also their assembly has lots of ASCII art comments.
2943 16:56 <@sulix> and macros.
2944 17:07 < matches> Yeah "just use GMP" is probably the answer
2945 17:08 < matches> Their Makefiles are pretty intimidating
2946 17:22 < matches> I like that they seem to store the sign as part of the size
2947 17:22 < matches> If something has a negative size it is negative and has |size| digits
2948 17:34 < matches> I guess I will try and write some sort of report about how we implemented Arbitrary Integers but they are terrible compared to existing implementations :P
2949 --- Day changed Thu Jul 24 2014
2950 14:44 < matches> So I was going to work on the project but existential dread
2951 14:45 < matches> About whether my major exists
2952 14:45 < matches> Do I exist?
2953 14:49 <@sulix> Are you thinking? Because cogito ergo sum.
2954 14:50 < matches> I'm not sure I was thinking when I picked this major...
2955 14:50 <@sulix> I did some project code yesterday and then a bug I thought I'd fixed reappeared so I got distracted failing to fix that.
2956 14:50 < matches> Haha
2957 14:53 < matches> Should I upset everyone and recommend freshers for wheel again...
2958 14:54 <@sulix> I'm all for it, but I think the consensus was that we need to make them actually do wheel-y projects first.
2959 14:55 < matches> That seems kind of hypocritical though
2960 14:55 < matches> Because nearly none of active wheel has actually done wheel-y projects
2961 14:55 < matches> Certainly not before getting on wheel
2962 14:56 <@sulix> My current random guess is that the problem is that people used to fix the desktops and stuff, and now everyone has their own laptops to break.
2963 14:56 < matches> But I'll be quiet or people might decide I need to be removed due to lack of doing useful things
2964 14:56 < matches> Yeah
2965 14:56 < matches> That too
2966 14:57 <@sulix> I'll definitely bring it up at the meeting.
2967 15:02 <@sulix> So which CoderDojo forms do I need to fill out?
2968 15:02 < matches> Ooh!
2969 15:02 < matches> http://coderdojowa.org.au/volunteer
2970 15:02 < matches> This one
2971 15:03 < matches> But now you've said that I'm already adding you to the mailing list...
2972 15:04 < matches> There's a thing on Saturday in 2.01 in CS at 12:00pm
2973 15:04 < matches> I hope people actually show up because we are pretty short on presenters
2974 15:04 <@sulix> Yeah, I've got a programming competition then. I'm trying to work out how much of the schedule for the competitions exists.
2975 15:06 <@sulix> Okay, apparently there are programming competitions every saturday in August, which will be fun.
2976 15:07 <@sulix> Although half of them are "details TBD," which sounds ominous.
2977 15:07 <@sulix> Also there is a round 2 and a round 4 but no round 3.
2978 15:11 <@sulix> Okay: it looks like the only weeks I don't have programming competitions on are the last 3 on the form.
2979 15:13 <@sulix> Also I suspect they're running out of names for the programming competitions, because this Saturday's is called the "South Pacific Winter Programming Carnival".
2980 15:21 < matches> Haha
2981 15:27 <@sulix> Form submitted. Sorry for the snarkiness.
2982 15:40 < matches> Brilliant
2983 15:40 < matches> You can do a C or C++ workshop or something :P
2984 15:40 < matches> Or just talk about Commander Keen that'll work
2985 15:40 < matches> Or "Why Javascript is awful and you should forget all the lessons"
2986 15:42 <@sulix> "Intro to DOS programming." :P
2987 18:38 < matches> I did a sort of half hearted attempt at writing more about Arbints
2988 18:39 < matches> Maybe I'll try put fonts in
2989 18:39 < matches> That seems vaguely like not what I am supposed to be doing right now :P
2990 18:43 < matches> There's that virtual FPU sitting there doing nothing
2991 18:43 < matches> That I promised to do things with in my lit review
2992 18:43 < matches> That Tim is marking
2993 18:43 < matches> When I haven't actually done anything with it and he knows it...
2994 18:44 < matches> I can't help but feel like we need a more impressive thing to zoom in on
2995 18:44 < matches> Or even a way to draw things once we have zoomed in
2996 18:46 < matches> Does "We implemented Arbitrary Precision Integers but GMP did it better" count as research?
2997 19:13 < matches> Do we have a memory leak?
2998 19:13 < matches> I've been running it for a while and things are slowing down
2999 19:25 <@sulix> matches: With GMP or just doubles?
3000 19:27 <@sulix> My quick check has us not leaking memory with doubles.
3001 19:27 <@sulix> Well, X leaks memory and the nVidia driver leaks memory, but we're fine.
3002 19:35 < matches> I was actually running with singles :S
3003 19:35 < matches> See push to documents repo
3004 19:35 < matches> There is a pdf
3005 19:35 < matches> I did a thing
3006 19:38 < matches> I'm basing the assumption that x86-64 is IEEE compliant on the fact that it passed the "paranoia" program
3007 19:39 <@sulix> Yeah, x86_64 is IEEE compliant.
3008 19:39 <@sulix> x86_32 is "mostly" IEEE compliant if I recall.
3009 19:41 < matches> Well a picture tells a thousand words
3010 19:41 < matches> So I think I wrote 8000 words today
3011 19:41 < matches> Progress!
3012 19:43 <@sulix> http://cgit.freedesktop.org/mesa/mesa/tree/src/mesa/drivers/dri/i965/brw_defines.h?id=9d6166880da83887e3246fb4498c3a07d979cc3b#n162
3013 19:43 <@sulix> I'll see if I can find where they actually set it to non IEEE.
3014 19:43 < matches> Oh I was going to say fglrx did different things to nVidea but as I don't have nVidea that's difficult to do
3015 19:45 <@sulix> Although there's this: http://lists.freedesktop.org/archives/mesa-dev/2013-July/041555.html
3016 19:45 <@sulix> Yeah, nVidia, Intel and fglrx all seem to do different things.
3017 19:45 <@sulix> fglrx does the strangest things.
3018 19:45 <@sulix> nVidia does the most consistant things.
3019 19:45 <@sulix> Intel sits nicely in the middle.
3020 19:46 < matches> If we can get four screenshots of the different things at the same view bounds that might be useful
3021 19:47 < matches> Also working out more about what the jagged edges implies about the precision/rounding might be helpful
3022 19:47 < matches> All I've got is "This is clearly not a circle"
3023 19:48 < matches> So it's still doing that with the quad trees enabled, so I assume the quad trees aren't amazingly quadifying everything yet :P
3024 19:48 <@sulix> The quadtrees are basically doing nothing but occasionally causing bugs.
3025 19:48 <@sulix> http://davidgow.net/stuff/ipdf-nvidia.png
3026 19:48 < matches> That is progress at least
3027 19:49 < matches> Thanks
3028 19:49 < matches> I will update the others to be the same view bounds
3029 19:49 < matches> I *think* we have code to set the view bounds at start?
3030 19:51 <@sulix> http://davidgow.net/stuff/ipdf-intel.png
3031 19:51 < matches> Haha
3032 19:51 < matches> Well it's really obvious that that's different
3033 19:52 < matches> Can you rerun it with -b 0.0869386 0.634194 2.63295e-07 2.63295e-07
3034 19:52 < matches> Obsessive compulsive...
3035 19:52 < matches> Must all be same view bounds...
3036 19:53 < matches> I'd run it on my other laptop with intel integrated graphics except the keyboard still doesn't work
3037 19:55 < matches> Actually don't bother
3038 19:55 <@sulix> http://davidgow.net/stuff/ipdf-nvidia1.png
3039 19:57 <@sulix> http://davidgow.net/stuff/ipdf-intel1.png
3040 19:59 < matches> oah wierd stuff is happening with the quad tree
3041 19:59 < matches> There is a big circle and a little circle
3042 19:59 < matches> Is that supposed to be here...
3043 20:00 < matches> The distance between them is not constant :S
3044 20:09 <@sulix> Yeah, that doesn't happen on intel and is the bug I've been hitting my head against.
3045 20:09 <@sulix> Pretty certain I'm trying to render one more object than there actually is somewhere, maybe corruping memory in the process.
3046 20:32 < matches> Ok so it turns out the CPU is actually about as terrible as the GPU at those view bounds when you replace the "double" with "float" in the Circle Renderer :S
3047 20:32 <@sulix> That's what I expected.
3048 20:32 <@sulix> It looks like the nVidia one, right?
3049 20:32 < matches> But it does slightly different wrong things!
3050 20:34 < matches> It looks similar-ish
3051 20:35 < matches> It is blocky as opposed to zig zaggy
3052 20:39 < matches> As in it doesn't look as whack as intel
3053 20:43 <@sulix> Hmm... I'm not sure what I'm doing wrong, but I can't see any artefacts at all with CPU rendering w/ those bounds.
3054 20:55 < matches> They are hard coded as doubles
3055 20:55 < matches> Not floats
3056 20:55 < matches> Or reals rather
3058 20:56 <@sulix> This explains much.
3059 21:08 < matches> I have pushed a thing
3060 21:09 < matches> It almost sounds like a real paper
3061 21:09 < matches> Until you realise all it is is "we drew some circles and they look different"
3062 21:09 < matches> Also your screenshots had some kind of crazy blue glowy border
3063 21:10 <@sulix> Yeah, that's the KDE window shadow.
3064 21:10 < matches> Fancy
3065 21:10 <@sulix> It used to make the nVidia driver corrupt screenshots, but it seems to work now.
3066 21:11 < matches> I'm pretty pleased with that 4 way comparison figure...
3067 21:12 < matches> "One of these things is not like the others..."
3068 21:12 < matches> *cough* intel
3069 21:12 <@sulix> The conclusion is brilliant.
3070 21:15 < matches> If we assume the nVidia and x86-64 figures are what things are supposed to look like
3071 21:15 < matches> fglrx tries really hard
3072 21:15 < matches> But doesn't quite make it
3073 21:16 < matches> (I'm pretty sure that's just a particularly good view for it)
3074 21:16 < matches> (If you move it around it goes insane)
3075 21:16 < matches> I can respect the intel shader
3076 21:16 < matches> It isn't afraid to blatantly disregard the rules
3077 21:17 < matches> intel driver rather
3078 21:17 < matches> Not sure why I put "shader" there
3079 21:17 < matches> This has not been as productive as I hoped
3080 21:18 < matches> Still
3081 21:18 < matches> We finally have something written that Tim can pass judgement on
3082 21:18 < matches> He's still in the country right?
3083 21:19 < matches> He might want to finish passing judgement on my literature review first :S
3084 21:21 <@sulix> I think he's still in the country, but don't hold me to that.
3085 21:22 < matches> Ok, so if you translate around with the CPU things don't go insane, but they do on the GPU. That might be caused by something else though.
3086 21:23 < matches> I'll be at University tomorrow
3087 21:24 <@sulix> I might head in, too, then.
3088 21:25 <@sulix> Do things still "go insane" on the GPU with CPU-side coordinate transforms.
3089 21:26 < matches> Yeah
3090 21:26 < matches> It looks like there is a tear
3091 21:27 < matches> So you get this rectangle pattern
3092 21:27 < matches> If you move it around on the CPU it maintains its shape
3093 21:28 <@sulix> I think that fglrx (or maybe just the AMD hardware) calculates the coordinates per-triangle rather than per-vertex or something.
3094 21:28 < matches> Under fglrx the bottom part of it sort of maintains its shape but there is a big diagonal line and the stuff above that changes
3095 21:28 <@sulix> That's pretty weird.
3096 21:29 < matches> Also the bottom part doesn't have concave bits but the top occasionally gets one
3097 21:29 < matches> concave/overhanging whatever
3098 21:29 <@sulix> The whole thing maintains its shape on nvidia (and even intel)
3099 21:29 < matches> Well, the bottom half (and also the CPU/nVidia entire thing) looks kind of like a stair case
3100 21:29 < matches> The top bit gets all these sticky out bits and overhangs
3101 21:30 < matches> Which brings us to our next paper
3102 21:30 < matches> The geology of fglrx
3103 21:30 <@sulix> Intel also does the "staircase" on the other side of the circle.
3104 21:30 < matches> On the other side...
3105 21:30 < matches> Hmm the plot thickens
3106 21:31 < matches> Oh well I need to sleep
3107 21:31 < matches> Why do I feel like I have actually lost sleep over the holidays...
3108 21:32 < matches> I am not ready for semester to start :(
3109 21:32 <@sulix> I know exactly what you mean.
3110 21:33 < matches> I seem to have been roped into unpaid work with physics
3111 21:34 <@sulix> Oh dear. More lab demonstrating or something more interesting?
3112 21:34 < matches> Hopefully if I visit ECM they won't make me do GENG5505 yet
3113 21:34 < matches> Fixing my honours experiment I think...
3114 21:35 <@sulix> Another pipe corroded through?
3115 21:35 < matches> Haha
3116 21:35 < matches> They were wondering where all the electronics went
3117 21:35 < matches> (I have most of it)
3118 21:35 < matches> (Also it's no longer functional)
3119 21:36 < matches> (I may have taken some of it apart...)
3120 21:36 < matches> (Although really the sensible option would have been to burn it with fire)
3121 21:37 < matches> Goodnight anyway
3122 21:37 <@sulix> I'm required to "correct" anything they want to change with this project after submitting it.
3123 21:37 <@sulix> On the morrow, then!
3124 --- Day changed Fri Jul 25 2014
3125 11:24 < matches> I'm at UCC
3126 11:39 <@sulix> Okay, I'll show up once I've sorted out this acedemic record stuff.
3127 12:05 < matches> I went to ECM but apparently just a few other people are having difficulties...
3128 18:24 < matches> So steam browser supports html5 canvas but not keyboard events
3129 18:24 < matches> This seems to happen with a lot of browsers
3130 18:25 < matches> We support html5 canvas (which is pretty much solely designed for web based games)
3131 18:25 < matches> But we don't support any way to actually pass input to the page that doesn't suck
3132 18:25 < matches> (I'm looking at you Safari)
3133 18:31 < matches> It sort of spoils the "HTML5 is AWESOME" message the HTML5 people are aiming for
3134 18:31 < matches> The best bit of html5 is <canvas> and that you no longer need to use html because you can just draw everything in the canvas...
3135 18:32 < matches> Which is scarily similar to using PostScript
3136 --- Day changed Mon Jul 28 2014
3137 10:05 < matches> In 2.07 now
3138 10:07 < matches> You removed SDL from contrib
3139 10:07 < matches> So now it won't compile on the lab machines
3140 10:07 * matches gets out the laptop then
3141 --- Day changed Tue Jul 29 2014
3142 09:59 < matches> I seem to have this bizarre illness
3143 10:00 < matches> Where half my head feels fine
3144 10:00 < matches> And the other half feels like I've been trying to read brainfuck
3145 10:00 < matches> I hope you don't catch it
3146 10:00 < matches> The side that is sick is the side that was closest to the group hug
3147 10:00 < matches> Coincidence!?
3148 10:00 < matches> I think not!
3149 10:08 < matches> You know how there's this theory about the two halves of the brain being semi independent and sort of subconsciously able to think by themselves
3150 10:08 < matches> I think the left side of my brain is dead
3151 10:08 < matches> It's not responding to pings
3152 10:09 < matches> I think the project working side might be the alive one though, so we shall see if I can actually do something useful
3153 --- Log closed Tue Jul 29 23:13:08 2014
3154 --- Log opened Wed Jul 30 12:03:38 2014
3156 12:03 -!- Irssi: #ipdf: Total of 2 nicks [1 ops, 0 halfops, 0 voices, 1 normal]
3157 12:03 -!- Irssi: Join to #ipdf was synced in 0 secs
3160 --- Day changed Sun Aug 03 2014
3161 12:43 < matches> So I found my moronic mistake with the virtual FPU and it seems to actually work
3162 12:44 < matches> I think it's even slower than using Rational<Arbint>
3163 12:44 < matches> Unless everything's slow because of the QT now
3164 12:45 < matches> There's an awful lot of "Rendering QT node" debug
3165 12:46 < matches> No its the VFPU that makes sense
3166 12:47 < matches> Since every floating point operation now requires writing 18 hex characters over a unix domain socket
3167 12:47 < matches> Hooray VHDL
3168 12:48 < matches> (I'm sure if you have $$$$ and like Enterprise(TM) software VHDL is great)
3169 12:48 < matches> I'll just do some miscellaneous things and hopefully some of them will be useful
3170 12:48 < matches> Bye
3171 14:14 < matches> So with the VFPU and doing everything on the GPU we have 2.2 FPS
3172 14:14 < matches> It seems to be able to cope with rectangles on the CPU
3173 14:15 < matches> The circle may present some difficulties
3174 14:16 < matches> It would be have been pretty cool to change the vfpu and see if anything makes it look like intel
3175 14:16 < matches> But ultimately useless I suppose
3176 14:37 < sulix> There might be some info in the hardware docs, though they're pretty incomprehensible: https://01.org/linuxgraphics/documentation/2013-intel-core-processor-family
3177 14:38 < sulix> AMD also have manuals: http://developer.amd.com/resources/documentation-articles/developer-guides-manuals/
3178 14:38 < sulix> nVidia are sworn to secrecy.
3179 --- Day changed Tue Aug 05 2014
3180 12:38 < matches> I'm looking at the atril source
3181 12:38 < matches> To see if I can compile it without the oppressive 400% max zoom
3182 12:39 < matches> It is rather intense
3183 12:39 < matches> I think the zooms are set depending on the language?
3184 12:41 < matches> There are po files which are definitely not position independent object files
3185 12:43 < matches> #define ZOOM_MAXIMAL(zoom_levels[n_zoom_levels - 1].level)
3186 12:43 < matches> The plot thickens!
3187 12:45 < matches> Heh this is actually sort of making sense
3188 12:45 < matches> They have a "cut-n-paste/zoom-control" directory
3189 12:48 < matches> I'm confused as to how the maximum zoom is 6400% and yet my version will only zoom to 400%
3190 12:48 < matches> But I guess I shall compile it from source and see what actually happens
3191 12:51 < matches> They also have a script to configure their configure script...
3192 12:51 < matches> Oh dear xml is involved
3193 12:53 < matches> There needs to be a way to save the status of configure so you don't have to rerun it from scratch every time you fix an error
3194 12:56 < matches> Our document viewer may only be able to render beziers and circles
3195 12:56 < matches> But at least it doesn't have any dependencies!
3196 12:56 < matches> Well many
3197 12:56 < matches> I suppose since SDL is in contrib we should put GMP in contrib too
3198 12:57 < matches> But yes, I don't care about caja plugins dammit I want to zoom on a pdf
3199 13:00 < matches> "Warning: Comparison between pointer and integer"
3200 13:00 < matches> Aaaah
3201 13:00 < matches> It compiles...
3202 13:01 < matches> But where did it put the binary...
3203 13:02 < matches> Oh great it expects things to be in /usr
3204 13:03 < matches> I guess I will run "install-sh" and hope I will still have a usable pdf viewer
3205 13:07 < matches> There's a LOT of G_ and g_ variables
3206 13:07 < matches> Actually functions
3207 13:07 < matches> I think g_ does not mean global in this context
3208 13:07 < matches> Maybe
3209 13:11 < matches> So it's not just a matter of ZOOM_MAXIMAL being redefined to ridiculously huge numbers having any effect
3210 13:11 < matches> I guess they use some other library that has zooming
3211 13:13 < matches> It somehow detects what the maximum zoom should be based on the document as well
3212 13:13 < matches> This is just over engineering!
3213 13:19 < matches> ev_view_can_zoom_in (EvView *view)
3214 13:19 < matches> return view->scale * ZOOM_IN_FACTOR <= ev_document_model_get_max_scale (view->model);
3215 13:19 < matches> Let
3216 13:19 * matches casually makes it return true
3217 13:19 < matches> Also wtf
3218 13:19 < matches> They have a "gboolean" type
3219 13:19 < matches> Is "bool" inferior?
3220 13:19 < matches> What more do you need than true and false
3221 13:20 < matches> Is bool used as a variable name for something
3222 13:20 < matches> sigh
3223 13:24 < matches> No ok it still won't let me zoom in more
3224 13:25 < matches> g_return_if_fail in the zoom functions
3225 13:25 < matches> They don't look important
3226 13:28 < matches> There are a surprising amount of places with a "if zoom or scale is bigger than the maximum, return"
3227 13:28 < matches> Like, are three checks really necessary
3228 13:29 < matches> I think you can set model->sizing_mode to "EV_SIZING_FREE" but I have no idea what that will do
3229 13:29 < matches> So I'll just continue commenting out anything that looks like it will return early from the zoom process...
3230 13:30 < sulix> Sounds like a plan!
3231 13:31 < matches> The use of g_ for function names makes me want to stab someone :P
3232 13:31 < matches> I assume it's related to gtk or maybe just gnome
3233 13:32 < matches> Also about half their doubles are "gdouble"
3234 13:32 < matches> And the other half are "double"
3236 13:32 < sulix> gtk had their own crazy, let's implement everything from scratch idea.
3237 13:32 < sulix> They have their own "classes" in C, IIRC.
3238 13:32 < matches> It's pretty object orientated yes
3239 13:34 < matches> Ok I seem to now be able to zoom in a *bit* more and then it just stops drawing things
3240 13:35 < matches> No wait it's just hopelessly slow
3241 13:37 < matches> It also zooms in on the wrong spot and then draws it and then moves the view back to the mouse and draws again
3242 13:38 < matches> But yes, it doesn't cope well with zooming at all
3243 13:38 < matches> I was kind of hoping it would work
3244 13:39 < matches> Maybe if I use a simpler document
3245 13:43 < matches> Congratulations anyway
3246 13:43 < matches> We can zoom further on a circle than atril can
3247 13:44 < matches> Unfortunately the reasons atril can't zoom are probably not related to precision
3248 13:44 < matches> Maybe I need to find some "if zoom > thing don't draw anything" and comment those out too
3249 13:44 < matches> Sigh
3250 14:07 < matches> It is resistant to my valgrind attempts
3251 14:07 < matches> It seems to have spawned 9 processes
3252 14:10 < matches> Oh "atril" is actually a series of hideous bash scripts that load libraries or something
3253 14:12 < matches> You know
3254 14:12 < matches> I'm sure if we tried
3255 14:12 < matches> We could make a less shitty pdf viewer out of our program
3256 14:29 < matches> Interesting
3257 14:30 < matches> "ev_document_render" got called 10.6 Billion times
3258 14:30 < matches> I don't think I changed the view that many times...
3259 14:32 < matches> Ok, so just working on our own viewer is probably better at this point than working out why atril has a heart attack if you zoom
3260 14:33 < matches> I'm going to pretend this wasn't a waste of time...
3261 14:41 < matches> "existing pdf viewers cap the view because otherwise they break due to bugs that aren't actually related to precision"
3262 15:43 < matches> So I suspect the GPU Bezier rendering is buggy
3263 15:44 < matches> Or maybe it's a quad tree thing
3264 15:45 < matches> Or both...
3265 15:46 < matches> I'm pretty sure the appearance of an individual bezier should not depend on what other beziers are in the document :P
3266 15:47 < matches> Yeah there's wierd quad tree shit going on
3267 15:48 < matches> I'm going to try and add some sort of glyph/font like thing
3268 15:49 < matches> It's either that or create hand drawn beziers in the current view
3269 15:50 < matches> But I feel it desperately needs an example of being able to actually zoom in and create something at an "arbitrary" level
3270 18:54 < matches> So we sort of have a very broken, very badly written, SVG parser
3271 18:55 < matches> A bunch of bugs showed up in the bezier rendering
3272 18:56 < matches> Although it might just be fglrx
3273 22:23 < sulix> The GL code is now consistantly broken, rather than sporadically broken.
3274 22:24 < sulix> Trying to render SVGs with it might count as modern art, though.
3275 22:39 < sulix> Man, the buggier the bézier code gets, the more beautiful its output.
3276 22:53 < sulix> Voilà: http://davidgow.net/stuff/ipdf-koch-svg.png
3277 23:11 < sulix> And on the CPU: http://davidgow.net/stuff/ipdf-koch-svg1.png
3278 --- Day changed Wed Aug 06 2014
3279 11:00 < matches> That looks amazing!
3280 11:04 < matches> I'm sorry
3281 11:04 < matches> All I seem to do is add bugs :S
3282 11:05 < matches> And XML
3283 11:13 < sulix> I suspect you removed a lot of bugs with the -DQUADTREE_REMOVED. :P
3284 11:14 < matches> It's the bit where parts of my Bezier code have been wrong for more than a month that make me feel guilty :S
3285 11:14 < sulix> The GPU code was wrong in the same way.
3286 11:15 < sulix> Also it used to corrupt all of the memory.
3287 11:23 < matches> So you're not using the koch.svg file?
3288 11:23 < matches> That breaks for me
3289 11:23 < matches> Which isn't surprising considering the svg parsing code
3290 11:24 < sulix> I was using koch1.svg from your Lit Review
3291 11:25 < matches> Oh right
3292 11:25 < matches> koch.svg isn't actually a valid svg file :S
3293 11:26 < sulix> I tried it with a random, complex SVG from the internet, and segfault.
3294 11:27 < matches> Haha
3295 11:27 < matches> Yeah it works
3296 11:27 < matches> We can zoom in further than atril!
3297 11:28 < sulix> Victory!
3298 11:34 < matches> So I was trying to think about shading things... and stroke thickness... and all those other things in SVG...
3299 11:34 < sulix> Yeah, I thought about those.
3300 11:35 < sulix> I came to the conclusion that I should probably try to make the quadtree work first.
3301 11:37 < matches> Probably :P
3302 11:58 < sulix> btw, your computer's clock is slightly out.
3303 11:58 < sulix> alternatively there is a lot of time travel going on
3304 11:58 < sulix> (secretly, I'm hoping it's the latter)
3305 11:59 < sulix> (im my mind, I'm good enough to have bugfixed the SVG code before it was written)
3306 17:10 < matches> Yeah I removed ntpd because my boot kept stalling at it for several minutes
3307 17:17 < matches> The GPU Bezier renderer seems a lot more conservative about how many lines it uses
3308 17:18 < matches> I thought I remembered just copying the GPU algorithm for the CPU but maybe I didn't
3310 17:38 < matches> Some of the shader stuff confuses me...
3311 17:38 < matches> Like: "pixize = vec2(pixel_w/bounds_w, 100*pixel_h/bounds_h)"
3312 17:39 < matches> Random factor of 100 ?
3313 17:39 < matches> As far as I can tell pixel_w,pixel_h is always 640,480
3314 17:57 < matches> Hmm
3315 17:57 < matches> But the window isn't 640,480
3316 17:57 < matches> Wait
3317 17:57 < matches> Is it
3318 17:58 < matches> Yeah
3319 17:58 < matches> The screen is 800 x 600
3320 17:58 < matches> That is only slightly confusing...
3321 17:58 < matches> Should I change all the 640,480 to 800,600 ...
3322 17:59 < matches> Or will this break things
3323 18:01 < matches> We are actually passing the width and height of the viewport around so I figure that should probably be used instead of 640 and 480
3324 18:02 < matches> Also I suspect we want just a few more lines on the beziers since they often start looking more like trapezoids
3325 18:04 < matches> But only if they are big to begin with
3326 18:05 < matches> We can probably also do a "if x1,y1 == x2,y2 just use one line"
3327 18:06 < matches> Or alternately actually implement lines as distinct from beziers but that's silly :P
3328 18:06 < matches> If I can work out how to do shading then we won't even need rectangles and circles any more
3329 18:06 < matches> I should probably focus on precision of things but I sort of want to be able to draw something cool first
3330 18:15 < matches> Oh damn merge
3331 19:19 < matches> So a series of curveto commands isn't just a series of individual beziers for each three points
3332 19:20 < matches> I guess it's time to actually read the SVG spec a bit
3333 19:20 < matches> Oh maybe they are cubic beziers
3334 19:21 < matches> I thought they were just quadratic unless you used a special command
3335 21:08 < matches> The order of your coefficients in the bezier geometry shader seems reversed...
3336 21:08 < matches> But maybe that's because I reversed it in the CPU first...
3337 21:08 < matches> I don't know
3338 21:08 < matches> I will leave it in this order and hope it doesn't break
3339 21:08 < matches> (I'm making the beziers cubic because it seems like a thing we want)
3340 21:08 < matches> And is required for SVG paths
3341 21:27 < matches> Hmm
3342 21:27 < matches> SVGs also have quadratic beziers and as much as setting (x3,y3) == (x2,y2) *almost* looks the same...
3343 21:29 < matches> Close enough I guess
3344 21:30 < matches> I don't particularly want to add seperate objects for quadratic beziers since I've only ever seen cubic ones in SVGs anyway
3345 21:49 < matches> http://szmoore.net/ipdf/svgshape.png
3346 21:49 < matches> Woo!
3347 21:50 < matches> I'm actually pretty excited about having a semi usable svg viewer that can zoom, if not indefinitely, a lot further than any of the open source image viewers will let you zoom :P
3348 21:51 < matches> Shading... that's hard
3349 21:51 < matches> Stroke thickness is kind of hard too...
3350 21:51 < matches> Blergh
3353 --- Day changed Thu Aug 07 2014
3354 21:24 < sulix> So it turns out ttf fonts were not very difficult to add at all: http://davidgow.net/stuff/ipdf-ttf.png
3355 21:42 < sulix> And fixed the glyphs.svg rendering: http://davidgow.net/stuff/ipdf-glyphs.png
3356 22:06 < sulix> So all of those Béziers make Gmpint very slow.
3357 22:07 < sulix> It also runs out of memory very quickly.
3358 22:11 < sulix> About 30 second of dragging an svg around uses ~15G of ram.
3359 22:13 < sulix> It takes (with Rational<Gmpint> CPU rendering) ~10 G to render the default frame of glyphs.svg
3360 22:18 < sulix> Okay, it just broke a computer w/ 32 GB of ram.
3361 22:20 < sulix> Code is pushed, btw, for your computer-wreckingly-awesome enjoyment.
3362 22:32 < sulix> I tried fixing Rational::ToDouble() and using GPU w/ CPU coordinate transform.
3363 22:32 < sulix> Still runs out of memory.
3364 22:33 < sulix> But it works, even if it's still a tad slow.
3365 22:45 < sulix> Okay, memory use is not a problem at all if we delete out GMP integers after we're finished with them.
3366 22:46 < sulix> (There was a TODO: to that effect)
3367 --- Day changed Fri Aug 08 2014
3368 16:11 < matches> Wait you just did the things *I* was supposed to do for this week :P
3369 16:11 < matches> I hope you don't want me to fix quadtrees...
3370 16:13 < matches> Ah sorry about the GMP
3371 16:14 < sulix> I gave the QuadTrees another go as well, but random half-letter-"b"s were everywhere.
3372 --- Day changed Mon Aug 11 2014
3373 10:56 < matches> Ok I think I need to throw all design principles out the window adding a keyboard handler
3374 10:57 < matches> I know you were keen on having a mouse handler independent of the Screen class
3375 10:57 < matches> But the Screen class detects the events
3376 10:57 < matches> It really makes sense for the event handlers to just be member functions
3377 10:57 < matches> Maybe virtual in the unlikely event that there are ever different types of Screen
3378 10:58 < matches> Probably the View should handle the events
3379 10:58 < matches> Then we wouldn't have this convoluted thing where View has a reference to a Screen but Screen has a pointer to a View...
3380 10:59 < matches> But I need to let that go and actually do useful things
3381 10:59 < matches> So KeyboardHandler is now a member of Screen
3382 10:59 < matches> Delicious spaghetti
3383 11:02 < sulix> I think I can bring myself to forgive you. :)
3384 --- Day changed Tue Aug 12 2014
3385 13:48 < matches> I can draw rabbit_simple.svg
3387 13:48 < matches> Almost
3388 13:50 < matches> IT IS BEAUTIFUL
3389 13:52 < matches> Wait I think quad trees are enabled
3390 13:53 < matches> Nope they aren't
3391 13:53 < matches> Oh wel
3392 13:54 < matches> http://szmoore.net/ipdf/infinite-precision-document-FOX.png
3393 13:54 < matches> That didn't take very long
3394 13:55 < matches> I suspect the wierd bits in the wrong spot are because there are translations applied to things
3395 13:55 < matches> Because inkscape
3396 13:57 < matches> Hmm yeah
3397 13:58 < matches> I wonder why there are random straight lines between things though
3398 13:58 < matches> They are filled regions of the same colour and no stroke but obviously we just read the beziers and draw the outlines
3399 14:50 < matches> Dammit SVG
3400 14:50 < matches> So the y coordinate of text refers to the bottom
3401 14:50 < matches> The y coordinate of *everything* *else* refers to the top
3402 14:51 < matches> It's alright this is ipdsvg we aren't constrained by stupid standards
3403 14:59 < matches> Ergh somehow fonts are broken in eye of mate...
3404 14:59 < matches> How did I even...
3405 15:00 < sulix> Fonts are usually handled with y=0 being the "baseline" of characters.
3406 15:10 < matches> There are a few wierd things going on with straight lines
3407 15:11 < matches> Hmm only at the default zoom
3408 15:11 < matches> Horizontal lines on both cpu and gpu are kind of wobbly
3409 17:23 < matches> Ok I am suffering from an attack of matrix algebra
3410 17:23 < matches> It was bound to happen sooner or later I guess
3411 17:24 < matches> The spaghetti is cooking nicely
3412 17:26 < matches> The fact that our document is all in GL coordinates and SVGs are not is causing me way more confusion than it really should be
3413 17:27 < matches> All I need to do is set an initial transformation matrix then everything else should Just Work
3414 17:27 < matches> Of course it does help if you actually use matrices instead of Rect's
3415 17:29 < matches> Also I think functions that modify arguments passed by references is one of the things that tpg hates
3416 17:29 < matches> But there are a lot of them here
3417 17:29 < matches> They are so convenient!
3418 22:03 < matches> If I can get transforms and groups actually working properly, we can probably hack some sort of recursive thing together and use view reparenting
3419 22:03 < matches> Somehow
3420 22:05 < matches> I'm thinking putting the fox in the rabbit's eye and so on recursively would actually be pretty damn awesome
3421 22:05 < matches> It is easy to say these things though...
3422 22:06 < sulix> So, I have view reparenting "working" in the quadtree code.
3424 22:06 < matches> Cool
3425 22:06 < sulix> (The rest of the quadtree doesn't work, so the point is somewhat moot, though)
3426 22:06 < matches> Wait you already had that working, but you have rendering bugs
3427 22:06 < matches> Yeah
3429 22:08 < sulix> I will push that now, actually.
3430 --- Day changed Wed Aug 13 2014
3431 00:34 < matches> Yes
3432 00:34 < matches> I have defeate
3434 00:34 < matches> Basic matrices
3435 00:36 < matches> So it is pretty ugly and inefficient but meh
3436 00:36 < matches> We have translate, scale and matrix
3437 00:37 < matches> skew will be pretty easy to add but probably useless
3438 00:37 < matches> rotate is a bit harder
3439 00:38 < matches> Also the skewing operations obviously don't work on rectangles
3440 00:38 < matches> Or anything that isn't defined in terms of bezier paths I guess
3441 00:38 < matches> But translating and scaling will
3442 00:40 < matches> Ok, so SVG has a "defs" thing that lets you define groups without drawing them
3443 00:41 < matches> And a "use" thing that lets you insert a group
3444 00:41 < matches> Actually element in general
3445 00:41 < matches> Doesn't have to be a group
3447 00:41 < matches> I wonder if doing recursive magic with that works
3448 00:48 < matches> Not in normal svg viewers it would seem
3449 00:49 < matches> Hmm
3450 00:51 < matches> (eom:7883): librsvg-WARNING **: Circular SVG reference noticed, dropping
3451 00:51 < matches> That's just boring!
3452 00:52 < matches> So I think we will need some fairly major changes to our Document structure to get much more SVG stuff working
3453 00:53 < matches> I wonder if we actually need to store a DOM if we want that to work
3454 00:53 < matches> Also I thought I fixed transformations but they still break for fox-vector.svg :(
3455 00:58 < matches> I think "broken" will probably be the most commonly occuring word in our git commit messages
3456 11:05 < matches> Oooh
3457 11:06 < matches> svg-tests/recursive.svg is slightly recursive viwed in firefox
3458 16:10 < sulix> The QuadTree works again!
3459 16:10 < sulix> (I have finally worked out how to multiply by two, it seems...)
3460 16:11 < sulix> Also s/ again//
3461 17:23 < matches> Project complete!
3462 18:05 < sulix> Man, nVidia has an OpenGL extension where you can just pass the "d" attribute of SVG paths in and have it render them for you.
3464 18:05 < matches> Welp
3465 18:07 < sulix> It's like: glPathStringNV(path, GL_PATH_FORMAT_SVG_NV, strlen(d), "M100,180, L40,10, etc...")
3466 18:08 < matches> Presumably you could just replace 90% of the SVG stuff with that then :S
3467 18:09 < sulix> It only works on nVidia hardware, though, so it's not really practical.
3468 18:10 < matches> I think I need to actually do the Bezier's bounding rectangles correctly
3469 18:10 < sulix> I was just about to add that, actually.
3470 18:10 < matches> But its not just (x0,y0,x3-x0,y3-y0)
3472 18:10 < matches> Go ahead then
3473 18:10 < sulix> Nah, you have to take the x and y components separately and solve for min and max.
3474 18:11 < matches> Yes google is a lot faster than actually doing maths
3475 18:13 < matches> I can add that unless you've already written it?
3476 18:13 < sulix> Sure.
3477 18:13 < sulix> The internet seems to claim that you should use newton-raphson to find the roots of the derivatives, but I'm not really comfortable with doing that for cubic béziers,
3478 18:31 < matches> I just differentiated the parametric formula
3479 18:31 < matches> And got a quadratic
3480 18:31 < matches> Unfortunately there is a slight problem with Real
3481 18:31 < matches> Since we don't have a general "sqrt"
3482 18:31 * matches adds that to the growing list of "fix later"
3483 18:32 < matches> It doesn't compile for Rational<Gmpint> anymore anyway
3484 18:38 < matches> I am writing code that sjy would probably not consider elegant
3485 18:50 < matches> Well that totally broke everything
3486 18:53 < sulix> So it turns out that the view reparenting was only working by chance.
3488 19:20 < matches> Hmm it turns out finding the bounding box of a bezier has really annoying edge cases
3489 19:43 < sulix> Okay, automatic generation of new quadtree nodes "works" when zooming in.
3490 19:44 < sulix> It doesn't do any fancy clipping, so the actual rendering code hits precision issues, but it's still pretty cool.
3491 19:48 < sulix> It also get a bit buggy if you try to pan over the edges of quadtrees, as it only renders one node at a time.
3492 19:59 < matches> Nice
3493 21:14 < sulix> I have just achieved infinite precision with the quadtree!
3494 21:14 < sulix> Only on rectangles, only when zooming in, and only when the camera doesn't cross a quadtree boundary, but it works!
3495 --- Day changed Thu Aug 14 2014
3496 09:50 < matches> cool
3497 10:37 < matches> I'm beginning to suspect whoever said to use the Newton Raphson method to find Bezier bounding boxes was right
3498 10:43 < matches> I love how the sites I looked up said "Obviously there is a problem if a = 0 or b^2 - 4ac < 0, but we can just make sure we always pick beziers that don't cause those problems
3500 10:44 < matches> I have at least got an algorithm that returns finite sized bounding boxes now
3501 10:45 < matches> I think it even works!
3502 10:45 < matches> But only on the CPU
3503 10:45 < matches> Presumably the GPU does something to cope with the fact that the bounding boxes were totally wrong before?
3504 10:53 < matches> Oh it wasn't doing anything because they were {0,0,1,1}
3505 10:54 < matches> I have to transform the Bezier coefficients before they are uploaded to the GPU
3506 10:54 < matches> Blergh
3507 10:55 < matches> What I've basically done is change it to be the opposite of how it was designed
3508 10:55 < matches> -_-
3509 10:56 < matches> So before, the Bezier control points were relative to some bounding rectangle but when I parsed the SVG I just made it always {0,0,1,1} so the coordinates were absolute
3510 10:56 < matches> Now I have changed the Bezier control points to be absolute and then calculated a bounding rectangle from them
3511 10:56 < matches> I can maths
3512 10:57 < matches> So I could make the Beziers still have the {0,0,1,1} bounds which will fix the GPU renderer without having to transform the coefficients
3513 10:57 < matches> And just leave this "SolveBounds" function in for actually getting the bounds
3514 10:58 < matches> Except SolveBounds is disgusting
3515 10:58 < matches> And slow
3516 10:58 < matches> So I could add a Bounds member variable to the Beziers
3517 10:58 < matches> And then we have two bounds
3518 10:58 < matches> I seem to have coded myself into a catch 22 "No matter what you do it is terrible" situation
3519 10:59 < matches> Adding a Bounds member variable to the Bezier struct sort of defeats the whole point of the "Object of Arrays" idea
3520 10:59 < matches> You know what
3521 11:00 < matches> I think the "Object of Arrays" approach has caused more problems than it solved :P
3522 11:02 < matches> (Probably not actually)
3523 11:09 < matches> Spaghetti
3524 11:09 < matches> I'm going to transform the coordinates before uploading to the GPU shaders
3525 11:09 < matches> Because I think that is the least objectionable?
3526 11:09 < matches> I don't know
3527 11:09 < matches> I still don't like any of the solutions
3528 11:10 < matches> But having more than one bounding rectangle definitely seems dumb
3529 11:10 < matches> Especially if the other is *always* {0,0,1,1}
3530 11:11 < matches> I hope this doesn't totally break your quad tree
3531 11:15 < matches> Actually looking at the quad tree it seems that is the best way to make it more likely to work
3532 11:21 < matches> I feel like we should use SVGMatrix more and Rect less
3533 11:21 < matches> But oh well
3534 11:29 < sulix> Ah: so the Quadtree kinda relies on the Bézier coordinates being relative to the bounding rectangle.
3535 11:29 < matches> Yep
3536 11:29 < matches> I'll do that
3537 11:29 < matches> It's easier to do the transform when the coordinates are uploaded to the GPU
3538 11:29 < matches> Although probably less efficient
3539 11:30 < matches> On the other hand, the CPU renderer relies on having absolute coordinates
3540 11:30 < matches> Two renderers
3541 11:30 < matches> TWICE the matrices
3542 11:30 < matches> TWICE the confusion
3543 11:31 < matches> And Rects are not really convenient because they aren't a proper transformation matrix
3544 11:31 < matches> I can't just multiply some Rects together
3545 11:31 < matches> Or calculate an inverse
3546 11:32 < sulix> Yeah, there are problems there.
3547 11:34 < matches> Hmm
3548 11:34 < matches> I *almost* fixed it
3549 11:34 < sulix> Rects are really good for the QuadTree, though.
3550 11:34 < matches> We should leave it as rects
3551 11:34 < matches> You just manually work out what it needs to be anyway
3552 11:35 < matches> I think the inverse of the equivelant matrix will always work as a rect
3553 11:35 < matches> So there are random lines missing from the beziers now
3554 11:35 < matches> But the ones that are there are in the right spot!
3555 11:36 < matches> (On the GPU)
3556 11:36 < sulix> So, I looked at that. I see random missing lines on Intel, not on nVidia.
3557 11:37 < matches> Oh were they missing before?
3558 11:37 < matches> Well I don't recall them being missing before
3560 11:37 < matches> I know what it is
3561 11:38 < matches> When the bounding rectangle has width or height of zero
3562 11:38 < matches> Sigh
3563 11:38 < matches> Wait that can't be it
3564 11:38 < matches> Most of the missing lines aren't straight horizontal or vertical
3566 11:39 < sulix> I think it's just dodgy rounding on the GPU.
3567 11:40 < matches> I will push what I have I guess
3568 11:40 < matches> I have nice pretty debug rectangles (in *colour*) around the beziers when they are rendered on the CPU
3569 11:40 < matches> Somehow I don't think that is going to be sufficiently impressive progress in the meeting
3570 11:41 < sulix> Oooh... I'm looking forward to trying that.
3571 11:41 < sulix> I have an impressive Quadtree demo to do!
3572 11:41 < sulix> Assuming the bezier stuff doesn't suddenly break.
3573 11:41 < matches> ... it might
3574 11:41 < matches> Ok, I will `git stash` and then `git pull` as opposed to merging
3575 11:42 < matches> Yes, there are definitely not randomly missing beziers in your code
3576 11:44 < matches> Wait, was your quad tree working with beziers before?
3577 11:44 < matches> It wouldn't have been?
3578 11:44 < matches> All the bounds were wrong
3579 11:48 < matches> Ooh
3580 11:48 < matches> H and V
3581 11:48 < matches> What are they
3582 11:49 < sulix> Horizontal and Vertical lines
3584 11:50 < sulix> The QuadTree works with beziers, but doesn't increase precision.
3585 11:52 < matches> Hmm, now to work out which of the 5 or 6 edge cases is breaking the Bezier bounds algorithm
3586 11:53 < matches> It's kind of hard to debug
3587 11:53 < matches> The edge cases apply seperately to the y and x directions
3588 11:53 < matches> So you can't tell just by looking at a curve that it is an edge case
3589 11:54 < matches> Unless you are a mathemagician I guess
3590 11:55 < matches> I don't understand
3591 11:55 < matches> The bounds for the beziers that disappear look right
3592 11:59 < matches> Maybe I'm not uploading the data correctly anymore
3593 12:03 < matches> Ok now it only breaks for the horizontal and vertical lines which I at least understand
3594 12:03 < matches> I was iterating over the object bounds and uploading the ones that had a type of BEZIER
3597 12:04 < matches> Yeah because I added a GROUP as well
3598 12:04 < matches> And every <path> gets a GROUP now
3599 12:04 < matches> So the order of indexes must be off
3600 12:14 < matches> Face palm
3601 12:14 < matches> This is why we have objects.data_indices
3602 12:19 < matches> I wish I'd added the SVG stuff earlier
3603 12:19 < matches> It makes it a lot easier to do stuff when you can mess around with actual documents
3604 12:42 < matches> I have pushed some stuff
3605 12:43 < matches> Mostly it just looks like I added a bunch of colourful rectangles to the CPU rendering
3606 12:43 < matches> I think I broke the Quad tree too
3607 12:43 < matches> Go team
3608 12:43 < matches> Oh and there are a bunch of not so colourful GPU rectangles
3609 12:43 < matches> Due to the "GROUP" object just using the outline rectangle shaders
3610 12:44 < matches> So part of the quad tree seems to work on the GPU but not all of it
3611 12:44 < matches> And none of it seems to work on the CPU?
3612 12:48 < matches> Also instead of rendering a slightly broken fox it now renders a segfault
3613 12:48 < matches> Similarly for the koch snowflake
3614 12:48 < matches> It's alright we can still draw Humphrey
3616 12:53 < matches> It's just the quad tree that segfaults on the more complicated svgs
3617 13:38 < sulix> I have returned from the committee meeting, and that stupid 2/3rds majority thing has been revoked!
3618 13:38 < matches> Hooray!
3619 13:39 < matches> I should head towards University for the project meeting
3620 13:39 < matches> It is tempting to just stay home
3621 13:40 < matches> But that would probably not be wise
3622 13:41 < sulix> You should come.
3623 13:41 < sulix> I'm going to see if I can fix the Quadtree, but it looks like your changes have thoroughly broken it.
3624 13:41 < matches> Sorry!
3625 13:41 < sulix> By which I mean, it probably was only working if everything had bounds (0,0)-(1,1)
3626 13:41 < matches> Haha
3627 13:42 < matches> You can just undo my changes to demonstrate it if you want
3628 13:42 < matches> To the bus
3629 13:43 < sulix> I may yet do that...
3630 14:54 * sulix -> CSSE
3631 19:08 < matches> So shading on the CPU sort of totally breaks
3632 19:08 < matches> But it exists
3633 19:08 < matches> Progress!
3634 19:10 < matches> Bezier bounds are not as good as I'd hoped
3635 19:10 < matches> I think they only work if the end points form the bounds
3636 19:17 < sulix> Ah...
3637 19:21 < sulix> Btw, in the default "c" glyph, there are _two_ curves with bounds (0,0)-(1,1)
3638 19:21 < sulix> Numbers 9 and 18
3639 19:22 < matches> Yeah I found the wierd ebounds
3640 19:22 < matches> It's in the font code
3641 19:22 < matches> There is an AddBezierData with (0,0,1,1) instead of AddBezier (which automatically works out the bounds)
3642 19:22 < matches> For vlines
3643 19:22 < sulix> That'd do it.
3644 19:23 < matches> shape.svg seems to lose part of the curve now
3645 19:23 < matches> I've made the bezier control points relative
3646 19:23 < matches> Things mostly sort of work
3647 19:23 < matches> But the quad tree still looks fantastically broken
3648 19:25 < sulix> It always looks broken if you pan around or zoom out: it's my ugly baby.
3649 19:28 < matches> http://szmoore.net/ipdf/brokenbeziers.png
3650 19:28 < matches> To be fair the beziers are actually fine
3651 19:28 < matches> It's just the bounding rectangles that are broken
3652 19:28 < matches> Also shading should probably not be turned on unless it is actually a closed path
3653 19:29 < sulix> Hmm...
3654 19:29 < matches> Wait
3655 19:29 < matches> The beziers are rendering fine
3656 19:29 < matches> But the bounding rectangles are wrong
3657 19:29 < matches> Does this mean
3658 19:29 < matches> If I *fix* the bounding rectangles
3659 19:29 < matches> It will break something
3660 19:29 < matches> :-(
3661 19:29 * matches forges ahead nevertheless
3662 19:30 < matches> It does look suspiciously like it is only ever using P0 and P3 for the bounds
3663 19:30 < matches> Despite all that horrible solving for the turning points
3664 19:31 < sulix> Yeah.
3666 19:32 < matches> I suspect if I fix this it will totally break the absolute to relative bezier transform then
3667 19:35 < matches> The file for those beziers is kind of interesting
3668 19:35 < matches> inkscape has used absolute commands for all except one of the beziers
3669 19:35 < matches> Why?
3670 19:35 < matches> The first one was copied and pasted with the control points moved
3671 19:35 < matches> Also for some reason random "translate" applied to the whole group
3672 19:36 < matches> It might make some sense if it caused the coordinates of the actual paths to come out as integers maybe
3673 19:36 < matches> But they don't
3674 19:37 < matches> There is translate and then the initial "moveto" is not 0,0
3675 19:37 < matches> I guess we shouldn't be trusting an svg editor that segfaults if you try recursion anyway
3676 19:59 < matches> Ok, high school maths:1, 5th year science/engineering student:0
3677 20:15 < matches> Also my amazingly clever shading algorithm has
3678 20:15 < matches> ...
3679 20:15 < matches> issues
3680 20:15 < matches> When the paths are not in the view
3681 20:16 < sulix> Ah...
3682 20:18 < matches> In fact it has many issues in many cases
3683 20:27 < matches> Hmm
3684 20:27 < matches> I wonder if this is what causes problems with evince/atril
3685 20:27 < matches> Ok not this
3686 20:28 < matches> I mean, if they use a similarly terrible except actually working shading algorithm
3687 20:28 < matches> And just draw everything to a really big texture so they can use it
3688 20:28 < matches> As opposed to clipping things
3689 20:28 < matches> That seems a little unbelievable though
3690 20:29 < matches> Surely that would crash long before you even got to >1600%
3691 20:30 < matches> I suppose what I should really do is look up some papers on shading
3692 20:30 < matches> Blergh
3693 20:35 < matches> Things are pushed
3694 20:37 < matches> Ah crap
3695 20:38 < matches> Things will be pushed in 8 hours according to my clock
3696 20:38 < matches> Just don't do anything until after 4am tomorrow and everything will be in the right order
3698 20:39 < sulix> That shading is _amazing_
3699 20:40 < matches> Try replacing the colour with rand()%255 for even more amazingness
3700 20:40 < sulix> If you zoom in _just_ the right amount, all hell breaks loose. I love it!
3701 20:40 < matches> Haha
3702 20:40 < sulix> Slowly panning across it is also magical!
3703 20:41 < matches> Try it on rabbit_simple.svg
3704 20:42 < sulix> I think Humphrey might have come down with a case of stripey binary myxomatosis.
3705 20:43 < matches> So I will have to think about shading a bit more :P
3706 20:43 * sulix braves the unholy combination of shading and quadtree.
3707 20:43 < matches> Oh I didn't try that
3708 20:44 < sulix> It actually doesn't look any different.
3709 20:44 < sulix> I think the quadtree only breaks w/ the GPU.
3710 20:44 < matches> Yes
3711 20:45 < matches> You lose the debug message if the node # > 9 it seems too
3712 20:46 < matches> At least the quad tree works on the CPU
3713 20:46 < matches> But why doesn't it work on the GPU they are now using the same bounds
3714 20:47 < matches> So should I prepare a report for Tim?
3715 20:47 < matches> Progress Report: Arbitrary precision numbers are unsurprisingly, totally infeasible
3716 20:47 < matches> Therefore, we implemented random bits of SVG
3717 20:47 < matches> Figure 1: A totally broken shading algorithm
3718 20:48 < sulix> You should include this image of the quadtree: http://davidgow.net/stuff/ipdf-humphreys-ghost-face.png
3719 20:48 < matches> Yes
3720 20:48 < matches> I wonder if since we're allegedly engineers we should document what bugs go with what images
3721 20:48 < matches> Hmmm
3722 20:49 < sulix> That's why I name the screenshots like that, I know this goes with the "Humphrey's Ghost Face" bug.
3723 20:49 < matches> Haha
3724 20:49 < matches> So we have the svg-test images
3725 20:50 < matches> Would it be worth while to have a commit script that automatically drew them and did screnshots
3726 20:50 < matches> It might be tricky though because you have to test panning and zooming and things
3727 20:50 < sulix> Yeah... perhaps videos?
3728 20:51 < matches> We'd need to be able to control the view automatically
3729 20:51 < matches> Which is actually required for that list of things in the proposal we are supposed to do anyway
3730 20:52 < sulix> I think we'll want that feature at some point anyway for doing proper performance tests
3731 20:54 < matches> I kind of want some sort of debug overlay for things like the bounding boxes too
3732 20:54 < sulix> Hmmm... the bounds look right: http://davidgow.net/stuff/ipdf-humphrey-bounds-ahead.png
3733 20:54 < matches> And a menu system
3734 20:54 < matches> And the ability to "insert SVG here"
3735 20:54 < matches> And open a file browser
3736 20:54 < matches> And...
3737 20:54 < matches> And...
3738 20:54 < matches> Yeah
3739 20:54 < sulix> I've been seriously considering doing a menu system.
3740 20:54 < matches> Too bad christmas is after the thesis is due
3741 20:54 < matches> Do it!
3742 20:55 < matches> We're running out of mouse buttons to toggle things :P
3743 20:55 < matches> Are there any SDL based menu systems that we can use easily?
3744 20:55 < sulix> There are a couple of OpenGL-y ones that might work.
3745 20:56 < matches> Let's not port the project from SDL -> FreeGlut
3747 20:56 < sulix> Let's not.
3748 20:57 < sulix> (I'm looking for an SDL/OpenGL menu system so that I can replace FreeGLUT in the Graphics project(
3749 20:57 < matches> Could we use Qt or is that overkill?
3750 20:57 < sulix> I suspect it's probably overkill.
3751 20:59 < sulix> I've got a couple of old GUI libraries for SDL/OpenGL I wrote in high school I could dig up as a last resort.
3752 20:59 < matches> ... I have seen your high school code
3754 20:59 < sulix> (I think one of them uses a tiny bit of boost, though, so I'd rather avoid it)
3755 21:03 < matches> I kind of want to look at Qt at some point
3756 21:03 < matches> Maybe if we have the menu in a seperate window...
3757 21:03 < matches> Then we need threads...
3758 21:03 < matches> Then we have two problems...
3759 21:09 < matches> All the links to SDL libraries are broken
3760 21:34 < sulix> I'm not sure what changed, but now ipdf crashes my OpenGL debugger.
3761 21:42 < sulix> The QuadTree/GPU code works again!
3762 21:42 < sulix> Somehow the bug I fixed last night reappeared, but I've added the needed *2s back in.
3763 --- Day changed Fri Aug 15 2014
3764 08:43 < matches> So it seems to work well but things still go wierd on the GPU
3765 08:44 < matches> I'm in Quadtree node 32 looking at the tip of Humphrey's eyebrow
3766 08:44 < matches> The bottom line moves normally
3767 08:44 < matches> Vertically only sorry
3768 08:44 < matches> It doesn't move horizontally
3769 08:44 < matches> The top line stays still and every so often jumps to a new (wrong) position
3770 08:45 < matches> On the CPU it appears fine though
3771 08:45 < matches> Although on the CPU it doesn't show the "Current View QuadTree Node: 32"
3772 08:46 < matches> Hrm, I should add a 2d view of a car to the svg-tests
3773 08:47 < matches> Then I can be like "IT IS RELEVANT TO MECHANICAL ENGINEERING BECAUSE BEZIERS"
3774 08:47 < matches> In the conference/thesis offense
3775 08:48 < matches> My concern is actually more how I'm going to present the arbitrary precision number representations because they don't even work
3776 08:49 < matches> Well I should probably try Gmp rationals and Gmp floats before totally giving up I guess
3777 08:49 < matches> And also those p-adic things
3778 08:50 < matches> I never looked them up
3779 --- Day changed Sun Aug 17 2014
3780 14:38 < matches> Hey Qt has a lot of stuff in it
3781 14:38 < matches> They have their own iostreams that seem identical to std::iostream
3782 14:39 < matches> It's kind of cool I guess except there are "Qs" everywhere
3783 14:41 < matches> Ok we need a window icon
3784 14:41 < matches> This is vital
3785 15:54 < sulix> I've got SDL2 window icon setting code if you want it: https://github.com/sulix/omnispeak/blob/master/src/icon.c#L555
3786 16:04 < matches> Cool
3787 16:39 < matches> I'm bringing in spaghetti
3788 16:39 < matches> I mean Qt
3789 16:40 < matches> In theory this means the Debug functions should have a mutex if we want to debug things in the Qt stuff
3790 16:40 < matches> But in practice screw it
3791 16:40 < matches> What could go wrong?
3792 16:41 < matches> (Don't worry, I have a CONTROLPANEL_DISABLED define)
3793 16:42 < matches> This View/Document/Screen thing could probably be tidied up a lot
3794 16:42 < matches> But effort
3795 16:42 < matches> I think I'll just give ControlPanel direct access to document, screen and view
3796 16:42 < matches> Currently everything has to go through view
3797 16:43 < matches> Which would make more sense if some of screen was in view maybe
3798 16:43 < matches> Bah screw design
3799 16:51 < matches> So now we have 4 main classes!
3800 16:51 < matches> 4 x the fun!
3801 17:29 < matches> Bargh Qt needs its own version of make
3802 17:29 < matches> That's stupid
3804 21:10 < matches> I have a mysterious LMS deadline that suddenly appeared
3805 21:10 < matches> "Progres Report"
3806 21:10 < matches> Description
3807 21:10 < matches> "asadf"
3808 21:10 < matches> Due
3809 21:10 < matches> 19th August, 11:01PM
3810 21:10 < matches> What
3811 21:14 < sulix> Oh dear.
3812 21:15 < matches> I have sent an email
3813 21:15 < matches> I hope that's not the conference abstract
3814 21:15 < matches> I also really hope we're allowed to use our own computer and things in the conference
3815 21:16 < matches> Also I really hope that if I can use my own computer it doesn't decide to shit itself anyway just to spite me
3816 --- Day changed Mon Aug 18 2014
3817 17:06 < matches> Good news and bad news!
3818 17:07 < matches> The kerning now appears to kern in the right direction
3819 17:07 < matches> But this means "BleedingCowboys.ttf" works
3820 20:30 < sulix> It's a pain to look up clipping of beziers, because there's a separate technique for doing something else called "bézier clipping"
3821 20:40 < matches> Baha
3822 20:41 < matches> I'm basically just procrastinating with Qt
3823 20:41 < matches> I mean
3824 20:41 < matches> Doing *awesome* things with Qt
3825 20:48 < sulix> I am looking with considerable envy at SolveQuadratic().
3826 20:49 < sulix> SolveCubic() will get very ugly.
3827 20:49 < sulix> Assuming I don't just sample the cubic at 100 points and look for roots.
3828 20:49 < matches> Haha
3829 20:49 < matches> Newton Raphson?
3830 20:50 < sulix> Very tempting.
3831 20:50 < matches> Then you can solve any bezier!
3832 20:50 < matches> Then we can put in Quintics!
3833 20:50 < sulix> Solving cubics exactly is not fun.
3834 20:50 < sulix> Dear god no.
3835 20:50 < matches> Imagine how many loops a quintic could have...
3836 20:51 * sulix cries quietly in the corner.
3837 20:53 < matches> So I'm calling AddText with sane values and nothing is happening :(
3838 20:56 < sulix> With or without quadtree?
3839 20:56 < sulix> With quadtree, everything is broken if you try adding things.
3840 20:56 < sulix> Because they won't be added to a node.
3841 20:57 < matches> No quad trees I think
3842 20:57 < matches> Yeah no quad tree
3843 21:00 < matches> Wait I think we need to rebuild the buffers or something
3844 21:07 < matches> Ah, ForceRenderDirty()
3845 21:14 < matches> I don't think the scale of our text is quite right?
3846 21:14 < sulix> Well, the scale of everything is wrong, but it's entirely possible that the text is _extra wrong_
3847 21:16 < sulix> In theory, the "scale" input to addText is the "height" of a character.
3848 21:19 < matches> Mmm
3849 21:19 < matches> I'm trying to add text relative to the current view
3850 21:19 < matches> But it isn't consistent
3851 21:19 < matches> When I go to a different view it is definitely not adding it in the same relative place
3852 21:20 < matches> The scale and positions aren't *already* relative are they?
3853 21:22 < sulix> Not to the view...
3854 21:24 < matches> What are they relative to?
3855 21:24 < matches> Anyway instead of fixing that bug I'm adding more
3856 21:25 < sulix> Umm... The bounding box maybe?
3857 21:25 * sulix investigates.
3858 21:26 < sulix> Scale and position should (I think) be normal, document coordinates.
3859 21:26 < matches> Odd
3860 21:26 < matches> Oh well
3861 21:26 < matches> Having Qt is totally worth it by the way
3862 21:27 < matches> Although it is still slightly embarrassing that it has its own SVG parser
3863 21:27 < matches> Ours is better!
3864 21:27 < matches> It has Reals everywhere!
3865 21:27 < matches> Except where it constructs Real from floats...
3866 21:27 < matches> And converts them to floats...
3867 21:28 < sulix> Thesis title should be "Keepin' it real: Staying afloat on the sea of document precision"
3868 21:31 < matches> Bahaha
3869 21:31 < matches> That is brilliant
3870 21:32 < matches> The Qt stuff is never deleting anything
3871 21:32 < matches> I suppose I should fix that at some point
3872 21:33 < matches> So basically it is a state machine and you use the menu to change the state
3873 21:33 < matches> Pressing OK does something depending on the state
3874 21:34 < matches> And there is a single text edit you can type stuff into
3875 21:34 < matches> In theory it can hide/show different widgets depending on the state
3876 21:34 < matches> I feel like I should add a "Generate Thesis Title" widget now...
3877 21:35 < matches> Also I can sort of see the appeal of lambdas for GUI programming now
3878 21:35 < matches> But I have a feeling they would totally break Qt anyway
3879 21:49 < matches> ...
3880 21:49 < matches> Ok sometimes the text gets added upside down...
3881 21:49 * sulix disclaims any and all responsibility for that.
3882 21:51 < matches> Yes!
3883 21:51 < matches> I finally did that thing I was saying I'd do for like 3 weeks
3884 21:51 < matches> Zoom in to floating point limit
3885 21:51 < matches> Add SVG
3886 21:51 < matches> Watch it explode!
3887 21:52 < sulix> Oooh...
3888 21:52 < matches> It is glorious
3889 21:54 < matches> We need to be a bit careful with thread safety in qt
3890 21:55 < matches> Really we should be able to request that the qt thread call UpdateAll itself
3891 21:55 < matches> Rather than calling it directly from the viewer thread
3892 21:55 < matches> But I'm not sure how that's done
3893 21:55 < matches> There's all this stuff about "Cross thread signals" on the internet
3894 21:56 < matches> It seems you can hide/show things but not much else
3895 22:47 < matches> Bah you pushed things!
3896 22:55 * sulix proclaims the new features as excellent and promptly goes to bed.
3897 --- Day changed Tue Aug 19 2014
3898 18:08 < matches> My conference is in the week starting 13th October
3899 18:08 < matches> Is Tim back then?
3900 18:08 < matches> "your supervisor will be invited to attend and mark your presentation"
3901 18:09 < matches> So if he can't go... I don't get a mark?
3902 18:12 < matches> It sounds suspiciously like they are going to get everyone's supervisors to mark them and then just scale everyone :S
3903 18:13 < matches> Surely at least one other person is going to read the final report...
3904 18:13 < matches> I've heard bad things about the engineering scaling system
3905 18:14 < matches> Your mark becomes inversely proportional to how well your supervisor's students did the previous year or something like that
3906 18:15 < matches> Anyway, two months left to accomplish something!
3907 20:05 < sulix> Worked out why the font bounds were weird.
3908 20:06 < sulix> AddText takes (msg, scale, x, y) not (msg, x, y, scale), so the coordinates were all bunk.
3909 20:09 < sulix> Fonts looks _amazing_ with floating pt errors, too.
3910 20:09 < sulix> http://davidgow.net/stuff/ipdf-i-think-you-mean-comic-fail.png
3911 20:29 < sulix> The Art of Computer Programming briefly mentions the Stern-Brocot tree, btw.
3912 20:29 < sulix> I can't see it being really useful, though.
3913 21:06 < sulix> My difficulty clipping bézier paths may be because it is actually impossible.
3914 21:06 < sulix> I think I've just found a paper saying that it is only possible for béziers which happen to be straight lines.
3915 22:55 < sulix> Never mind, it is possible, it's just very nasty.
3916 23:15 < sulix> Also C++14, because we need more lambdas: https://isocpp.org/blog/2014/08/we-have-cpp14
3917 --- Day changed Wed Aug 20 2014
3918 12:05 < matches> I guess we've accomplished the goals for this week's meeting then!
3919 12:05 < matches> (That shading algorithm totally counts as a "first attempt")
3920 12:07 < sulix> I'm about 2/3rds of the way through deriving the reparametrisation of béziers.
3921 12:08 < sulix> It basically just boils down to pages and pages of rearranging polynomials.
3922 12:38 < sulix> I managed to get the qt4/qt5 monstrosity compiling on my laptop.
3923 12:38 < sulix> It has a "moc-qt4" program, which works, whereas "moc" is qt5.
3924 16:28 < matches> I think I need to stop adding this log to git...
3925 23:08 < sulix> matches: But then, how would history record that I've finally got a maybe-correct derivation for cubic bézier reparametrisation: http://davidgow.net/stuff/cubic_bezier_reparam.pdf
3926 23:08 < sulix> (Code is in git)
3927 --- Day changed Thu Aug 21 2014
3928 12:57 < sulix> So it turns out that quadtrees are hard, and reparametrising béziers is why.
3929 13:03 < matches> I don't seem to see anything anymore with the quad tree enabled
3930 13:10 < sulix> The quadtree doesn't let you add things in realtime.
3931 13:10 < sulix> Quadtree nodes are immutable once created.
3933 13:35 < sulix> debug: ReParametrise (bezier.h:190) - (0.000000,0.000000),(1.000000,1.000000),(1.000000,1.000000),(1.000000,1.000000) -> (-nan,-nan),(-nan,-nan),(-nan,-nan),(-nan,-nan)
3934 13:36 < sulix> Looking good!
3935 13:54 < sulix> Clearly, the quadtree rendering is now perfect: http://davidgow.net/stuff/ipdf-this-is-definitely-a-fox.png
3936 --- Day changed Sun Aug 24 2014
3937 17:51 < matches> ...
3938 17:51 < matches> I am parsing CSS in the SVGs
3939 17:51 < matches> WHAT HAVE I DONE
3940 17:52 < matches> Blame inkscape for always using css to control the properties
3941 17:57 < matches> http://szmoore.net/ipdf/death-by-shading.png
3942 17:59 < matches> Anyway there is a proper "Group" data type like with the Beziers
3943 18:00 < matches> So if you want to cache things related to groups you can put them in that I guess
3944 18:02 < matches> Really it should be called "PATH" not "GROUP" but I vaguely thought it might be able to be used as either
3945 --- Day changed Mon Aug 25 2014
3946 10:39 < matches> I was thinking about how to do shading more
3947 10:40 < matches> And I basically arrived at the idea Loop and Blinn have that you kept talking about
3948 10:40 < matches> Except I didn't realise it at first because my version was all using rectangles instead of triangles
3949 10:40 < matches> I think I will try to get my scanline fill to work first though
3950 13:03 < matches> It sort of works
3951 13:15 < matches> ... or not
3952 --- Day changed Tue Aug 26 2014
3953 14:25 < matches> I wonder if I'm going to end up just reverse engineering an algorithm that I was too lazy to understand properly
3954 14:26 < matches> I really like the idea of flood fill better than doing insane triangulation maths
3955 14:26 < matches> Surely GPUs can flood fill
3956 14:28 < sulix> I think they can, but either very slowly or only on very new hardware...
3957 14:28 < sulix> I think you'd need to do it in several passes, maybe?
3958 14:28 < matches> The problem I had with flood fill was working out where to start
3959 14:28 < matches> I have something that almost works
3960 14:28 < matches> That requires no clever maths...
3961 14:29 < matches> But it is CPU only
3962 14:29 < matches> And probably breaks in a bunch of special cases
3963 14:30 < matches> It doesn't have precision problems though (once you've drawn the boundary that is)
3964 14:31 < matches> It would still be cool to go through the Loop/Blinn algorithm on CPU and GPU
3965 14:31 < matches> Because that is going to have floating point maths in it
3966 14:31 < matches> flood fill has no floats
3967 14:31 < matches> Therefore it is infinite precision
3968 14:31 < matches> Objective complete
3969 14:32 < matches> Actually I don't think my algorithm is proper flood fill, it's sort of scanline flood fill but not?
3970 14:32 < matches> When in doubt write random code
3971 14:34 < sulix> How are you handling the case where, due to a path being half offscreen it becomes two paths?
3972 14:34 < matches> ... I am not
3973 14:35 < matches> If part of the path is offscreen it is annoying
3974 14:36 < sulix> Hopefully, once some of the quadtree code is more functional, the path clipping code can be used to clip paths to the screen.
3975 14:39 < matches> Compositing is also totally broken because the fill just looks for black pixels
3976 14:40 < matches> But if I can get it doing fonts then that's sort of not totally useless
3977 18:35 < matches> I am very tempted to implement a "Please click on the location to fill this path from" interface at the moment
3978 18:39 < matches> Alright, I have shading somewhat stack overflowing
3979 18:39 < matches> I mean
3980 18:39 < matches> Working
3982 18:39 < matches> ********
3983 18:49 < matches> http://szmoore.net/ipdf/shading-the-only-svg-that-doesnt-segfault.png
3984 20:43 < matches> Argh
3985 20:43 * matches stabs coordinate transforms
3986 20:43 < matches> stab stab stab
3987 20:44 < matches> My filling algorithm is to find the 4 extrema of the path
3988 20:44 < matches> And try and flood fill inwards from each one
3989 20:45 < matches> But the extrema pixel coordinates keep transforming to *outside* of the path
3990 20:45 < matches> By one pixel
3991 20:45 * matches stabs
3992 21:52 < matches> I guess I should just actually implement loop and blinn shading
3993 21:52 < matches> Flood fill is not as easy as I had hoped
3994 21:53 < matches> I have to start doing annoying geometry maths to work out what point I can start the flood fill at
3995 21:55 < matches> Sigh
3996 --- Day changed Wed Aug 27 2014
3997 22:33 < sulix> Suddenly: ToAbsolute on ToRelative does not give original Bezier
3998 22:34 < sulix> I may need to rethink what my quadtree code is doing to those poor curves.
3999 22:51 < sulix> Never mind, that assertion happens _whenever_ trying to load cubicbeziers.svg.
4000 23:03 < sulix> Oh wow: you should see what happens if you start splitting and reparametrising beziers with bad precision.
4001 23:03 < sulix> 'tis "whacked"
4002 23:08 < sulix> http://davidgow.net/stuff/ipdf-a-slight-curve.png
4003 23:15 < sulix> And the Quadtree infinite precision for béziers now works*
4004 23:15 < sulix> *for some beziers.
4005 23:15 < sulix> * for some large finite value of "infinite"
4006 --- Day changed Thu Aug 28 2014
4007 11:35 < sulix> What time is it? It's NaN time!
4008 11:44 < sulix> Ah, so... horizontal lines have a height of zero in their bounds. This makes divides by zero happen in places.
4009 11:48 * sulix apologieses profusely for his latest "break absolutely everything" commit.
4010 11:57 < sulix> Yeah, making divide by zero crash rather than generating NaNs breaks a lot of things.
4011 11:57 < sulix> You'd be surprised by how often we divide by zero.
4012 11:57 < sulix> Mostly in the Quadtree code.
4013 11:58 < sulix> Because, damnit, mathematics can't beat me.
4014 12:20 < matches> Yes, lines have zero width... stupid lines
4015 12:21 < matches> I stopped the segfaults in shading by adding a depth counter to FloodFillOnCPU
4016 12:21 < matches> Which makes interesting shading patterns...
4017 12:22 < matches> I think it's time to fall back to the "Use the second algorithm on Wikipedia" approach
4018 12:23 < matches> I can still use a Flood Fill but hopefully a less segfaulty one
4019 12:29 < matches> I'm not sure about ToAbsolute and ToRelative...
4020 12:29 < matches> I thought they should be inverses but they seem not to be
4021 12:29 < matches> Oh wait, the divide by zero would do that
4022 12:30 < matches> That curve looks cool
4023 12:30 < matches> You should talk about it lots in the meeting
4024 12:31 < matches> For the whole meeting
4025 12:31 < matches> Nothing else
4026 12:31 < matches> Don't mention the total lack of progress I made...
4027 12:32 < matches> How much can I trust SolveCubic?
4028 12:40 < matches> Basically I'm going with randomly picking points until I find one inside the shape then starting a flood fill from that point
4029 12:41 < matches> Checking the pixels for the edges is not feasable
4030 12:42 < matches> If you scan along a bresenham line you could jump over the edge
4031 12:42 < matches> ... I see an "if false" in SolveCubic
4032 12:42 < matches> Slightly worrying
4034 12:43 < matches> I approve of SolveCubic
4036 13:31 < matches> Did you introduce the floating point exception or did I...
4037 13:45 < sulix> I just did.
4038 13:49 < sulix> SolveCubic may have small problems.
4039 13:50 < matches> Yes I am trying to rewrite it now...
4040 13:50 < matches> I swear we did this in CQM
4041 13:52 < sulix> Solvecubic also sometimes tries to take the square root of a negative number.
4042 13:52 < matches> Yep
4043 13:54 < sulix> I've pushed slightly less broken code.
4044 13:55 < sulix> Ny which I mean slightly more broken.
4045 13:55 < matches> ...
4046 13:55 < sulix> I'm avoiding divide by zeros by "if (denominator = 0) denominator = 1" style shenanigans.
4047 13:56 < matches> I'm slightly scared by the change to Rect::TransformRectCoordinates
4048 13:56 < matches> Oh right
4049 13:56 < matches> Yeah
4050 13:56 < matches> That seems...
4051 13:56 < matches> Scary
4052 13:56 < sulix> It works, scarily enough.
4053 13:56 < matches> Anyway I already ifdef'd the thing out of SolveCubic
4054 13:56 < matches> It's too slow for me though
4055 13:57 < sulix> And by works, I mean I haven't found any issues yet.
4056 13:57 < matches> I am brute forcing a point inside the shape
4057 13:57 < matches> This requires finding the intersections with horizontal and vertical lines for every single bounding bezier
4058 13:57 < matches> Then the "even or odd" test
4059 13:57 < matches> You only have to do it once
4060 14:05 < matches> I think SolveCubic would best be done by doing Newton Raphson or the Shooting Method between the endpoints and the turning points
4061 14:05 < matches> I'll see
4062 14:06 < matches> I'll try implement it and see if it's faster
4063 14:06 < matches> You may have to demonstrate my totally broken shading for me
4064 14:06 < matches> I haven't committed in a while
4066 14:41 < matches> SolveCubic may not have been the bottleneck there...
4067 14:41 < matches> -_-
4068 14:41 < matches> The failure to update the variable for a while loop miiiight be more of an issue
4069 14:44 < sulix> Good, 'cause I'm thinking of "upping the SolveCubic accuracy" a bit.
4070 14:44 < sulix> Anyway...
4072 14:46 < matches> I've reimplemented SolveCubic
4073 14:46 < matches> In theory it works
4074 14:55 < matches> I reckon Binary Search is the only algorithm I am good at implementing :S
4075 14:57 < sulix> Oooh...
4076 14:57 < matches> What?
4077 18:56 < matches> So, the fact that we are not solving cubics analytically is probably important in terms of precision issues
4078 18:58 < sulix> Yeah, that's why I started trying to solve them analytically and then stopped and cried.
4080 18:59 < matches> There is a solution, it's just ugly
4081 18:59 < sulix> Nah: for the Quadtree, it didn't matter much. As long as the error went in the correct direction, you'd just lost a tiny bit of efficiency.
4082 19:00 < matches> Hmm
4083 19:00 < matches> I don't expect shading using arbitrary precision arithmetic will be particularly fast anyway...
4084 19:01 < sulix> Well, ideally most of the work in the shading will be done in screen-space anyway.
4085 19:01 < matches> Yes
4086 19:02 < sulix> And, of course, you don't need infinite precision for that because pixels/integers/etc.
4087 19:03 < sulix> So, fingers crossed, things shouldn't matter too much.
4088 19:03 * sulix realises that that is a maddeningly vague statement.
4089 19:03 < matches> Well
4090 19:04 < matches> I got the fill point to actually be in the correct location by actually solving the intercepts correctly
4091 19:04 < matches> But
4092 19:04 < matches> If you zoom *out* far enough, it can end up no longer inside the shape when transformed to screen space
4093 19:04 < sulix> Okay, new plan.
4094 19:04 < sulix> No zooming out.
4095 19:05 < matches> I wish there were other people in this channel so I could justify putting that in qdb
4096 19:20 < matches> There is a temptation to not require all those references to be const during rendering
4097 19:20 < matches> We might want the Path to be able to alter itself based on the view
4098 19:20 < matches> But I guess we'll take that path when we get to it
4099 19:22 < matches> It might be worthwhile caching Path's for fonts
4100 19:22 < matches> But I'd have to change the coordinates to be relative
4101 19:22 < matches> Too many coord transforms
4102 19:24 < sulix> Yup, that can be the subtitle of the thesis: "Too many coord transforms"
4103 19:26 < matches> "A coordinated approach to graphics"
4104 19:26 < matches> Oh wow
4105 19:26 < matches> That should be a textbook
4106 19:26 < sulix> "Transformers: Matrices in disguise"?
4107 19:30 < sulix> Okay, it turns out (if denominator is zero, make denominator 1) has one flaw.
4108 19:30 < sulix> Sometimes you should make the denominator... negative 1.
4109 19:30 < matches> ...
4110 19:30 < sulix> Somehow, I did not see this coming.
4111 19:30 < matches> I'm not sure how this maths works
4112 19:30 < matches> 1/0 != 1/-1
4113 19:30 < sulix> That's what they want you to think.
4114 19:31 < sulix> (Or maybe the bug I'm having is unrelated to that, but it seems like a fun guess)
4115 21:16 < matches> Floating Point Exception in fglrx!
4116 21:16 < matches> glXDestroyContext
4117 21:16 < matches> It only happens when the program is about to close
4118 21:16 < matches> So I'll be fine
4119 21:18 < sulix> Ah, maybe enabling floating point exceptions was a bad idea.
4120 21:18 < sulix> ...if it causes lots of "fglrxceptions"!
4121 21:21 < matches> Well fglrx is exceptional
4122 21:37 < matches> I'm falling back on just testing a crap tonne of points for "inside" and adding all the ones that are to a vector
4123 21:37 < matches> I think there are some false negatives in Path::PointInside
4124 21:37 < matches> Also FloodFill misses some edge pixels which makes no sense
4125 21:38 < matches> If it's next to a filled pixel it should get filled!
4126 21:47 < matches> So the disadvantage of using a breadth first search is that when your queue gets really big instead of getting a segfault you just get OOM
4127 21:48 < matches> segfaults are quick
4128 21:48 < matches> segfaults are like a pistol and oom is like drowning
4129 21:54 < sulix> There is no way a BFS floodfill should ever OOM
4130 21:54 < sulix> How many pixels are there, and how much are you storing per one?
4131 21:58 < matches> Probably max 256x256 pixels, 4 way fill
4132 21:58 < matches> It does not like rabbit_simple.svg
4133 22:04 < matches> Yeah there is *something* causing it to get in a loop
4134 22:04 < matches> Because obviously my BFS didn't check for loops
4135 22:14 < matches> -_-
4136 22:14 < matches> It is in a loop
4137 22:14 < matches> Because there is a region with fill colour of white
4138 22:14 < matches> And the flood fill is just "If it isn't white, stop"
4139 22:14 < matches> Maybe I should set the background to have 0 alpha or something
4140 22:15 < matches> Alpha seems to not actually do anything at the moment
4141 22:27 < matches> http://szmoore.net/ipdf/who-the-hell-needs-antialiasing-anyway.png
4142 22:44 < matches> http://szmoore.net/ipdf/shady-the-fox.png
4143 22:47 < matches> rabbit_simple.svg actually doesn't work as well as rabbit.svg
4144 22:48 < matches> I think one of the fill points of the outer ear is inside the inner ear
4145 22:54 < sulix> Oh my... They're beautiful!
4146 23:07 < matches> Ah crap
4147 23:07 < matches> I will be pushing
4148 23:07 < matches> 8 hours in the future
4149 23:07 < matches> I really should fix my system clock
4150 23:07 < matches> One of these days...
4151 --- Day changed Mon Sep 01 2014
4152 17:05 < matches> The wobbly lines on things are irritating me
4153 17:10 < matches> There doesn't seem to be anything wrong with Bresenham, which means the actual line endpoints are wrong
4154 17:11 < matches> The GPU does everything from floats and magically they go to the right spot in the buffer
4155 17:11 < matches> The CPU has to calculate integer pixel coordinates
4156 17:16 < matches> I need a magnifier I can't see the pixels that are in the wrong spot
4157 17:16 < matches> I know they are there
4158 17:21 < matches> ... ok maybe the Bresen Ham was a little past its use by date
4159 19:48 < matches> The more I try and get these lines to not look dumb, the more convinced I become that glDrawLines is doing something fancy
4160 19:56 < matches> You end up with wiggles in the line because the end points round to a different pixel to what you end up at by drawing the line
4161 19:57 < matches> If you make sure you start from where the last line ended, you still get a bend because of the left over error from the last line
4162 19:58 < matches> ...
4163 19:58 < matches> Drawing a single line when you know the Bezier is actually a single line is probably going to be less rage inducing than trying to fix this
4164 21:01 < matches> The Beziers are now classified according to Loop and Blinn
4165 21:01 < matches> Well
4166 21:01 < matches> Maybe
4167 21:01 < matches> It at least knows what lines are
4168 21:01 < matches> ...
4169 21:02 < matches> Ok all the ttf beziers are SERPENTINE and LOOP so that's probably wrong
4170 21:43 < matches> Ah, floating point precision strikes again
4171 21:43 < matches> That's kind of annoying
4172 21:47 < matches> But I suppose it's relevant
4173 21:47 < matches> Unless I'm totally screwing something up
4174 21:48 * matches adds in arbitrary fabs(x) < 1e-6 checks
4175 21:58 < matches> On the other hand, if you use DeCasteljau and stop the sub dividing whenever you get to something that classifies as a line, it's actually pretty cool
4176 21:59 < matches> But if you zoom in too far it starts to get classifications totally wrong
4177 22:02 < matches> Yeah I have an abstract due in 2 weeks
4178 22:02 < matches> so screwed
4179 --- Day changed Tue Sep 02 2014
4180 13:42 < matches> Ok, let's not use floating point values for colours
4181 13:42 < matches> Baad
4182 13:43 < matches> Bad bad
4183 13:43 < matches> Bad sheep
4184 13:43 < matches> The project isn't about precision of representable colours
4185 13:44 < matches> So I'm perfectly OK with using uint8 for colour components
4186 18:53 < matches> Is it too late to change the aim of the project to "render svg-tests/fox.svg but without antialiasing"
4187 18:54 < matches> "Or using the GPU"
4188 18:55 < matches> Well, I understand why shading doesn't usually use flood fill now anyway
4189 18:55 < matches> But I still kind of think flood fill is better at high resolutions
4190 18:56 < matches> Not parallelisable I guess
4191 18:56 < matches> ... Or is it
4192 18:57 < matches> I was going to try and do arbitrary precision stuff but it seems we can no longer compile unless Real == double or float
4193 --- Day changed Wed Sep 03 2014
4194 17:26 < matches> It would seem Mr Munroe has beaten us...
4195 17:28 < matches> In Javascript
4196 17:28 < matches> We should die of shame
4197 17:29 < matches> Hahaha
4198 17:29 < matches> "// there is no elegance here. only sleep deprivation and regret"
4199 17:31 < matches> On the bright side, we now have a reason to reference xkcd
4200 17:31 < matches> I guess?
4201 17:59 < sulix> I just saw that, and had planned to use that comment as my next commit message.
4202 --- Day changed Thu Sep 04 2014
4203 10:59 * sulix sulix also engages panic mode.
4204 11:00 * sulix apparently also needs to engage sleep mode.
4205 11:00 * sulix almost engaged sheep mode by mistake.
4206 11:49 < sulix> How sure are you that SolveCubic works?
4207 18:37 < matches> Well, sure within 1e-4
4208 19:40 < matches> Rational<Gmpint> compiles again
4209 19:41 < matches> It just takes... several minutes... to load "The quick brown fox jumps over the lazy dog"
4210 21:04 < matches> I have a new ingenius plan
4211 21:06 < matches> This will probably not work but hopefully it won't take me too long...
4212 21:06 < matches> Oh dear what am I saying
4213 21:06 < matches> It's going to take a month now
4214 22:24 < matches> I really should be integrating one of those two libraries Rowan suggested into the project
4215 22:24 < matches> As opposed to implementing what I like to call the "Paranoid Number"
4216 22:25 < matches> It starts as a float
4217 22:25 < matches> It checks each operation for an exception, and if it does it adds the operation to a linked list instead
4218 22:25 < matches> After trying to simplify the next elements in the list
4219 22:26 < matches> This is going to be awful
4220 22:26 < matches> But it's been a while since I wrote a good linked list
4221 22:32 < matches> ... I think the last linked list I wrote had a nasty tendency to randomly segfault...
4222 22:32 < matches> Nevertheless we shall forge ahead
4223 23:14 < matches> You know you haven't had enough sleep when you start to terminate lines with ':' instead of ';'
4224 23:14 < matches> I blame python
4225 23:14 < matches> Not that it was valid python syntax either
4226 23:15 < matches> But it's a good scapesnake
4227 23:42 < matches> Oh recursive list based functions
4228 23:42 < matches> How I have missed you
4229 23:42 < matches> void Foo() {if (m_next != NULL) m_next->Foo();}
4231 --- Day changed Fri Sep 05 2014
4232 09:26 < matches> It sort of kind of works
4233 09:26 < matches> Except when it breaks
4234 09:26 < matches> ParanoidNumber::Simplify needs improvement
4235 09:26 < matches> Also operator+=
4236 09:26 < matches> operator-=
4237 09:26 < matches> operator*=
4238 09:26 < matches> and operator/=
4239 09:26 < matches> But it's definitely a number
4240 09:27 < matches> It just occasionally does things in totally the wrong order'
4241 09:28 < matches> So basically the idea is if you have an inexact (or overflow/underflow) operation, you delay it, and then in theory you can rearrange the operations so you have a smaller error...
4242 09:28 < matches> It seems to give better results than floats for tests/paranoidcalculator
4243 09:29 < matches> But that might be because it is initialised off 3/10 and the float starts at 0.3
4244 09:30 < matches> It could all be totally useless
4245 09:30 < matches> It probably is totally useless
4246 09:31 < matches> I have a hunch that even if it's terrible it might beat true arbitrary precision arithmetic
4247 09:31 < matches> In terms of actually being able to use it
4248 09:31 < matches> Who needs mathematics when you have blind intuition...
4249 --- Day changed Sun Sep 07 2014
4250 11:37 < sulix> Explicit mentions of hitting floating point precision issues in the xkcd comic: http://chromakode.com/post/notes-on-xkcd-pixels
4252 17:33 < unmercifulfish> mwahahaha
4253 17:34 < unmercifulfish> you can't escape my stalkerish tendencies now
4254 --- Day changed Mon Sep 08 2014
4255 11:10 < matches> This channel seemed like such a good idea at the time...
4256 11:19 < sulix> (So did the quadtree...)
4258 12:08 < Pommers> http://chromakode.com/post/notes-on-xkcd-pixels
4259 12:10 < sulix> It's gettign crowded in here.
4260 12:10 < sulix> Pommers: We were looking at that yesterday. The XKCD guys had it easy.
4261 12:11 < Pommers> Lets get all the people in here
4262 12:42 < unmercifulfish> IPDF PARTY
4263 13:49 < matches> Hmm, maths is harder than I thought
4264 15:41 < Pommers> Speaking of math, I know a guy who is looking for people versed in maths and willing to work on psuedo random number sets
4265 --- Day changed Wed Sep 10 2014
4266 11:50 < matches> So, compiling with mpfr reals is not as simple as including the mpfr real header and doing a typedef
4267 11:50 < matches> As there appear to be compiler errors in mpfr
4268 11:51 < matches> Does anyone actually check that things compile before including them in debian..
4269 11:53 < matches> I'd expect this sort of thing from Fedora
4270 11:58 < matches> Never mind it's namespace issues
4271 11:58 < matches> I'm sorry debian
4272 12:15 < matches> Something is horrendously broken about mpfrc++
4274 12:17 < matches> #if REAL >= REAL_RATIONAL was not a wise choice
4275 12:21 < sulix> Ah...
4276 12:24 < matches> mpfrc++ is at least as precise as doubles but slower
4277 12:24 < matches> I never actually found the limit for doubles
4278 12:24 < matches> I should do that first
4279 12:26 < matches> Hmm, things start to disappear at around 4e-11 but you can still add stuff and it looks fine
4280 12:27 < matches> Ok doubles go crazy at 1e-17
4281 12:27 < matches> 1e-15 even
4282 12:30 < sulix> Is that zooming in on the origin, or somewhere further away?
4283 12:31 < matches> Somewhere further away
4284 12:31 < matches> If you translate towards the origin you can actually see the shapes start to get better
4285 12:31 < matches> It's actually pretty cool
4286 12:35 < matches> There's something dodgy with starting the View with low width/height as opposed to zooming in to the same width/height
4287 12:37 < matches> mpreal doesn't seem to want to show anything
4288 12:39 < matches> I think it's well past time we automated the zooming/translating process
4289 12:39 < matches> Sitting here zooming for 20 minutes is not really an effective use of time...
4290 12:47 < matches> Speaking of which, the CPU shading evaluates the fill points lazilly now, so you don't have to wait 5 minutes even with doubles just to load an SVG
4291 15:39 < sulix> For all of the hangers on... quadtrees are now slightly more functional.
4292 15:39 < Pommers> "functional"
4294 18:40 < iceweaselOS> Apparently talking in ICC isn't allowed
4295 18:41 < iceweaselOS> ICC
4296 18:41 < iceweaselOS> dammit
4297 18:41 < iceweaselOS> it has autocorrect
4298 18:42 < iceweaselOS> Humphrey the rabbit does not work :(
4299 18:42 < iceweaselOS> it gets stuck in pause
4301 --- Day changed Thu Sep 11 2014
4302 10:53 < matches> I wonder who that was
4303 10:59 < matches> "... for some reason in irram they decided to let x<y loop forever if x=y ..."
4304 10:59 < matches> That's...
4306 10:59 < matches> Fine
4307 11:00 < matches> sulix: How many "x < y" checks do we have?
4308 11:10 < sulix> A finite number..., I guess?
4309 11:10 < matches> iRRAM appears to do some sort of fancy wrapping of MPFR
4310 11:11 < matches> So I'm not hopeful
4311 11:11 < matches> Considering MPFR seems to slow down hideously
4312 11:11 < sulix> Wrap all of the numbers inside other numbers.
4313 11:11 < matches> I did some more ParanoidNumber stuff
4314 11:11 < matches> Maybe one day it will actually give the correct result
4315 11:11 < matches> Then I can actually implement Simplify() ...
4316 11:12 < sulix> http://en.wikipedia.org/wiki/Binary_GCD_algorithm
4317 11:12 < sulix> (If you want fast gcd for bigints, btw)
4318 11:14 < matches> Should I plead "I had assignments due" or "I was seeing if Humphrey The Rabbit would work on a Firefox OS phone" in the progress meeting...
4319 11:14 < matches> iRRAM does seem to have a friendly install script
4320 11:15 < sulix> As long as you don't plead the FITH, you're probably fine. :P
4321 11:16 < matches> I'm fairly certain that floats are the best possible number representation and you need to do something more clever than just change your number representation if you want an arbitrary precision document
4322 11:16 < matches> Something like... some kind of... quad... tree
4323 11:17 < sulix> Do not use those words. You may not be able to vanquish what you summon.
4324 11:23 < matches> I need to write an abstract...
4325 11:24 < matches> "We compare different number representations and find them all to be useless"
4326 11:24 < matches> Hmm, something non commital
4327 11:25 < matches> "By the date of this conference we will hopefully have something interesting to say"
4328 11:25 < matches> "We discuss the drawbacks of adding IRC channel logs to git repositories"
4329 11:26 < matches> Ah nooo
4330 11:26 < matches> iRRAM uses "REAL"
4331 11:28 < matches> Also someone's been using too much javascript because there are all these extra ';' in their headers and it annoys -Werror=pedantic
4332 11:58 < sulix> ... :-(
4333 15:04 < matches> Spot the slight mathematical flaw in iRRAM
4334 15:04 < matches> inline LAZY_BOOLEAN operator == (const REAL& x, const REAL& y){ return (y<x)&&(x<y) ; }
4335 15:05 < matches> inline LAZY_BOOLEAN operator != (const REAL& x, const REAL& y){ return (y<x)||(x<y) ;}
4336 15:06 < matches> inline LAZY_BOOLEAN operator <= (const REAL& x, const REAL& y){ return (x<y); }
4337 15:06 < matches> They don't seem to have equality
4338 15:06 < matches> But that doesn't stop them implementing the operators...
4339 18:21 < sulix> Clearly the way to fix your computer's clock issues: http://jackf.net/bezier-clock/
4340 18:22 < matches> Against all better judgement I am still trying to fix ParanoidNumber
4341 18:22 < matches> And oh my goodness that is amazing
4342 18:23 < matches> But
4343 18:23 < matches> It still thinks it's 02:23:31
4344 18:23 < sulix> Unless I fix the quadtree view code, infinite precision only works if the coordinate you're zooming in on can _not_ be represented as a float.
4347 18:25 < sulix> So if you zoom in on a coordinate that is exactly representable as a sum of powers of two, then you're always zooming in on the boundary between two quadtree nodes.
4348 18:25 < matches> Haha
4349 18:26 < matches> Could you use an extra "centre" node?
4350 18:26 < matches> That would be redundant though
4352 18:26 < sulix> Nah, I just need to actually get around to supporting displaying content from more than one node at a time.
4353 18:32 < matches> ParanoidNumbers seem to actually construct properly now
4354 18:32 < matches> At least, if you construct them and convert back to doubles, they are within 5% of the double
4355 18:33 < matches> Now... simplifying...
4356 18:33 < sulix> With float ParanoidNumbers or double ParanoidNumbers?
4357 18:33 < matches> doubles
4358 18:33 < matches> But I'm testing the traversal of the list(s) when converting back
4359 18:34 < sulix> Ahhh...
4360 18:34 < matches> You need two recursive functions and I really hope I don't get stack overflows...
4361 18:35 < matches> You start with "value = Head.value * Head.MultiplyFactors + Head.AddTerms"
4362 18:35 < matches> Then define MultiplyFactors as "1 * f1.AddTerms * f2.AddTerms * ..."
4363 18:36 < matches> Then AddTerms is "0 + t1.MultiplyFactors + t2.MultiplyFactors + ..."
4364 18:36 < matches> It gets confusing how you deal with division and subtraction...
4365 18:37 < matches> I'm sure you can write that as a non recursive function... somehow...
4366 18:37 < matches> I don't seem to be overflowing the stack though
4367 18:38 < matches> So 666 operations have taken 5 minutes
4368 18:38 < matches> But that's without simplifying anything
4369 18:38 * sulix is currently traversing quadtree nodes via needlessly complicated recusion.
4370 18:38 < matches> I am quitely confident that I may be able to render a frame in at most 10 years
4371 18:39 < sulix> Clearly in your seminar, you should start zooming in and when it gets slow just say "If you'll bear with me for a moment" and then just leave the room.
4372 18:39 < matches> The thought has crossed my mind...
4373 18:41 < matches> Something that is kind of cool is I can change the digit to be something like uint8_t and see how well it converts to floats
4374 18:41 < sulix> That is pretty cool.
4375 18:41 < matches> It will be when I implement it that is
4376 18:41 < sulix> How well _does_ it convert to floats?
4377 18:42 < sulix> (Also, sudden thought: is NaN a valid mark for this project, as I am getting the feeling it should be...)
4378 18:42 < matches> My paranoid number tester does occasionally fail with "NaN != NaN"
4379 18:42 < matches> That's good!
4380 18:42 < matches> They both got NaN!
4381 18:43 < matches> Well, ideally I will be able to make it so that the ParanoidNumber *doesn't* get NaN
4382 19:17 < sulix> Hmm... I think the ext4 driver just crashed on my computer.
4384 19:18 < sulix> I didn't know that was possible.
4385 19:25 * sulix -> other computer
4386 19:31 < matches> I'm inventing cool terminology
4387 19:31 < matches> Paranoia
4388 19:31 < matches> Is the number of ParanoidNumbers...
4389 19:33 < sulix> "Peeking around corners looking for NaNs, wearing a tinfoil hat to guard against surious SIGFPEs, it's... ParanoidNumber!"
4390 19:33 < matches> That is the best description
4391 19:33 < matches> Well
4392 19:34 < matches> It actually kind of blindly walks around the corner into the SIGFPE and then freaks out and runs back around it
4393 19:35 < sulix> It's like running away from the fireball in slow motion.
4394 19:37 < matches> I'll at least win an award for most original number representation
4395 19:37 < matches> (He hopes)
4396 19:37 < matches> (It's probably not)
4397 19:38 < matches> (It's probably equivelant to one of the many alternative number representations that I didn't look up...)
4398 19:38 * sulix also hopes there is an award for most buggy implemention of a quadtree.
4399 19:38 < matches> (Specifically, the ones in the Handbook of Floating Point arithmetic that were described as "curiously useless")
4400 19:40 < sulix> (Googling "curiously useless" yields the phrase: "My boss is curiously useless. He fanatically creates an abundant amount of work for everyone while at the same time manages to get nothing done.")
4401 19:40 < sulix> (I suspect they're referring to iRRAM)
4402 19:40 < matches> Bahaha
4403 19:43 < matches> I think that simply replacing floats with int8_t actually does work
4404 19:44 < matches> Which is cool because it is easier to test if an operation with int8_t can be represented exactly
4405 19:44 < matches> But on the other hand there will be fewer operations that can be represented exactly
4406 19:45 < matches> I think this may actually work
4407 19:46 < matches> You don't need arbitrary precision to represent screen space and you are converting from view space to screen space
4408 19:47 < matches> So maybe it won't be that terrible, even though it would need a lot more digits to represent the same arbitrary precision value
4409 19:47 < matches> I think
4410 19:47 < sulix> You need more than an int8_t to represent screen space.
4411 19:47 < sulix> (An int16_t would do it, though)
4412 19:48 < matches> You just need to be able to represent each individual factor in the operation
4413 19:48 < matches> factor and term
4414 19:48 < sulix> You may (probably will) need factors > 127
4415 19:48 < matches> Probably
4416 19:50 < matches> If I just implement an "Exact" template function...
4417 19:50 < matches> Uh oh, templates
4418 19:51 * sulix thinks that he's approaching the worst recursive subroutine ever written.
4419 19:52 < sulix> (It has four switch statements in it so far)
4420 19:57 < matches> Template errors
4421 19:57 < matches> Aargahasdf
4422 19:58 < matches> Compilers are so good at identifying exactly what error you made and so useless at actually providing any hint of how you can fix it
4423 19:58 < matches> It's totally different from like "expected ';' before ..."
4424 19:59 < sulix> I saw a feature request for Google translate to support "gcc -> English".
4425 19:59 < matches> I want to specialise some template static member functions...
4426 19:59 < matches> That was my first mistake...
4427 20:00 < matches> Ok, they don't need to be static member functions, they can just be ... whatever you call a function that has no namespace associated with it
4428 20:01 < sulix> Functions...?
4429 20:01 < matches> Bah
4430 20:01 < matches> I would say "Regular functions" but they tend to be less common in C++
4431 20:02 < sulix> I've heard them called "first-class functions" before.
4432 20:03 < matches> That's a little elitist
4433 20:03 < sulix> I'm sort-of enamoured with the idea of "business-class functions" and "economy-class functions," though.
4434 20:11 < sulix> BTW, I just commited something that doesn't compile. Fixes will come shortly, it was just the easiest way to get it from computer A -> computer B.
4435 20:13 * sulix has made a typo and copy-pasted it into 16 places.
4436 20:13 < matches> copy-paste is best practice...
4437 20:14 < matches> It irks me that "a -= b" is equivelant to "a += (-b)" but there is no "a /= b == a *= (-b)"
4438 20:14 < matches> Well there is sort of, but only for rationals
4439 20:14 < sulix> The operator you're looking for is (1 / b)
4440 20:14 < matches> Yes, but - is quicker
4441 20:14 < sulix> This is true.
4442 20:15 < matches> I mean 3 can be represented exactly but 1/3 cannot
4443 20:15 < matches> With floats
4444 20:15 < matches> Most annoying
4445 20:15 < matches> I'm currently trying to reduce all the superflous "{0 + ...}" in ParanoidNumber
4446 20:16 < matches> It likes to create 1000 element lists with 900 zeroes in a row
4447 20:35 < unmercifulfish> this is gripping reading
4448 20:35 < unmercifulfish> so glad I am a part of this
4449 20:35 < matches> This is the cutting edge of useless research
4450 20:41 < sulix> "The bézier tiptoed out of bed. What an adventure it would be to go beyond the quadtree node in which it had lived its entire life. Its mother had told it sternly that it was dangerous to wander the dark wastes beyond the root node. Bobby Bézier had never believed the stories, though: there were no such things as wild NaNs that ate all of the coefficients of naughty boys. Bobby looked up at the sky: the infinite void of night penetrated only by the deb
4451 20:43 < Pommers> Oh yeah. One of the people in cs has a nice post on their door about rabbits that I want to grab a copy of
4452 20:50 < matches> I have a "TrustingOp" and "ParanoidOp"
4453 20:50 < matches> I think I'm having too much fun...
4454 20:54 * sulix has decided to give up actually coding and take up writing bézier curve fanfiction full time.
4455 20:55 < sulix> (People say it's derivative, but it's just about to reach a turning point)
4456 21:15 < unmercifulfish> qdb
4457 23:09 < matches> Oh dear
4458 23:09 * matches curses kahan
4459 23:10 < matches> I do not understand how this works, but if I delete terms that are "-0" things break
4460 23:13 < matches> So I got to the point where ParanoidNumber gave results that were reasonably close to just using a normal double without simplifying them
4461 23:14 < matches> And now I am basically adding random lines of code and running the tester and seeing if it still works :S
4462 --- Day changed Fri Sep 12 2014
4463 00:15 < matches> This reminds me of the James Micken comment about addition and multiplication basically being the same thing...
4464 01:19 < matches> Oh my goodness
4465 01:19 < matches> I think ParanoidNumber just worked for one very specific case!
4466 01:19 < matches> ParanoidNumber: 0.3 + 0.3 + 0.3 = 9/10, ToDouble() -> 0.9000000000000000222044604925031308084726
4467 01:20 < matches> double: 0.3 + 0.3 + 0.3 = 0.9000000000000001332267629550187848508358
4468 01:20 < matches> And strtod("0.9") = 0.9000000000000000222044604925031308084726
4469 01:23 < matches> For 0.2+0.2+0.3 = 7/10 it has an error of 1e-16 compared to doubles which apparently have no error
4470 01:24 < matches> But as floats it has an error of 1e-8 compared to 4e-8
4471 01:24 < matches> Hmm, why does 7.0 / 10.0 give a bigger error than 0.2 + 0.2 + 0.3
4473 01:24 < matches> It'll be denormals probably
4474 01:25 < matches> That is... quiet irritating
4475 01:27 < matches> So... it is sometimes better than a float but only if there is a full moon and you sacrifice thrice the number of one tenth of unity goats three times
4476 01:28 < matches> The other times it breaks because subtraction is totally FITH
4477 01:29 < matches> Oh, it should be able to do 1/3 + 1/3 + 1/3 except my calculator tester doesn't allow me to enter non decimal numbers
4478 01:32 < matches> I think 0.3+0.3+0.3 is the only thing it can do better than regular floats
4479 01:32 < matches> Woo
4480 01:40 < matches> Ah it is better at beating doubles when it is using doubles as the digit type
4481 01:40 < matches> But it still is sometimes worse if you have a small number of operations
4482 01:40 < matches> Probably because one division is less precise than several additions
4483 01:42 < matches> I'm going to keep telling myself that this is useful
4484 01:42 < matches> After I sleep
4485 01:51 < Pommers> It isn't currently a full moon...
4486 01:51 < Pommers> Also
4487 01:51 < Pommers> Go to bed
4488 11:27 < matches> It was a full moon after I sacrificed the goats
4489 11:40 < Pommers> Stop toying with space matches
4490 --- Day changed Sat Sep 13 2014
4491 09:25 < unmercifulfish> I feel that the logs for this chat should just be added to qdb
4492 09:25 < unmercifulfish> < matches> It was a full moon after I sacrificed the goats
4493 12:36 < Pommers> Should make a QDB bot for submitting them
4494 16:20 < matches> When sulix is awarded the nobel prize for his work on quad trees, I will be proud to be a part of this channel
4495 16:21 < Pommers> Until then you won't be?
4496 16:23 < matches> I've been thinking of adding a "grep -v matches" before the automatic log commit
4497 17:45 < matches> I am a horrible person
4498 17:45 < matches> I am exploiting that "delete" acting on a NULL pointer has no effect
4499 17:46 < matches> "delete Operation(new ParanoidNumber(a), ADD)"
4500 17:47 < matches> Where Operation will either return the first argument or NULL
4501 17:50 < matches> At the moment I'm very pleased with how cleverly I am about to shoot myself in the foot
4502 17:50 < matches> We'll see how I feel after I pull the trigger
4503 17:53 < matches> I have either written an extremely elegant or an extremely segfaulting function
4504 17:57 < matches> The latter is looking more and more likely
4505 17:57 < matches> But I'm sure there is beauty underneath
4506 17:57 < matches> Like the ugly duckling!
4507 17:59 < matches> There's still hope, it turns out I'm segfaulting way before I even get to that
4508 18:01 < matches> Apparently 2+1 is 6, but doubles think 2+1 is 5
4509 18:02 < matches> "Freedom is the freedom to say 2+2=4, if that is granted, all else follows"
4510 18:06 < matches> Ok, 1 + 1 is 2
4511 18:06 < matches> I think we can call that a result
4512 18:06 < matches> Write the thesis and we're done here
4513 22:18 < matches> 1 + 0.3 = 13/10 and 13/10 + 0.2 = Segfault
4514 22:18 < matches> Getting close
4515 22:19 < matches> Now I'm doomed to not sleep until it works because I'm never going to remember how this works
4516 22:20 < Pommers> Got a whiteboard?
4517 22:20 < matches> I do actually but it's in the other room
4518 22:20 < matches> And it does not have space for this
4519 22:22 < matches> I don't think I've ever used so many delete operators...
4520 22:23 < matches> Also constructing a 4 diretional list on the stack then stealing one of the branches
4521 22:23 < matches> It's a game of "dodge the double delete!"
4522 22:24 < matches> Which reminds me, we have copious memory leaks in an external library
4523 22:24 < matches> 'ld' ?
4524 22:25 < matches> /lib/x86_64-linux-gnu/ld-2.19.so
4525 22:25 < matches> Not sure if that's something I'm doing wrong
4526 22:26 < matches> Turns out you can't pipe valgrind's output to less
4527 --- Day changed Sun Sep 14 2014
4528 18:23 < matches> http://szmoore.net/ipdf/sam/abstract.txt
4529 18:24 < matches> You know
4530 18:24 < matches> It sounds a lot more boring than it has been
4531 18:24 < matches> I also need a title
4532 18:24 < matches> "Number Representations and Precision in Vector Graphics" ?
4533 18:24 < matches> I'm too scared to pick something punny...
4534 18:26 < matches> Libreoffice thinks that is 147 words and pluma thinks it is 150...
4536 18:26 < matches> hyphenated IEEE-754
4537 18:28 < matches> I think the first sentence could probably be less mind numbingly boring
4538 18:28 < matches> My last thesis started with "In this project" which seems terribly not academic
4539 18:29 < matches> (Every time I read that thesis I get a little more convinced that the markers didn't actually read any of it)
4540 18:29 < matches> ParanoidNumbers are not going well
4541 18:30 < matches> Like, it's hard to debug where things go wrong when you have a list like this:
4542 18:30 < matches> debug: main (tests/paranoidcalculator.cpp:75) - a is: {3*(23*(398123+(1/10+(2/100+(3/1000+(5/10000+(1/100000+(2/1e+06+(2/1e+07+(3/1e+08)))))))))+(3*(398123+(1/10+(2/100+(3/1000+(5/10000+(1/100000+(2/1e+06+(2/1e+07+(3/1e+08)))))))))/10+(4*(398123+(1/10+(2/100+(3/1000+(5/10000+(1/100000+(2/1e+06+(2/1e+07+(3/1e+08)))))))))/100+(1*(398123+(1/10+(2/100+(3/1000+(5/10000+(1/100000+(2/1e+06+(2/1e+07+(3/1e+08)))))))))/1000+(2*(398123+(1/10+(2/100+(3/1000+(5/10000+(1/100000+(2/1e+06+(2/1e+07+(3/1e+08)))))))))/10000+(3*(398123+(1/10+(2/100+(3/1000+(5/10000+(1/100000+(2/1e+06+(2/1e+07+(3/1e+08)))))))))/100000))))))/10+(1*(23*(398123+(1/10+(2/100+(3/1000+(5/10000+(1/100000+(2/1e+06+(2/1e+07+(3/1e+08)))))))))+(3*(398123+(1/10+(2/100+(3/1000+(5/10000+(1/1 ...
4543 18:31 < matches> That's without simplifying it
4544 18:31 < matches> But when it gives the wrong results before you even simplify it, that's bad
4545 18:32 < matches> I'm preeety sure it's not just giving the wrong results because it has a stupid number of terms/factors
4546 18:32 < matches> It's wrong enough that there's an extra factor or a factor missing or I don't even
4547 18:34 < matches> Hmm
4548 18:34 < matches> I want to add "Unlike paper, ..." before the third sentence
4549 18:34 < matches> I'm going to Captain Obvious
4551 18:38 < matches> Panic mode is well and truly activated now
4552 --- Day changed Mon Sep 15 2014
4553 14:00 < sulix> matches: You need to put the title in: "Keepin' it real: Staying afloat on the sea of document precision"
4554 --- Day changed Tue Sep 16 2014
4555 12:25 < matches> I think I'm sinking
4556 12:27 * sulix has failed to come up with a suitable sinking/syncing pun.
4557 12:27 < sulix> Or, for that matter, a reason why the quadtree clipping sometimes breaks.
4558 12:33 < matches> Hey, at least multiplication doesn't sometimes break
4559 12:33 < matches> That's fairly major
4560 12:34 < matches> My graph traversal for the paranoid number is somewhat flawed
4561 12:34 < matches> Yeah it's a giant graph
4562 12:34 < matches> Smell the memory cooking
4563 12:34 < matches> I should do some performance tests
4564 12:34 < matches> Even if I don't have anything to performance test...
4565 12:34 < matches> We did promise graphs
4566 12:35 < matches> (The other kind of graph)
4567 12:35 < sulix> How are you traversing the graph?
4568 12:35 < sulix> (The first kind)
4569 12:36 < matches> Originally I was alternating between multipying and adding, but that doesn't work if the first term in a factor has factors of its own
4570 12:36 < matches> I could probably still use that but I'd have to add zeroes or something
4571 12:37 < matches> Thinking of it as a linked list is wrong anyway, as factors can have terms and terms can have factors and I'm not making any sense am I
4572 12:37 < matches> Maths can be made up of maths
4573 12:37 < matches> I kind of regret basically promising to have this done in my abstract...
4574 12:37 < sulix> Hmmm...
4575 12:40 < matches> Anyway, either I am creating the graph wrong and traversing it correctly, or creating it right and traversing it incorrectly, or they are just both totally wrong
4576 12:41 < matches> Leaning towards the latter
4577 12:43 < sulix> You need to traverse it from the bottom-up, I think.
4578 12:50 < matches> Yeah, I need to change both the creation and traversal
4579 12:50 < matches> Repeated divisions need to change into a division of a number multiplied by a number
4580 12:51 < matches> Then the traversal is relatively trivial
4581 12:55 < matches> Division is annoying
4582 12:55 < matches> Why does maths need division
4583 12:55 < matches> All the other operations make sense
4584 12:56 < matches> Someone should reinvent mathematics without a concept of division
4585 12:56 < matches> We'd all be happier
4586 12:56 < sulix> I'm pretty certain there's a formal proof of that.
4587 12:57 < sulix> Most number systems don't have the concept of division.
4588 12:57 < matches> Haha
4589 12:57 < sulix> (The integers, for example)
4590 12:57 < matches> They have division with remainder though?
4592 12:58 < matches> But you don't get a single number out
4593 12:58 < matches> You get two
4594 12:58 < matches> So it's not a proper operation?
4595 13:00 < sulix> It's not a binary operation, and it's not the inverse of division.
4596 13:00 < sulix> *multiplication
4597 13:07 < Pommers> Guild debate. Pretty sure I have the best camera angle here
4598 13:43 < matches> Guild debates are beyond the scope of this project
4599 13:44 < matches> Urgh I'm regretting distinguishing between ADD and SUBTRACT now
4600 13:44 < matches> Just ADD and MULTIPLY should be all the universe needs
4601 13:51 < matches> You have to be doing something wrong when you are using a triple pointer
4602 13:57 < matches> Delicious triple pointers
4603 13:57 < matches> Now I feel like a triple chocolate milkshake
4604 13:58 < matches> Ok, I have integers 0->9 adding and subtracting preeettty well now
4605 14:13 < sulix> Ah, the power of modern technology!
4606 14:29 < sulix> When in doubt, more debug output!
4607 14:55 < sulix> Ah ha! I have solve (read: hacked horribly around) the quadtree bug.
4608 14:56 < Pommers> http://xkcd.com/859/
4609 14:56 < sulix> Basically, adding a bunch of fake roots in fixes it.
4610 14:56 < sulix> Details at 11.
4611 14:56 < sulix> Pommers: You are a terrible person.
4612 14:57 < Pommers> Oh wait. It's there. I should sit closer to my screen
4613 15:32 < matches> Ok, I'm pretty sure Paranoid Numbers are the worst idea ever
4614 15:32 < matches> Possibly worse than iRRAM
4615 15:32 < matches> I mean, in theory you can simplify things
4616 15:32 < matches> But when you can't, you just use ridiculous amounts of memory
4617 15:33 < matches> I don't think you can simplify that many things either
4618 15:33 < matches> If I get it working enough I can demonstrate that it is a terrible idea at least...
4619 16:46 < matches> Oh my goodness it is kind of working
4620 16:46 < matches> I've said this before...
4621 16:47 < matches> But it's at least at the point where it gives suspiciously similar answers to the right answer
4622 16:48 < matches> As opposed to suspiciously off by factors of a random number
4623 17:31 < matches> So I think all I have to do now is simplify the damn things
4624 17:31 < matches> There are all sorts of amazing optimisations like not using a std::vector for the terms and not using seperate std::vector for + and -
4625 17:31 < matches> And I think you can probably optimise it even more if you just use doubles
4626 17:31 < matches> But
4627 17:31 < matches> It is better at repeated division/multiplications
4628 17:32 < matches> See TestMulDivIntegers
4629 17:32 < matches> By better I mean slower
4630 18:50 < matches> I have pushed a thing
4631 18:50 < matches> Try not to judge me too harshly
4632 18:51 < matches> Also there's no way it will compile on Cabellera anymore
4634 23:06 < matches> I just made SVGs load into the centre of the view...
4635 23:06 < matches> And then I realised that is exactly the sort of thing that will break sulix's quad tree
4636 23:06 < matches> Sorry sulix
4637 23:07 < matches> My role in this project is basically to make sulix have to do more work...
4638 23:21 < Pommers> Why'd you break cabellera?
4639 23:30 < matches> I didn't break cabellera
4640 23:30 < matches> I used C++11 auto range iterators
4641 23:30 < matches> for (auto add : m_next[add])
4642 23:31 < matches> As opposed to: for (vector<ParanoidNumber*>::iterator i = m_next[ADD].begin(); i != m_next[ADD].end(); ++i) ParanoidNumber * add = *i; ...
4643 23:31 < matches> Cabellera's g++ is too old for such amazing things
4644 23:32 < matches> But I was at a point where I needed to change like 50 things from iterating over linked lists to iterating over some kind of hideous graph of std::vectors
4645 23:32 < matches> Stuff typing all that
4646 23:32 < matches> I broke the quad tree because I thought it would be nice to have SVGs automagically appear in the middle of the screen
4647 23:33 < matches> As opposed to the top left underneath all the debug output
4648 23:49 < sulix> matches: You are too kind.
4649 23:49 < sulix> The quadtrees are actually working fine at the moment, by virtue of me modifying SolveCubic to no longer solve cubics.
4650 23:51 < Pommers> return null;?
4651 --- Day changed Wed Sep 17 2014
4652 00:03 < matches> Wait what
4653 00:04 < matches> Did you remove Cubic Beziers
4654 00:04 < matches> Because I liked having Cubic Beziers
4655 00:04 < sulix> I'm still trying to work out why it works myself.
4656 00:04 < sulix> Nope, still cubic béziers, solvecubic now just returns some things which I'm pretty certain actually aren't roots as well.
4657 00:05 < Pommers> Do either of you want to trade? I'll work on iPDF and you can fix my housemates computer
4658 00:05 < matches> Hah
4659 00:05 < matches> Set up that Jenkins server and we'll talk
4660 00:06 < Pommers> The Container exists for Jenkins
4661 00:06 < Pommers> So it's been started
4662 00:06 < matches> I wish CS would stop thinking up names for things that are also names for other things
4663 00:06 < matches> Container
4664 00:06 < matches> Puppet
4665 00:06 < matches> Orchestration
4666 00:06 < matches> For god's sake
4667 00:06 < matches> Paranoid Number is totally a legit name
4668 00:07 < Pommers> Stop being paranoid about things matches
4669 00:07 < matches> "Tree" is acceptable because it at least looks like a tree
4670 00:07 < Pommers> Also, coming to dinner?
4671 00:07 < matches> There is dinner?
4672 00:07 < matches> When?
4673 00:07 < matches> It's midnight...
4674 00:07 < Pommers> Thursday
4676 00:07 < matches> I suppose
4677 00:08 < matches> Thursday is typically the post meeting freak the hell out day
4678 00:08 < matches> Monday Tuesday and Wednesday being the pre meeting freak the hell out days
4679 00:08 < matches> Friday is the I should really prepare for my Physics tute day
4680 00:09 < Pommers> Saturday is "o god coderdojo day"
4681 00:09 < matches> Yeah
4682 00:09 < matches> Sunday is I should really not be playing computer games but I will anyway day
4683 00:09 < matches> No wait last sunday was a freak the hell out day as well, until they made me play monopoly
4684 00:09 < matches> Sorry, monotony
4685 00:10 < matches> Anyway how do you get valgrind to print number of flops
4686 00:10 < sulix> I'm just s/.*day/PANIC day/g at the moment.
4687 00:10 < matches> Your thing works!
4688 00:10 < matches> Sort of
4689 00:11 < sulix> thing?
4690 00:11 < matches> Quadtrees
4691 00:11 < Pommers> "thing"
4692 00:11 < sulix> Technically it works*
4693 00:11 < Pommers> "it"
4694 00:11 < matches> I have about 4 things I still haven't even tried yet
4695 00:11 < matches> And a couple of things that will work if you wait the lifetime of the universe
4696 00:12 < Pommers> Hmm... I should make my bed while this backup happens
4697 00:12 < matches> Bed is tempting
4698 00:12 < sulix> *The roots I'm using might not actually be roots, it doesn't work if you either pan, zoom out or anything other than zoom in on a single point forever.
4699 00:12 < Pommers> But having too much fun coding?
4700 00:13 < matches> I'm working towards actually making graphs of something
4701 00:13 < matches> I'm still working out exactly what I will graph
4702 00:13 < matches> flops seems important
4703 00:13 < matches> I'm basically replacing a single inexact flop with a lot of usually but not always exact flops
4704 00:14 < matches> And several hundred MB of memory...
4705 00:18 < matches> Ok, this can't be right
4706 00:18 < matches> To render one frame of an empty document takes 137,708 F64 operations
4707 00:18 < matches> I assume that means double
4708 00:19 < Pommers> That could use some optimizing...
4709 00:19 < Pommers> Make all the functions return null and see how that speeds it up
4710 00:19 < matches> -_-
4711 00:19 < matches> This is using regular doubles
4712 00:19 < matches> Not Paranoid Numbers
4713 00:20 < matches> ParanoidNumbers takes about 5 minutes to do 20000 operations at the moment...
4714 00:20 < sulix> I could believe it.
4715 00:22 < Pommers> I'd be paranoid if it took that long
4716 00:22 < matches> Don't make fun of the Paranoid Numbers
4717 00:22 < matches> They are a work of art
4718 00:23 < sulix> Part of me really wants to try compiling the quadtree with them, and part of me is terrified at the prospect.
4719 00:23 < matches> There isn't a REALTYPE for them yet
4720 00:25 < matches> They might work better if you had a giant Trie of them but I have no idea how to structure that
4721 00:25 < matches> At the moment there is one tree per number and the tree can have a variable number of children for each node... :S
4722 00:31 < matches> Anyway what performance metric do we want
4723 00:31 * matches consults the proposal
4724 00:32 < Pommers> What will be easier to do? Time or space?
4725 00:32 < matches> Time
4726 00:32 < matches> But Rowan was kind of keen on Space
4727 00:32 < matches> Number of flops is probably the best
4728 00:32 < Pommers> Just put some ascii art daleks in
4729 00:34 < matches> Ok, so 1 frame = 30,897 F32, 129,042 F64 and 1000 frames = 4,364,459 F32, 6,933,526 F64
4730 00:35 < matches> 6800 doubles to render a blank document
4731 00:35 < matches> Actually
4733 00:35 < matches> There is a rectangle in it!
4734 00:35 < matches> Of width and height 0
4735 00:40 < Pommers> I almost have the container for jenkins fully set up now
4736 00:43 < Pommers> I just need to give it an IP and then just set up jenkins
4737 00:45 < matches> I was joking about Jenkins...
4738 00:46 < Pommers> I actually have a use for it...
4739 01:16 < Pommers> Housemates computer is fixed
4740 01:16 * Pommers sleeps
4741 01:29 < matches> Dammit
4742 01:29 < matches> I entered a timestamp of 10:56 instead of 22:56 and now my commit is at 13:27
4743 01:30 < matches> I tried git commit --amend --date= but it didn't quite work
4744 01:31 < sulix> Would it count as cheating on the performance metrics if time travel was involved?
4745 01:31 < matches> Nooo
4746 01:32 < sulix> As you can see, ParanoidNumber actually takes only -12 hours to render a frame by being so cautious it violates causality.
4747 01:32 < matches> So, the "spatial" metric can be: Number of pixels not the same, and mean minimum distance horizontally/vertically to the nearest set pixel in the original
4748 01:32 < matches> Because I said so
4749 01:32 < matches> A possible other metric would be the Fourier Transform
4750 01:33 < matches> But that's stupid
4751 01:33 < matches> -12 hours is pretty good
4752 14:43 < Pommers> Jenkins is set up and I can let you create an account if you want to play with it
4753 17:01 < matches> Best idea ever
4754 17:02 < matches> A performance test that saves the graph as an svg
4755 17:02 < matches> Then runs itself on the graph...
4756 17:05 < Pommers> recursive testing?
4757 17:05 < matches> Yes!
4758 17:07 < Pommers> Set your testing up in jenkins
4759 17:08 < matches> This is performance tests not sanity tests
4760 17:08 < matches> Or whatever they're called
4761 17:08 < matches> We don't need to make sure things work...
4762 17:08 < matches> We just need them to not work fastly
4763 18:35 < matches> SDL_WINDOW_HIDDEN is useful
4764 18:36 < matches> Actually I'm surprised it works
4765 18:37 < matches> As much as having ipdf constantly flashing up and stealing my window focus is exciting, I am relieved to be able to stop that
4766 18:37 < matches> Now I just need to stop the monitor from randomly flashing its menu and/or turning off...
4767 18:42 < matches> So I can show some graphs tomorrow, although they may not be graphs of actually useful things
4768 19:44 < matches> I hope you like ipython because we're using ipython
4769 19:50 < matches> git.ucc seems sick...
4770 --- Day changed Thu Sep 18 2014
4771 11:40 < matches> Good news!
4772 11:40 < matches> You can run ipdf on motsugo
4773 11:40 < sulix> I heard.
4774 11:40 < sulix> My good news is that I just added something to the quadtree.
4775 11:40 < sulix> (Sort-of)
4776 11:41 < matches> The GPU rendering doesn't work on motsugo though
4777 11:41 < sulix> I fail to be surprised.
4778 11:42 < matches> Does the quadtree work with the centred SVGs?
4779 11:44 < sulix> I haven't been brave enough to do the horrible merge that needs to happen.
4780 11:45 < sulix> I'm currently trying to work out how I'm going to pass which node that you want to add something to through all of the Add* code.
4782 11:57 < matches> I'm about to hack around C++ 'const' correctness
4783 11:57 < matches> What could possibly go left
4784 11:57 < matches> I mean wright
4785 11:57 < matches> I mean lng
4786 11:57 < matches> asdf
4787 11:58 < matches> segmentation fault
4788 12:00 < matches> Well it compiled at least
4789 12:00 * matches notes that if he ever sees a function declared 'const' again he should not assume that it is actually const...
4790 12:01 < matches> Now to overload the new and delete operators...
4791 12:04 * sulix hides quietly in a corner.
4792 12:05 < matches> ParanoidNumber needs to call new and delete an excessive amount
4793 12:05 < matches> Because you might need to create a new one but you might not, but you can't tell in advance :S
4794 12:06 < sulix> So: pool or arena?
4795 12:06 < sulix> (I'd probably recommend an arena)
4796 12:06 < matches> I've not heard of an arena, which probably means its better
4797 12:07 < matches> It sounds traumatic though
4798 12:07 < matches> Remember these numbers are paranoid as it is
4799 12:08 < sulix> Basically allocate a large block of memory, every time you allocate return the next <n> bytes of it.
4800 12:08 < sulix> Delete is a no-op, everything is freed when you delete the entire arena.
4801 12:10 < matches> Hm I thought that was a pool
4802 12:11 < sulix> A pool is basically an array of objects of the same type that get reused.
4803 12:12 < sulix> Though nomenclature is pretty dodgy generally wrt memory management.
4804 12:12 < matches> SO thinks they are the same
4805 12:12 < matches> http://stackoverflow.com/questions/13381123/whats-the-difference-between-a-memory-arena-and-a-memory-allocator
4806 12:12 < matches> Actually no that's not quite the question I was looking for
4807 12:13 < matches> I said pool and it says "allocator"
4808 12:13 < matches> (But some guy with upvotes says arenas are also called pools)
4809 12:13 < matches> (Democracy has decided!)
4810 12:13 < matches> Ok I see the difference
4811 12:13 < matches> Yeah, I was going to use a pool, but I guess arena would be better
4812 12:14 < sulix> http://en.wikipedia.org/wiki/Memory_pool
4813 12:15 < sulix> http://en.wikipedia.org/wiki/Region-based_memory_management
4814 12:28 < matches> So, PN gives "inf" for an operation that doubles do fine
4815 12:29 < matches> But on the plus side, Wolfram Alpha gives "Wolfram alpha does not understand your query -344+(6+8/10 +...)"
4816 12:29 < matches> "Showing results instead for "6+8/10 +3/100"
4817 12:30 < matches> (The maths is syntactically correct, or at least, I'm fairly sure it is. I'm not actually going to count all those brackets...)
4818 12:31 < matches> I find it amusing that there is a number representation that is just "That number is too complicated, here is a similar looking one"
4819 12:32 < matches> Actually, correction
4820 12:32 < matches> *addition rather
4821 12:32 < matches> "Insert more money to calculate the number"
4822 12:32 < sulix> "Upgrade to wolfram alpha pro to use numbers with more factors"
4823 12:32 < matches> Yes
4824 12:32 < matches> That is pretty much what it says
4825 12:34 < matches> Hmm, I think PN should take a couple of minutes per frame at the moment
4826 12:34 < matches> I should probably actually test it
4827 12:39 < matches> That's a whole lot of compiler errors
4828 12:39 < matches> Argh
4829 12:39 < matches> I hate the "overloaded call to ClassConstructor(int) is ambiguous" errors
4830 12:40 < sulix> I think the idea of "compiler errors" is clearly too negative.
4831 12:41 < sulix> We should have "constructive compiler feedback" instead.
4832 12:41 < sulix> "warnings" should become "suggestions"
4833 12:41 < matches> I don't understand that type of eror
4834 12:41 < matches> The compiler is supposed to silently typecast the primitive types right?
4836 12:41 < matches> So if you have a constructor for a double it will cast "0" to "0.0" and use that
4838 12:42 < matches> If you implement constructors for *both* it will get confused
4839 12:42 < sulix> (If there isn't another constructor that is equal or better)
4840 12:42 < matches> If you implement constructors for only *one* it *still* gets confused
4841 12:42 < matches> Aaskdfasdf
4842 12:43 < matches> It's totally confused
4843 12:43 < matches> I think
4844 12:43 < matches> It sees the int
4845 12:44 < matches> Then goes "I can construct a ParanoidNumber off that"
4846 12:44 < matches> Then goes "But I can construct a ParanoidNumber off the const ParanoidNumber &"
4847 12:44 < matches> That doesn't make sense
4848 12:44 < matches> Why is the copy constructor ambiguous
4849 12:45 < sulix> That's exactly what I hate about c++.
4850 12:45 < sulix> (Also boost)
4851 12:51 < matches> It gets Real(0) mixed up with Real(const char * str)
4852 12:51 < matches> Really
4853 12:51 < matches> Why
4854 12:51 < matches> FFS
4855 12:52 < matches> std::string it is then
4856 12:52 < matches> Don't know why it had to complain about every single other constructor being ambiguous as well
4857 13:01 < matches> I've just realised
4858 13:01 < matches> ParanoidNumber is going to have to be thread safe
4859 13:01 < matches> As long as we want the control panel to work anyway
4860 13:01 < matches> Blergh
4861 13:05 < matches> Oh boy
4862 13:05 < matches> You know you are in trouble when "this == NULL"
4863 13:06 < matches> I feel like I am messing with forces beyond my ken
4864 13:06 < matches> I haven't even overloaded new yet
4865 13:06 < sulix> this == NULL is actually perfectly possible.
4866 13:07 < matches> Yeah but where is it coming from
4867 13:07 < sulix> It's the correct result if you call a member function of a null pointer.
4868 13:07 < matches> Yeah I know
4869 13:07 < matches> But
4870 13:07 < matches> But
4871 13:07 < sulix> (Is it time for captain valgrind?)
4872 13:07 < matches> That spits out so many errors due to other libraries :(
4873 13:07 < matches> I'm learning to use gdb more effectively
4874 13:07 < matches> I finally learned how to set breakpoints
4875 13:08 < matches> And print variables
4876 13:08 < matches> You can even get it to call functions
4877 13:08 < matches> That's cool
4878 13:08 < sulix> Yeah, gdb is pretty amazing.
4879 13:08 < matches> So I have this four dimensional tree of death
4880 13:08 < sulix> What you should do is setup the libstdc++ pretty printers.
4881 13:08 < matches> The child links are implemented as a std::vector of pointers
4882 13:08 < sulix> It makes gdb print stl containers nicely.
4883 13:09 < matches> Somehow there are null pointers in the vector
4884 13:09 < matches> That should really not happen
4885 13:09 < matches> I could check that the pointers are not null but I suspect that won't solve the underlying issue
4886 13:09 < matches> Of null pointers ending up in places
4887 13:10 < matches> Couldn't I do something like: 'p (Pointer*)vector.data()'
4888 13:10 < matches> I should try that
4889 13:11 < sulix> PANIC PANIC PANIC
4890 13:12 < sulix> Apparently my draft dissertation is due tomorrow.
4891 13:12 < matches> Wahtasdfasdf
4892 13:12 < matches> What
4893 13:12 < matches> When did this happen
4894 13:13 < sulix> http://undergraduate.csse.uwa.edu.au/year4/Current/dates.html
4896 13:14 < matches> I thought Rowan said you had two weeks...
4897 13:14 < matches> Or was that two weeks ago..
4898 13:14 < matches> Also I know I don't technically have to submit a draft dissertation but I share some level of panic
4899 13:15 < sulix> Yeah, I thought I had more time, too.
4900 13:15 < matches> I'm sure you can organise it with Rowan
4901 13:16 < matches> My last thesis was somewhat of a blur
4902 13:16 < matches> I'm not sure there was a draft
4903 13:17 < matches> You'll get through it
4904 13:20 < sulix> (The question is not so much if I'll get through it, but if I'll get through it in the next hour and forty minutes... :P)
4905 13:22 < matches> I give you permission to copy/paste my lit review
4906 13:22 < matches> Although to be honest it could do with less rambling about svg and postscript
4907 13:26 < sulix> btw, I've found a way to make the quadtree run as slowly as the rationals.
4908 13:28 < matches> How?
4909 13:28 < sulix> Totally broken linked lists?
4910 13:28 < sulix> (I think)
4911 13:29 < sulix> QuadTree nodes can now "overlay" other quadtree nodes, which is the way I'm hacking mutability in.
4912 13:29 < sulix> For some reason, if you zoom in and then zoom back out again, it ends up creating a new overlay for each individual curve.
4913 13:30 < sulix> Each overlay runs the entire View::Render function again, independently.
4914 13:30 < matches> Damn
4915 13:30 < matches> I don't understand how these vectors have NULL in them
4916 13:30 < matches> I have "assert(b != NULL)" before the only two places where things get added to them
4917 13:31 < sulix> Are asserts enabled? :P
4918 13:32 < matches> I thought that was Java
4919 13:32 < matches> Also python
4920 13:34 < sulix> I'm not actually sure about the standard asserts, but most of them do get disabled if you're not in a debug build.
4921 13:34 < matches> Ooh
4922 13:34 < matches> oh oh oh
4923 13:34 < matches> I always do this
4924 13:34 < matches> std::vector::vector does not construct with a size
4925 13:35 < matches> It constructs with elements
4926 13:35 < matches> So vector n(0) is baad
4927 13:35 < sulix> I always use vector(size, initialvalue)
4928 13:35 < matches> Nope that isn't it...
4929 13:35 < sulix> Or just use the default constructor if I want "empty".
4930 13:36 < matches> I'm already wrapping the billions of 'new's in a check that it isn't null
4931 13:36 < matches> What is going on
4932 14:40 < Pommers> matches: Fix your NTP already
4933 16:28 < matches> sulix: "That's actually not too bad, I may be rationalising it, but..."
4934 16:28 < matches> (We wouldn't want the spectators to miss any of the puns)
4935 16:28 < sulix> (For context: talking about the performance and correctness of rational numbers)
4936 23:35 < matches> Your predictions of Simplify leading to infinite loops were remarkably accurate
4937 23:36 < matches> That just means the number can be infinitely Simplified!
4938 23:49 < sulix> Simplify() { return 0; /* it can't get much more simple */ }
4939 23:49 < matches> Haha
4940 23:50 < matches> Well I just spent 15 minutes learning why I had the stupid ".test" extension on all the testers in the first place
4941 23:50 < matches> (Which I recently removed because it was stupid)
4942 23:50 < matches> Turns out it was there so that make clean could remove all the testers without needing to keep a list of them all
4943 23:51 < matches> And that no matter how many times I ran make clean, I was still running the same buggy executable from 7 hours ago...
4944 23:51 < matches> "Not so stupid after all!" says past matches
4945 23:51 < matches> "Shut up" says present matches
4946 23:52 < matches> "When are we going to sleep" says future matches at approximately 3am
4947 23:53 < matches> Things get wierd when you remove NTP
4948 23:53 < matches> I have a watch on my wrist but it's not the same
4949 23:55 < matches> "You know what the .test extension is still dumb, I'll just remember to rm them when they need rebuilding" says a slightly older present matches
4950 --- Day changed Fri Sep 19 2014
4951 00:00 < matches> Ah, good old copy paste and naming variables 'n' and 'm'
4952 00:01 < matches> Mever fails to segfault
4953 01:05 < matches> PN Yields: 1.0000000000000000000000000000000000000000
4954 01:05 < matches> PN is: 1
4955 01:06 < matches> Doubles Yield: inf
4956 01:06 < matches> TAKE THAT KAHAN
4957 01:06 < matches> Hmm
4958 01:06 < matches> I could multithread the paranoid number...
4959 01:07 < matches> At least, the Convert to double part of it
4960 01:11 < matches> Ah, it fails on the very next test though
4961 01:11 * matches regrest uncommenting the "return 0;"
4962 01:11 < matches> grects
4963 01:11 < matches> regrects
4964 01:11 < matches> dammit
4965 01:11 < matches> it is too aerly for this
4966 01:21 < matches> So ipython stores pngs as giant base64 encoded strings
4967 01:21 < matches> In JSON format
4968 01:21 * matches shudders at the commit diff
4969 01:23 < matches> Good luck sulix
4970 09:47 < sulix> "Bang, zoom, straight to the rounding error"?
4971 12:17 * sulix falls over in a heap of insanity and shame.
4972 17:20 -!- mode/#ipdf [+o matches] by OperServ
4973 17:20 -!- matches changed the topic of #ipdf to: IPDF: A Document Precision Playground
4974 17:20 <@matches> This is good reading
4975 17:21 <@matches> I just read "At its core, IPDF breaks documents" and choked on my drink
4976 17:22 * sulix thinks we might have a new slogan.
4977 17:23 <@matches> I think it's important that you don't specify what the initials actually stand for
4978 17:23 <@matches> People will think it's a fancy PDF viewer
4979 17:23 <@matches> Except it's really a broken SVG viewer
4980 17:25 < sulix> I don't intend to give any more explanation than IPDF™.
4981 17:25 < sulix> Or possible IPDFⓇ
4982 17:25 <@matches> I am jealous of your compose key
4983 17:26 < sulix> Mwäháhà!
4984 17:33 <@matches> Ah yeah, precision vs zoom doesn't work to well for doubles for some reason
4985 18:24 <@matches> Look at ParanoidNumber go
4986 18:24 <@matches> Look at it go!
4987 18:24 <@matches> Well you can't because you're not here
4988 18:24 <@matches> I should do a live stream
4989 18:24 <@matches> It is so exciting
4990 18:24 <@matches> It's been going 5 minutes but it hasn't failed any tests since I commented out the comparison against doubles
4991 18:25 <@matches> Yes
4992 18:25 <@matches> Eat all the memory
4993 18:25 <@matches> Soon you will grow strong
4994 18:27 <@matches> Actually I probably *want* it to start failing tests
4995 18:28 <@matches> Because if it gives the same result as a double it's not any better than a double...
4996 18:28 <@matches> At the very least it is mathematically interesting
4997 18:29 <@matches> I'm not sure Mathematically interesting and "Chemical Engineering Conference" go together
4998 18:29 <@matches> Also I don't have enough Maths training to say anything about it other than it is interesting
4999 18:29 <@matches> It is a ring of complete hidden sets of fields
5000 18:29 <@matches> Yeah
5001 18:29 * matches uses maths
5002 18:30 < sulix> Technically, I don't think that things which use floats can ever be rings or fields due to the whole rounding error thing.
5003 18:34 <@matches> Well
5004 18:34 <@matches> It's a tree
5005 18:34 <@matches> I can hold on to that
5006 18:34 <@matches> I understand trees
5007 18:35 <@matches> They have trunks and bark and leaves
5008 18:35 <@matches> Hmm, I should add a Bark sub class
5009 18:35 <@matches> I wonder if I can bring myself to implement the same test algorithm in Mathematica and compare the results
5010 18:36 <@matches> I think using Mathematica in research requires you to sell your soul though
5011 18:41 <@matches> Also Frames thinks that enabling optimisation will magically make PN less terrible
5012 18:41 <@matches> I'm somewhat skeptical
5013 18:41 <@matches> But here goes
5014 18:42 <@matches> (I am anticipating it will make it much faster, in that it will segfault after 2 operations)
5015 18:45 <@matches> Well it didn't segfault, but apparently "219123 != 219123"
5016 18:48 <@matches> Actually
5017 18:48 <@matches> It seems a lot faster
5018 18:48 <@matches> Hooray Frames
5019 18:48 <@matches> Thesis done
5020 18:49 < sulix> Does it speed up the GMP and/or Arbint rationals much?
5021 18:49 <@matches> Haven't tried yet
5022 18:49 <@matches> I need to adapt Paranoidtester into a general tester
5023 18:49 <@matches> Or was that realops
5024 18:49 <@matches> I forget
5025 18:49 <@matches> I have a lot of testers that I suspect do basically the same thing
5026 18:57 <@matches> Compiling with -Ofast does lead to
5027 18:57 <@matches> WARNING: main (main.cpp:29) - __STDC_IEC_559__ not defined. IEEE 754 floating point not fully supported.
5028 18:57 <@matches> (I am the best programmer I put in all the warning messages!)
5029 18:58 <@matches> And ipdf segfaults
5030 18:58 <@matches> But not the tester
5031 18:58 <@matches> In stbtt_FindGlyphIndex
5032 18:59 <@matches> Are we going to need to submit another pull request...
5034 19:17 <@matches> Adding all those "SanityCheck" calls may have slowed things down a tad
5035 19:17 <@matches> IPDF::ParanoidNumber::SanityCheck (this=0xe80a640, visited=std::set with 9775 elements = {...})
5036 19:23 <@matches> Sanity-Check-this - out
5037 19:24 <@matches> I think I need more sleep
5038 19:53 < Pommers> goto bed matches
5039 --- Day changed Sat Sep 20 2014
5040 21:22 <@matches> Overloading operator new and delete is kind of terrifying...
5041 21:25 <@matches> Things run blindingly fast
5042 21:25 <@matches> Until they segfault
5043 21:26 <@matches> I suspect std::vector does not take kindly to being destructed multiple times
5044 21:33 <@matches> Hmm
5045 21:33 <@matches> I need to make std::vector pointers instead?
5046 21:33 <@matches> That's getting a little convoluted...
5047 21:35 <@matches> Or what I really need is a delete operator that doesn't actually call the destructor automagically
5048 21:35 <@matches> (Even when you overload delete it still calls the destructor?)
5049 21:36 <@matches> I'm noticing a bit of a trend here
5050 21:36 <@matches> You can overload C++ operators but if you do it totally breaks everything
5051 21:36 <@matches> eg: You could implement a double() operator for casting to double but you're better off just writing Double() and using it everywhere because otherwise the compiler goes insane trying to use your operator in places
5052 21:37 <@matches> You can implement your own delete operator but that won't stop the compiler from doing everything it does with the standard delete...
5053 21:38 <@matches> The only operators you can overload (mostly) are the arithmetic ones
5054 21:39 <@matches> Also the compiler seems to occasionally just use the standard delete anyway which just makes everything TOTALLY INSANE
5055 21:39 <@matches> WHY
5056 21:39 * matches grumbles and prepares to write "Delete" and "New"
5057 --- Day changed Sun Sep 21 2014
5058 16:43 <@matches> I think I'm just an idiot
5059 16:47 <@matches> Arena doesn't help that much
5060 16:47 <@matches> Which is odd
5061 16:48 <@matches> If anything it is slower :(
5062 16:49 <@matches> It's hard to profile something that doesn't work in valgrind
5063 17:00 <@matches> On the other hand, removing all the "SanityCheck" calls is surprisingly effective...
5064 17:48 <@matches> Damn
5065 17:48 <@matches> I guess new and delete isn't as expensive as just doing a craptonne of branches
5066 22:33 <@matches> So, probably a better idea is to stop the CPU renderer from recalculating the object bounds every single frame
5067 22:33 <@matches> And update the object bounds when the view is changed instead
5068 22:34 <@matches> Since we don't really care what the original bounds are
5069 22:37 <@matches> This is basically what the quad tree does...
5070 22:37 <@matches> Well it changes coordinate systems as well
5071 22:37 <@matches> But I can zoom in a lot further now than I could before...
5072 22:38 <@matches> Until eventually there is a sigfpe
5074 23:19 <@matches> This is slightly worrying
5075 23:19 <@matches> I can zoom to {0.5,0.5,1e-41,1e-41} and insert things without any problem
5076 23:21 <@matches> The CPU renderer does tend to sigfpe
5077 23:27 <@matches> Man I feel really foolish now
5078 23:27 * matches makes plans to delete the IRC logs again
5080 23:33 <@matches> The good news is that GMP rationals are pretty effective when you do cumulative transforms on the object bounds as opposed to keeping the bounds static and transforming them to view coordinates
5081 23:34 <@matches> The bad news is that so are regular floats
5082 23:34 <@matches> I'm pretty sure it just means we need a better tester
5084 23:38 <@matches> We lie
5085 --- Day changed Wed Sep 24 2014
5086 12:33 <@matches> Ok, sulix do you want to help design these comparisons that we were going to use
5087 12:35 <@matches> We've got the "qualitative" comparisons pretty well...
5088 12:36 <@matches> It's the other 4 things
5089 12:36 <@matches> Performance vs number of primitives
5090 12:36 <@matches> Like
5091 12:36 <@matches> Do we add N primitives randomly?
5092 12:36 <@matches> Are our "test" documents just going to be Humphrey The Rabbit and Shady the Fox
5093 12:36 <@matches> Not that I have any objection to using them
5094 12:36 < sulix> I think so.
5095 12:36 <@matches> I'll add Emma the Cat and Mitch the Panda
5096 12:37 < sulix> Number of primitives should just be number of curves/objects.
5097 12:37 < sulix> I could do with some panda-monium. :P
5098 12:37 <@matches> Pommers beat you to that one
5099 12:37 <@matches> That's the name of the club's panda
5100 12:38 <@matches> I don't think I'll add Sam the Girraffe
5101 12:38 <@matches> It's embarrassing
5102 12:38 <@matches> Oh also did you read the wall of text?
5103 12:39 <@matches> Because I think we need a new way to demonstrate how much better the quad tree is than regular floats other than just zooming forever
5104 12:39 <@matches> Regular floats do quite well now
5105 12:39 <@matches> I should probably commit this
5106 12:39 < sulix> What about adding objects?
5107 12:39 <@matches> I also tried to get it to run without X for bus purposes
5108 12:39 <@matches> Well
5109 12:40 <@matches> Basically I threw out the idea of the view and just applied the transforms directly to all the object bounds
5110 12:40 <@matches> It will still print the equivelant view coordinates
5111 12:40 < sulix> Can you zoom out and then back in again?
5112 12:41 <@matches> I suspect if you go far enough there will be issues
5113 12:41 <@matches> The problem is going far enough takes a long time
5114 12:41 <@matches> We need to make ipdf take input so we can automate it?
5115 12:41 * sulix will do that this afternoon.
5116 12:41 <@matches> Or implement a "tester" that does it
5117 12:42 <@matches> HFP has a chapter on reducing error which boils down to rearranging the operands
5118 12:43 <@matches> Although they don't have anything as insanely dumb as ParanoidNumber
5119 12:43 < sulix> Isn't that what ParanoidNumber does?
5120 12:43 <@matches> Yeah
5121 12:43 <@matches> But ParanoidNumber doesn't assume any knowledge over what you are trying to do
5122 12:43 <@matches> Hence it is horribly inefficient
5123 12:44 < sulix> I suspect we need to sit down with Kahan and work out what the exact mathematical properties of floating-point error.
5124 12:44 <@matches> Their examples are like "If you know you are going to sum all these numbers, you sort them"
5125 12:44 <@matches> HFP goes through them
5126 12:45 <@matches> Sometimes you can represent the error as another float but with twice the precision?
5127 12:45 <@matches> It might be worth looking at whether MPFR includes the "Inexact" exception
5128 12:45 <@matches> Then you could just double the precision every time it occured?
5129 12:46 * sulix will be back shortly.
5130 12:47 <@matches> It's going to be a fun study break :S
5131 12:54 <@matches> Also PN manages to pass realops but somehow give totally whack behaviour when you actually use it
5132 12:54 <@matches> ...
5133 12:54 <@matches> And this is enforcing a size limit of "0" (which means it should just be equivelant to doubles!)
5134 12:55 <@matches> Also I should probably also write a draft thesis
5135 12:55 <@matches> Also other units suck, why can't the project count as 4 units
5136 12:56 <@matches> I'm finally starting to understand this properly, as we are running out of time at an alarming rate
5137 12:56 <@matches> I guess there probably won't be any experts on floating point arithmetic at my conference so I can probably pretend I know things
5138 13:02 <@matches> http://szmoore.net/ipdf/i-think-there-might-be-a-slight-maths-error-rabbit.png
5139 13:03 <@matches> I strongly suspect Copy/Paste errors
5140 14:07 < Pommers> Yeah. We have a club Panda
5141 14:10 -!- matches changed the topic of #ipdf to: IPDF: Infinite Panda Document Format
5143 14:13 <@matches> cat.svg causes a segfault
5144 14:23 <@matches> It has an empty path?
5145 14:23 <@matches> I regret somewhat making the path be from m_start to m_end *inclusive*
5146 14:37 <@matches> Have a git push
5147 14:37 <@matches> Possibly with a correct timestamp
5148 14:47 <@matches> Ok, things still go whack when you apply transforms to objects instead of the view
5149 14:48 <@matches> But because you can add objects in screen space rather than converting to view space and back, you don't get the precision issues adding a new object
5150 14:48 <@matches> I suspect
5151 14:48 <@matches> If you add a new object at say 1e-6
5152 14:48 <@matches> It looks fine
5153 14:49 <@matches> Then you zoom back out to 0.5,0.5,1,1 or something
5154 14:49 <@matches> Then zoom back in
5155 14:49 <@matches> It *should* look all wierd again
5156 14:49 <@matches> Then again it might not
5157 14:49 <@matches> Because floating point errors are crazy
5158 14:51 <@matches> It does!
5159 14:51 <@matches> YES
5160 14:52 <@matches> It looks totally screwed up in a different way
5161 14:52 <@matches> So, I guess this just makes it more important that you can actually add elements to the quad tree
5162 14:53 * sulix feels the pressure mounting.
5163 14:54 <@matches> Is the quad tree kind of equivelant to applying the transformations to the objects but making sure you don't lose the original bounds
5164 14:54 <@matches> When you zoom back out you get the original bounds back
5165 14:54 < sulix> Yeah, largely.
5166 14:54 <@matches> Cool
5167 14:54 <@matches> I'm sure someone trained in maths can say smart stuff about that
5168 14:55 <@matches> Applying transformations to the object bounds is probably the best way if you have an arbitrary precision number representation
5169 14:55 <@matches> You can use a lower precision representation when you work in screen space then
5170 14:55 <@matches> Although I'm not sure how I would actually implement that...
5171 14:56 <@matches> With some kind of
5172 14:56 <@matches> Quad tree...
5173 14:56 <@matches> maybe
5174 14:56 <@matches> I guess if two methods end up with the same pixels in the same spot then they are mathematically equivelant
5175 14:58 <@matches> You could also speed things up
5176 14:58 <@matches> If you exploited the Path bounds
5177 14:58 <@matches> Express all the beziers within a path relative to {0,0,1,1} then just apply your transformation to the path
5178 14:58 < sulix> This is a much better design.
5179 14:59 <@matches> Well that's sort of a compromise between transforming the view or transforming all the bounds
5180 14:59 < sulix> But it's sort-of too late to change it all now.
5181 14:59 <@matches> I think I might have to though
5182 14:59 <@matches> If I want to make GMPrat performant
5183 15:00 <@matches> Lets face it, I'm doing a better job at redesigning the document structure to be less terrible than I am at actually implementing arbitrary precision number types :S
5184 15:01 <@matches> I could probably redesign the CPU renderer, but the GPU renderer would be harder
5185 15:01 <@matches> PathRenderer already does everything BezierRenderer does, but it adds shading
5186 15:02 <@matches> I think right now I will try to stop PN from being quite so broken
5187 15:02 <@matches> I am perplexed
5188 15:03 <@matches> As there is basically an "assert(equivelant to double)" after every single operation
5189 15:03 <@matches> None of them fail
5190 15:03 <@matches> And it gives a totally wrong result
5191 15:45 <@matches> It is rendering at 666 FPS
5194 16:49 <@matches> There should be postive paranoid numbers soon
5195 16:49 <@matches> s/postive/negative
5196 16:49 <@matches> That thing that happens when you put a '-' sign in front of a number
5197 16:49 <@matches> Very important
5198 16:50 < sulix> Negative _and_ paranoid? I'm waiting for the "Number Theropy" classes next... :P
5199 16:50 <@matches> I've learned some valuable lessons about implementing number classes I guess
5200 16:50 < sulix> At least it isn't complex?
5201 16:50 <@matches> Make sure you test every single function against doubles
5202 16:51 <@matches> (Including constructors)
5203 16:51 <@matches> I'm really over bug fixing this class
5204 16:51 <@matches> I just want to make it faster dammit
5205 16:51 <@matches> I have ideas about how to make it faster
5206 16:53 <@matches> Admittedly they are mostly not good ideas
5207 16:53 <@matches> I'm expecting the arena may help more when I don't enforce a size limit of zero though
5208 16:54 <@matches> YES
5209 16:54 <@matches> I can render a paranoid rabbit
5210 16:54 <@matches> Oh wait
5211 16:54 <@matches> I didn't increase the size limit from zero
5212 16:54 <@matches> That will be the real test...
5213 17:02 <@matches> So are you going to implement a magical automated zoomer? :P
5214 17:03 <@matches> We kind of need to be able to choose what point to zoom in on depending on the particular test svg though :S
5215 17:04 <@matches> Right now it's hard to tell whether I am getting precision issues or I just accidentally moved the mouse a bit
5216 21:02 < sulix> That's about the most terrifying email it's possible to get, I think.
5217 21:59 < sulix> Very simple scripts for automating actions now exist.
5218 --- Day changed Thu Sep 25 2014
5219 08:27 < sulix> matches: Thoughts on a meeting time?
5221 10:58 <@matches> I sort of slept in
5222 10:59 <@matches> Well actually I didn't sleep in I just started practicing the piano and then it was 2 hours later...
5223 10:59 <@matches> Yeah that's a pretty terrifying email
5224 11:05 <@matches> I am excited by the script
5225 11:06 < sulix> Cool. I just added support for toggling some things to it.
5226 11:06 < sulix> When's best for you meeting-wise?
5227 11:06 <@matches> I can't make 11:15
5228 11:06 <@matches> Was there any other time that wasn't 5:45pm?
5229 11:07 <@matches> It'll be a struggle to make 9am
5230 11:07 <@matches> In fact apparently it's 7pm now
5231 11:07 <@matches> I must have really overslept
5232 11:07 <@matches> I'm sure I made progress this week
5233 11:07 <@matches> I just can't remember what it was
5234 11:07 < sulix> Well, I can make any time after ~12:00, now.
5235 11:08 < sulix> You did the whole cumulative transforms thing.
5236 11:08 <@matches> Yes, I'm not sure whether that was good or not
5237 11:08 <@matches> It sort of invalidates a whole bunch of earlier things
5238 11:09 <@matches> It is probably the only way that arbitrary precision bounds will ever be practical though
5239 11:09 <@matches> I also fixed ParanoidNumbers a bit I guess
5240 11:09 <@matches> But they are still terrible
5241 11:09 <@matches> I doubt it will get to the point where they ever aren't terrible
5242 11:12 <@matches> Do I do my GENG4402 assignment that's due on Friday or do I sacrifice it for the glory of IPDF?
5243 11:13 < sulix> I was going to say "IPDF is not a cult", but I'm not so sure.
5244 11:16 <@matches> On the one hand I feel sorry for you because you have the entire trip to Sydney as a meeting time
5245 11:16 <@matches> On the other hand I feel sorry for me because I haven't actually achieved anything
5246 11:17 <@matches> I have an abstract that says what I'm going to achieve...
5247 11:17 <@matches> It hasn't been achieved
5248 11:17 <@matches> Achievements are locked
5249 11:18 * sulix hasn't achieved anything but the scripting stuff in the last week. :/
5250 11:19 <@matches> You have a quad tree!
5251 11:19 <@matches> Did I tell you, GMPrat takes 5 seconds to pass tests/addsub and doubles take 5ms
5252 11:20 <@matches> Wait GMPrat was only 500ms
5253 11:20 <@matches> PN was 6 seconds
5254 11:20 <@matches> When restricted to size 1
5255 11:20 <@matches> There's a bug somewhere
5256 11:22 <@matches> They seem to grow a lot bigger than the size limit
5257 11:22 <@matches> Ah, probably because I don't check "m_size + b->m_size"
5258 11:27 <@matches> -_-
5259 11:27 <@matches> You need to stop adding Debug messages that assume Real is a double
5261 11:28 <@matches> Some of us are trying to compile with other number representations here
5262 11:29 <@matches> I guess it does seem to actually work even though it's horrendously slow
5263 11:29 <@matches> That's an achievement of some sort
5264 11:30 <@matches> GMPrat also works and is horrendously slow but less horrendously slow, but I didn't actually invent that representation
5265 13:45 <@matches> The script has loadsvg now
5266 13:45 <@matches> It's tempting to make it turing complete... like some kind of... Post... Script
5267 13:45 <@matches> But that would be silly
5268 13:50 <@matches> I am loving this script thing
5269 13:50 <@matches> Think of all the calories I can save
5270 13:51 <@matches> My goodness I can even eat food whilst watching it zoom
5271 13:51 <@matches> My hand is freed
5272 13:51 <@matches> Freed from the tyranny of endless zooming
5273 13:56 < sulix> Cool: I was going to add loadsvg, but it still has some quadtree problems.
5274 13:58 <@matches> I think I will add "label" and "goto" so that I can zoom into the rabbit's eye *forever*
5275 13:58 <@matches> Well
5276 13:58 <@matches> Hmm that's actually work
5277 13:58 <@matches> But I still think it's necessary
5278 13:59 <@matches> You need to see at which point the Quad tree breaks
5279 13:59 <@matches> Note that we don't need it for anything I've done because everything I've done breaks before single precision floats
5281 14:10 <@matches> I have just implemented goto
5282 14:10 <@matches> However there is still time to rename it to "goat"
5283 14:11 <@matches> Once you have a goto you really need a branch...
5284 14:15 <@matches> Go GMPrat
5285 14:15 <@matches> A mighty 3 FPS
5286 14:16 <@matches> The downside of my hands being free is I can type more things into IRC
5287 14:17 <@matches> You can do it GMPRat
5288 14:17 <@matches> Noo
5289 14:17 <@matches> 2.5FPS
5290 14:17 <@matches> We're still going strong
5291 14:18 <@matches> Oh that's the total accumulated average
5292 14:18 <@matches> It's actually at 0.8 FPS now
5293 14:18 <@matches> And falling rather dramatically
5294 14:20 <@matches> The fundamental flaw with these arbitrary precision types is the representation never gets smaller
5295 14:21 <@matches> GMP rationals would calculate the GCD which could get smaller (when you zoom back out for example)
5296 14:21 <@matches> But the GMP integer representation will still only ever increase in size
5297 14:21 <@matches> I think they just always use the largest precision
5298 14:22 <@matches> I was hoping that ParanoidNumber would do better since it actually has the ability to shrink
5299 14:22 <@matches> But it's not amazing
5300 14:22 <@matches> Hmm, I seem to have lost the rabbit
5301 14:22 <@matches> That's not meant to happen
5302 14:22 <@matches> GMPrat is supposed to be exact
5303 14:22 <@matches> Where did he go!
5304 14:34 <@matches> Every frame is agonising
5305 14:39 <@matches> Maybe it's time to switch to motsugo for CPU rendering tests
5306 14:40 < sulix> At least we now have a good way of burning up time in meetings?
5307 14:44 <@matches> I'm somewhat confused
5308 14:44 <@matches> With the cumulative transforms, the GPU renderer should work
5309 14:44 <@matches> (It is equivelant to using CPU for transformations)
5310 14:44 <@matches> But it seems to still break around 1e-6
5311 14:44 < sulix> Yes...
5313 14:45 <@matches> Right I commented out the TRANSFORM_OBJECTS_NOT_VIEW line
5314 14:45 * matches tries again...
5315 14:45 * sulix is really regretting the 4-switch statement, 16-case recursive function in the quadtree at the moment.
5316 14:46 <@matches> I'll make a deal, I won't look at quadtree.cpp if you don't look in paranoidnumber.cpp
5317 14:47 < sulix> Fair enough.
5318 14:49 <@matches> I'm not seeing anything in the GPU renderer past 1e-7
5319 14:49 <@matches> With GMPrat
5320 14:50 <@matches> I think I'm going to have to go through and fix the more obviously terribly designed parts of things
5321 14:50 <@matches> For GMPrat to have any chance
5322 14:53 <@matches> I'm wishing I had a quad tree so I didn't have to transform the bounds of things that would end up outside the screen space :P
5323 14:53 <@matches> Maybe there's hope for an unholy union of Quadtree and GMP rationals yet
5324 14:56 < sulix> "And the fifth angel sounded, and I saw a star fallen from heaven to earth: and to him was given the key to a bottomless abyss."
5325 15:00 <@matches> I think the GMP rational -> float conversion is fith?
5326 15:00 <@matches> It doesn't make any sense :S
5327 15:00 <@matches> An object that is visible should always have bounds somewhere in {0,0,1,1}
5328 15:00 <@matches> Which should be totally fine
5330 15:01 <@matches> Is this the bit where I made GMPrat::ToDouble return 0 if it was too big...
5331 15:02 <@matches> No it returns mpq_get_d
5332 15:02 <@matches> Blergh
5333 15:03 < sulix> An object which is visible should have bounds that _intersect_ {0,0,1,1}
5334 15:03 < sulix> Perhaps not contained within {0,0,1,1}
5335 15:03 <@matches> Yeah that
5336 15:04 <@matches> Maths
5337 15:04 <@matches> 0 <= x <= 1
5338 15:04 <@matches> No not that
5339 15:04 <@matches> Thing
5340 15:04 < sulix> Unless you're using the Quadtree, which does clip everything to {0,0,1,1}
5341 15:04 <@matches> Anyway
5342 15:04 <@matches> It should work
5343 15:05 <@matches> Wait
5344 15:05 <@matches> There is this blen thing in the shader
5345 15:05 <@matches> Is that passed via the view width or something
5346 15:05 <@matches> Hmm
5347 15:05 < sulix> That's kinda FITH.
5348 15:07 <@matches> Anyway, somehow there is a multiplication by width/height and then probably a division afterwards somewhere else
5349 15:07 <@matches> I'm rather confused
5350 15:09 * matches struggles to navigate the treacherous waters of view.cpp
5351 15:10 <@matches> That's right, RenderRange does the actual things
5352 15:10 <@matches> We've got PrepareRender -> Render -> RenderRange/QuadtreeNode
5353 15:11 < sulix> RenderQuadtreeNode just calls RenderRange a bunch of times and then breaks the document.
5354 15:13 <@matches> It's alright, we can save to bitmap, we don't need the document structure to not break
5355 15:13 <@matches> So how the hell is the shader getting rounding errors
5356 15:13 <@matches> It should always get glbounds of {0,0,1,1}
5357 15:13 <@matches> (With cumulative transforms)
5358 15:15 <@matches> Then RenderRange -> ObjectRenderer::RenderUsingGPU...
5359 15:17 <@matches> This does not make sense
5360 15:17 <@matches> Things cannot be broken in this way such that they work
5361 15:18 <@matches> Hello
5362 15:18 <@matches> GraphicsBuffer::BufferUsageStaticDraw
5363 15:18 <@matches> Does that do a thing?
5364 15:19 <@matches> Also
5365 15:19 <@matches> Should //m_objbounds_vbo.Invalidate(); be commented out
5366 15:19 <@matches> I'm still confused as to how it works at all
5367 15:20 < sulix> BufferUsageStaticDraw tells the driver that we're not planning to modify this buffer frequently.
5368 15:21 < sulix> The Invalidate() stuff should be commented out, IIRC.
5369 15:23 <@matches> Hmm when in doubt make clean
5370 15:30 <@matches> Anyway
5371 15:30 <@matches> I hope you get the quadtree working with adding stuff
5372 15:30 <@matches> Because the only point at which doubles fail is when they get a sigfpe
5373 15:31 <@matches> Floats do start to do strange things before that
5374 15:32 <@matches> Actually no, doubles would screw up it's just the script is not testing in that range
5375 15:33 < sulix> I'm still trying to get panning working with the quadtree at the moment.
5376 15:33 < sulix> I was scratching my head for quite a while before I realised TRANSFORM_OBJECTS_NOT_VIEW was on.
5377 15:34 <@matches> Sorry
5378 15:34 <@matches> It really helps the non quad tree rendering
5379 15:34 < sulix> (You'd be surprised how totally it breaks the quadtree)
5380 15:34 <@matches> Oh I'm not that surprised
5381 15:34 <@matches> The quad tree is basically doing cumulative transforms after all
5382 15:34 <@matches> In a very loose sense
5383 15:35 < sulix> The quadtree does both.
5384 15:35 < sulix> It's an equal-opportunity data structure.
5385 15:37 <@matches> The more I think about this, the more I am convinced the quadtree is the only sensible solution
5386 15:37 <@matches> Which is unfortunate
5387 15:37 <@matches> I had hoped to actually have something to present
5388 15:37 <@matches> Other than "And you should all attend my partner's presentation next week!"
5389 15:38 <@matches> I guess I should produce some performance graphs or something
5390 15:38 <@matches> What I did was a terrible idea and this is why
5391 15:39 <@matches> Sort of thing
5392 15:40 <@matches> I wonder where the bottleneck is though
5393 15:40 <@matches> GMPrats get very slow even with the GPU renderer
5394 15:40 < sulix> Converting to doubles?
5395 15:40 < sulix> Doing gcd() all day?
5396 15:40 <@matches> Maybe
5397 15:40 <@matches> gcd is 50% of execution time according to perf
5398 15:41 <@matches> perf is pretty cool but it is not as easy to use as valgrind
5399 15:41 <@matches> Unfortunately valgrind doesn't fully emulate the FPU
5400 15:41 <@matches> So ParanoidNumbers don't work at all using it
5401 15:42 <@matches> Which is very frustrating because there have to be things I can speed up in them
5402 15:42 <@matches> I have a git stash where I tried to implement Copy-On-Write but it was a bit of a disaster
5403 15:42 <@matches> It was so much of a disaster that even with COW disabled they still broke (despite working fine in the previous commit...)
5404 15:43 <@matches> They do spend an excessive amount of time copy constructing each other though
5405 15:50 <@matches> If I can express bezier bounds relative to the containing path it might speed things up slightly
5406 15:51 < sulix> Ideally we'd ditch the "BEZIER" object and just have paths.
5407 15:51 < sulix> But that sounds like a lot of work.
5408 15:53 <@matches> I'd be less inclined to want to do it if things weren't so amazingly slow
5409 16:00 <@matches> It looks like Bezier::GetType is slowing the CPU rendering down at least
5410 16:01 <@matches> Clearing pixels is also slow
5411 16:01 <@matches> Although that doesn't depend on the zooming
5412 16:02 <@matches> GPU rendering isn't much better though, hmm
5413 16:46 <@matches> The Arena sort of helps things
5414 16:47 <@matches> Until it gets to allocating 40 million x 180 Bytes ...
5415 17:25 <@matches> Protip, #define and spelling mistakes and #ifdef do not work together well...
5416 --- Day changed Fri Sep 26 2014
5417 01:40 <@matches> I came up with a compromise between not doing GENG4402 to work on IPDF and doing GENG4402
5418 01:40 <@matches> The only casualty was sleep
5419 01:41 <@matches> And no one likes him anyway
5420 01:41 <@matches> All sides were pleased by the deal
5421 16:21 * matches regrets upgrading debian jessie after the shellshock thing
5422 16:21 <@matches> fglrx is gone again
5423 --- Day changed Sat Sep 27 2014
5424 12:28 <@matches> 5 minutes to go in ACM...
5425 12:29 <@matches> You can do it Dijkstras on Bijkstras
5426 12:30 <@matches> Oh 5 minutes until they freeze the score board
5427 12:30 < Pommers> 5mins? Timer says an hour
5428 12:30 <@matches> Apparently they freeze the score board
5429 12:30 <@matches> I don't know how these work
5430 12:30 <@matches> I only just started watching
5431 12:31 <@matches> An hour is plenty of time
5432 12:31 < Pommers> So teams can't see each others scores change?
5433 12:31 <@matches> Yeah
5434 12:31 <@matches> Mickelback just got C wrong
5435 12:31 < Pommers> They secretlyu have them all solved and are just waiting to waste time
5436 12:31 <@matches> Yes
5437 12:31 <@matches> Nice camera work...
5438 12:31 < Pommers> Getting it right won't get them into first though
5439 12:32 <@matches> I can't see them
5440 12:33 < Pommers> UWA needs labs like that
5441 12:33 <@matches> It's still anyone's game!
5442 12:33 <@matches> They said it
5443 12:33 <@matches> You can do it Dijkstras on Bijkstras
5444 12:33 < Pommers> Cmon Dijksras in Kijkstras
5445 12:34 * Pommers fails
5446 12:34 <@matches> Go for G
5447 12:35 < Pommers> Yeah. Be the winner and Get G right now
5448 12:35 <@matches> Also D
5449 12:35 <@matches> and E
5450 12:35 < Pommers> B D E G
5451 12:35 <@matches> All you need to do is get all the hard ones right without anyone else getting them right...
5452 12:35 < Pommers> Nah
5453 12:36 < Pommers> You want to get all the easy ones as soon as possible
5454 12:36 < Pommers> Then work out the harder ones
5455 12:36 <@matches> It's a bit late now..
5456 12:36 <@matches> Sink everything into the hard ones!
5457 12:37 < Pommers> The teams still get feedback in the last hour right?
5458 12:37 <@matches> I assume so
5459 12:37 <@matches> This camera is totally biased
5460 12:37 <@matches> I haven't seen sulix yet
5461 12:37 <@matches> Just shots of people eating sandwiches
5462 12:38 < Pommers> Sandwich eating stream
5463 12:38 < Pommers> Best Stream
5464 12:39 < Pommers> I saw Rowan before
5465 12:39 <@matches> ANU is doing well
5466 12:39 <@matches> I think this calls for some sabotage
5467 12:40 < Pommers> Where's axtens when we need him
5468 12:40 <@matches> In Canberra
5469 12:40 <@matches> Also he'd probably be supporting ANU
5470 12:41 < Pommers> Probably
5471 12:42 <@matches> I'm waiting for it to all come together
5472 12:42 <@matches> And all the boxes to turn blue
5473 12:42 <@matches> Commentator is a bit behind
5474 12:42 < Pommers> So they've submitted but no public feedback
5475 12:42 <@matches> Or the stream is a bit behind I guess
5476 12:42 < Pommers> 45 seconds
5477 12:42 <@matches> ANU made that submission 3 minutes ago
5478 12:43 <@matches> Singing!
5479 12:43 <@matches> This stream has it all
5480 12:43 < Pommers> I have the stream muted
5481 12:43 <@matches> Ooh a shot of the other room
5482 12:44 <@matches> sulix: Stand up!
5483 12:44 < Pommers> Move out the way so we can see into the room
5484 12:44 < Pommers> up and down the bolevard
5485 12:44 <@matches> Hahaha
5486 12:45 <@matches> That *could* be sulix in the background
5487 12:47 <@matches> I guess it would be intrusive to have the camera in the room with the contestants
5488 12:47 <@matches> Also for 14 viewers
5489 12:48 <@matches> I love the sign off
5490 12:51 <@matches> Gozz!
5491 12:51 <@matches> Gozz what are you doing!
5492 12:51 <@matches> Get back to the desk!
5493 12:54 <@matches> These coaches(?) look rather worried
5494 12:55 <@matches> Oh well, I guess I should actually work on IPDF
5495 12:55 <@matches> Not all of us get free trips to Sydney :P
5496 13:00 < Pommers> So yeah. They get there own score line
5497 13:01 < Pommers> I missed Gozz!
5498 13:03 <@matches> They must be doing well if they have time for sandwich breaks
5499 13:03 <@matches> Have all the solutions ready to submit in the last 5 minutes
5500 13:03 <@matches> Shock victory
5501 13:03 < Pommers> Or they have given up
5502 13:03 <@matches> No!
5503 13:03 <@matches> How do you read the scores anyway?
5504 13:03 <@matches> The first column is number of problems solved
5505 13:03 <@matches> What's the second
5506 13:04 < Pommers> I explained in #ucc before
5507 13:04 <@matches> Ah damn
5508 13:04 < Pommers> 12:09 <@Pommers> Ranking is number of challanges sorted: More is better
5509 13:04 < Pommers> 12:09 <@Pommers> Numeric number: Less is better
5510 13:04 < Pommers> 12:09 <@Pommers> Cause the number is minute challenmge was solved + 10 * number of failed attempts
5511 13:04 < Pommers> 12:10 <@Pommers> 20, not 10
5512 13:04 <@matches> I thought they were doing well because they had a high number
5513 13:04 < Pommers> 12:10 < bob> so those teams who haven't finished anything and have a score of zero are winning?
5514 13:04 < Pommers> 12:11 < bob> oh I get you now
5515 13:04 < Pommers> 12:11 <@Pommers> Nope. Because number of challenges sorted is more important
5516 13:04 < Pommers> 12:11 < bob> really shittily labeling on that table
5517 13:04 < Pommers> 12:11 < bob> so mickelback is coming second
5518 13:04 <@matches> Yes yes I can change windows
5519 13:04 < Pommers> 12:11 <@Pommers> The numbers on the left show their current position
5520 13:04 < Pommers> 12:11 <@Pommers> Yep
5521 13:04 < Pommers> 12:12 <@Pommers> The scoring system is pretty much the same for all coding competitions
5522 13:04 < Pommers> 12:12 <@Pommers> GCJ is different though
5523 13:04 < Pommers> 12:16 <@Pommers> Numbers down the bottom are attempts/correct/time was first solved
5524 13:04 < Pommers> Nope
5525 13:05 < Pommers> But now it's in your logs too!
5526 13:05 <@matches> Meh anything that skews the statistics away from me talking to myself is fine by me
5527 13:05 <@matches> Just copy paste all of #ucc in here please
5528 13:06 <@matches> Mikelback could still win
5529 13:06 < Pommers> 13:05 <@matches> Just copy paste all of #ucc in here please
5530 13:06 < Pommers> If you say so
5531 13:06 <@matches> Nonononono
5533 13:07 <@matches> ll just delete the git repository when this is over
5534 13:08 < Pommers> Sounds like a good idea
5535 13:13 <@matches> So you can go back to earlier parts of the stream
5536 13:13 <@matches> "I tell you, in this life there are only two things to die for: Your country, and these sandwiches"
5537 13:15 < Pommers> Yep
5538 13:15 <@matches> Back to live action, and there is a close up of someone eating a sandwich
5539 13:27 <@matches> There's still time!
5540 13:28 <@matches> Why are things not turning blue
5541 13:29 <@matches> Mickelback appears to be struggling on B
5542 13:29 <@matches> Haha
5543 13:29 <@matches> "It could be anyone... anyone... well probably not Auckland Queensland or Adelaide"
5544 13:29 <@matches> Anyone of the top five
5545 13:31 <@matches> I'm still waiting for the 5 simultaneous submissions from Dijkstras on Bijkstras
5546 13:32 <@matches> More sandwiches
5547 13:34 < Pommers> Now we wait
5548 13:36 <@matches> Reading these logs back it sounds a lot wierder when you don't notice the timestamps
5549 23:23 <@matches> So, after much spaghetti bolognaise
5550 23:24 <@matches> I managed to perform the arbitrary precision transformations only on the bounds of paths as opposed to the bounds of every single object.
5551 23:24 <@matches> It helps a lot but at the end of the day it's still fantastically useless.
5552 23:24 <@matches> Project complete I guess.
5553 23:24 <@matches> Just Use Quad Trees
5554 23:25 <@matches> If I had some way to clip the objects to the view so I only had to transform things that would be shown...
5555 23:25 <@matches> I would have a quad tree
5556 23:25 <@matches> If I don't do that, then you still have the exponentially growing Gmprat
5558 23:26 <@matches> Oooh
5559 23:26 <@matches> Actually
5560 23:26 <@matches> I had the CPU renderer on
5561 23:27 <@matches> The GPU renderer loves this
5562 23:27 <@matches> The CPU renderer seems to generally just suck
5563 23:27 <@matches> I suspect there is actually a bug in it
5564 23:27 <@matches> It shouldn't be so slow
5565 23:28 * matches is getting really excited
5566 23:28 <@matches> I actually zoomed in to 1e-9 using floats for everything except the Path bounds
5567 23:29 <@matches> It would be nice the CPU renderer didn't suck, but I guess "Transform the bounds with arbitrary precision on the CPU and do rendering with floats on the GPU" is actually the better approach
5568 23:32 <@matches> Also... I think somehow I got an updated fglrx that is less shit at rounding floats
5569 23:32 <@matches> Because eye_of_the_rabbit doesn't go all funny anymore...
5570 23:32 <@matches> Unless you make it zoom a little bit more
5571 23:36 <@matches> I still think the quad tree will beat this in terms of performances
5572 23:36 <@matches> But at least it's like, in the game
5573 23:36 <@matches> As in it doesn't take so long to do the performance test that it's not worth bothering actually doing a comparison
5574 23:36 <@matches> Now we can have a graph at least
5575 23:46 < Pommers> Very nice
5576 23:51 <@matches> The CPU renderer is stalling in DeCasteljau
5577 --- Day changed Sun Sep 28 2014
5578 00:18 <@matches> sulix: You need to get eye_of_the_rabbit working under the quadtree so I can compare it with my horrible hacks to View,Path,BezierRenderer,PathRenderer...
5579 00:18 <@matches> Document...
5580 00:18 <@matches> Not a single class escaped the wrath of the #ifdef
5581 00:18 <@matches> No pressure, glhfdd
5582 00:19 <@matches> Unfortunately I still get a bunch of SIGFPEs but I think those are in places that don't affect the rendering, so I can probably fix them
5583 00:19 <@matches> Or set the program to ignore them
5584 00:20 <@matches> I never really worked out how you turned them on in the first place :S
5585 00:20 <@matches> Also well done at ACM and all that
5586 00:21 <@matches> The #define's you'll want to turn off to get the quadtree working again are both the ones in view.h and path.h
5587 00:21 <@matches> Oh wait I put in a check just for you
5592 --- Log opened Sun Sep 28 09:58:47 2014
5594 09:58 -!- Irssi: #ipdf: Total of 1 nicks [1 ops, 0 halfops, 0 voices, 0 normal]
5595 09:58 -!- Irssi: Join to #ipdf was synced in 0 secs
5596 10:04 -!- matches changed the topic of #ipdf to: Motsugo OOMed, this would be a great excuse to extend the project if we were actually using Motsugo
5597 --- Log closed Sun Sep 28 10:08:08 2014
5598 --- Log opened Sun Sep 28 10:08:20 2014
5600 10:08 -!- Irssi: #ipdf: Total of 1 nicks [1 ops, 0 halfops, 0 voices, 0 normal]
5601 10:08 -!- Irssi: Join to #ipdf was synced in 2 secs
5604 --- Day changed Wed Oct 01 2014
5605 15:13 <@matches> Urgh the code is disgusting
5606 15:13 <@matches> I can't look at it
5607 15:16 <@matches> The quadtree does some... interesting things
5608 15:17 < sulix> I think by "interesting," you mean "horrible."
5609 15:18 < sulix> ... or possibly "wrong"
5610 15:22 <@matches> Well, you appear to be able to add things to the document
5611 15:22 <@matches> They are just in the wrong spot
5612 15:22 <@matches> Also things seem to disappear as well
5613 15:28 <@matches> Also I can't seem to compile without the quadtree anymore
5614 15:29 <@matches> You're going to have to deal with the Makefile because I'
5615 15:29 <@matches> m moving quadtree.o out of the defaults
5616 15:29 <@matches> It expects things in IPDF::Document that won't exist if the quadtree is disabled
5617 15:35 < sulix> Oops...
5618 15:35 < sulix> I have started using the makefile, though, so that's good.
5619 15:39 <@matches> I'm somewhat regretting the use of Real in the Bezier
5620 15:39 <@matches> I need another Real
5621 15:39 <@matches> Real2 ?
5622 15:39 <@matches> Or I can do terrible hacks with #undef
5623 15:39 <@matches> Hmmm
5624 15:39 <@matches> Terrible hacks with #undef it is!
5625 15:40 <@matches> Possibly we need a seperate header for all these defines
5626 15:40 <@matches> I have #defines in path.h that only work if there is also a #define in view.h ...
5627 15:41 <@matches> But the #define in view.h will still work without the one in path.h
5628 15:43 <@matches> Possibly using different branches would have been cleaner
5629 15:43 <@matches> Possibly I should really stop writing code and start writing a report anyway
5630 15:43 <@matches> So many regrets
5631 15:44 <@matches> I explained the project to some Engineers today!
5632 15:44 <@matches> They understood it!
5633 15:44 <@matches> They questioned the point of it all...
5634 15:44 <@matches> I said "The point is that you can zoom into this point here..."
5635 15:44 <@matches> "And it turns into a rabbit!"
5637 15:46 <@matches> Also "Why isn't there any shading"
5638 19:24 <@matches> I'm having a little trouble with the performance metrics
5639 19:24 <@matches> How do I pick something that is a legitimate performance metric
5640 19:24 <@matches> "We decided to render this Fox and this Rabbit"
5641 19:25 <@matches> Also I really wish ipdf were good enough to do the report in it, because then it could demonstrate itself
5642 19:25 <@matches> Now I have to link to a video?
5643 19:26 <@matches> I mean
5644 19:26 <@matches> "Figure 1 is the rabbit and Figure 2 is the rabbit that looks exactly the same when you zoom in!"
5645 19:26 <@matches> Is not really impressive
5646 19:26 <@matches> "Figure 1 is the rabbit and Figure 2 is the rabbit that looks terrible when you use floats and you'll just have to trust us when we say we fixed this"
5648 19:28 <@matches> Also causing difficulty...
5649 19:28 <@matches> The zoom level I am at is not representable as a double
5650 19:28 <@matches> Which means it cannot be plotted in gnuplot unless I scale it to the range of doubles first...
5651 19:28 <@matches> Which sort of defeats the point...
5652 19:29 <@matches> I guess I could plot "Exponent"
5653 19:29 <@matches> If you clear the document after every 3 rabbits, eye_of_the_rabbit appears to go for a very long time
5654 19:30 <@matches> So that's the performance vs zoom I guess
5655 19:30 <@matches> But only take a data point after every 3 rabbits otherwise you are also altering number of primitives
5656 19:30 <@matches> Panning is pretty much the same
5657 19:33 <@matches> I'm just going to sit and watch it zoom for a while I guess
5658 19:33 <@matches> Mesmerising
5659 19:34 <@matches> Should I help fix the quad tree?
5660 19:34 <@matches> I'd like to be able to compare it to the "actually we'll only use arbitrary precision for *some* things" approach
5661 --- Log opened Thu Oct 02 13:55:13 2014
5663 13:55 -!- Irssi: #ipdf: Total of 1 nicks [1 ops, 0 halfops, 0 voices, 0 normal]
5664 13:55 -!- Irssi: Join to #ipdf was synced in 3 secs
5666 14:44 <@matches> Hello
5667 14:44 <@matches> Are we having a meeting today?
5668 14:45 < sulix> Rowan is still in Sydney.
5670 14:45 <@matches> That would make it difficult
5671 14:45 <@matches> Well, I'
5672 14:45 < sulix> (Also every time I touch the quadtree it ends up more broken)
5673 14:45 <@matches> m in UCC trying to actually get some sort of performance graphs
5675 14:46 <@matches> Yeah, it turns out Gmprat may beat floats considerably but it isn't that great compared to doubles
5676 14:46 <@matches> It's interesting actually
5677 14:46 <@matches> It only really starts to slow down right at the limit of double precision
5678 14:46 <@matches> Then again, doubles will screw up before that (I hope)
5679 14:47 <@matches> You can represent 1e-38 with a float but it still screws up at 1e-6 after all
5680 14:47 <@matches> Also no one understands the eye_of_the_rabbit demo
5681 14:48 <@matches> I don't know why
5682 14:48 <@matches> It's pretty obvious it's *zooming to infinity*
5683 14:48 <@matches> (TM)
5684 14:48 <@matches> Righht?
5686 14:48 < sulix> Their puny minds cannot contain the majesty that is eye_of_the_rabbit.script
5687 14:50 <@matches> A demo that actually added text where the text was the scale might be cool
5688 14:50 <@matches> Except debugscript is not that advanced
5689 14:50 < sulix> If you want to add string formatting to debugscript, I will not be involved. :P
5690 14:51 <@matches> I would probably just hack on a command
5691 14:51 <@matches> "Do this one particular demo"
5692 14:51 <@matches> Instead of making the scripting language turing complete...
5693 14:53 <@matches> In other news, Gozz taught me the difference between %d and %i :P
5694 14:53 < sulix> My greatest shame...
5695 14:54 <@matches> I normally use %d but I never actually realised they were differenc
5696 14:54 <@matches> Well they aren't when printf'ing
5697 14:55 < sulix> Four hours (+ an 80 minute penalty) because of random octal...
5698 14:55 <@matches> Well, if you'd won you wouldn't be able to finish IPDF
5699 14:57 < sulix> (Yeah, but maybe I'd've been good enough to not make the QuadTree horribly buggy...)
5700 14:58 <@matches> I should look at it
5701 14:58 <@matches> When I tried it, it was adding objects but they were appearing in totally whack places
5702 14:58 <@matches> Also different places depending on what renderer was used
5703 14:59 <@matches> I'm pretty much in favour of just focussing on the GPU renderer now though
5704 14:59 < sulix> At the moment it should only add things to the root node.
5705 14:59 <@matches> I have encountered the issues you mentioned where the CPU renderer suddenly decides to crash X
5706 15:00 < sulix> It seems to just suddenly start eating all of the ram in the world.
5707 15:00 <@matches> I don't know why though
5708 15:00 <@matches> It does it with regular floats as well
5709 15:01 <@matches> Unless it's allocating the pixel buffer every frame but why the hell would I have done that
5710 15:06 <@matches> Also! Really annoying! GMP can only print rationals as num/den
5711 15:06 < sulix> I guess that's because not all rationals have a finite decimal expansion.
5712 15:06 <@matches> It has Rational -> Double but not Rational -> Arbitrary Precision Double
5713 15:07 <@matches> Yeah, but I wish you could go "Print in scientific notation to 4 sig fig"
5714 15:08 <@matches> I resorted to log(a/b) = log(a) - log(b)
5715 15:08 <@matches> Except it doesn't actually have logarithm either
5716 15:08 <@matches> It has "Number of digits needed in base X"
5717 15:08 <@matches> Close enough...
5718 15:09 <@matches> Actually no it's awful
5719 17:32 <@matches> I'm going to have to do some disgusting template-yness things to Rect and Bezier I think
5720 17:47 <@matches> Blergh
5721 17:47 <@matches> Having a universal Real type is really damn annoying now that I don't want a universal Real type
5722 17:48 <@matches> I can do: template <struct T = IPDF::Real> struct Rect {...};
5723 17:48 <@matches> But then you have to replace Rect with Rect<> everywhere
5724 17:49 <@matches> I could make very single class a template class with Real as the template
5725 17:49 <@matches> Because that isn't horrifying
5726 17:52 * matches sobs and makes Rect a typedef for "TRect"
5727 18:22 <@matches> A is for ambiguous
5728 18:31 * matches curses templates
5729 18:31 <@matches> They attempt to solve the flaws in preprocessor macros by not actually solving anything
5730 18:32 <@matches> So that you will give up and just not write the program
5731 18:32 <@matches> The purpose of templates is to stop people from programming
5732 18:42 <@matches> Maybe don't pull the latest commit if you want anything to work at all
5733 --- Day changed Fri Oct 03 2014
5734 14:53 <@matches> The GMP attitude to logarithms seems to be "You can just write your own" which is nice, except that writing your own is not actually that easy to do well
5735 15:39 <@matches> This is rather annoying
5736 15:39 <@matches> When running the "Performance vs View Bounds" test, the bottleneck is in printing the View bounds
5737 15:40 <@matches> Because it requires a log of logs
5738 15:40 <@matches> *log of lots
5739 15:40 <@matches> *lot of logs
5740 15:40 <@matches> *lots of lumberjacks
5741 15:40 <@matches> *whatever
5742 15:41 <@matches> At least I seem to have made it O(1) where 1 is depressingly large constant
5743 15:55 <@matches> And so, after a day and a half, I can hopefully actually print bounds that are outside the range of representable doubles
5744 15:56 <@matches> Then I guess I'll have to do some kind of hack to get the "e-XXX" part so I can actually import them in numpy/gnuplot
5745 --- Day changed Sat Oct 04 2014
5747 13:55 <@matches> So, time investment
5748 13:55 <@matches> Do I make graphs
5749 13:55 <@matches> Or do I draw a turtle.svg
5750 13:55 <@matches> So that we can make the "Turtles all the way down" reference
5751 13:55 <@matches> Tough decision
5752 13:56 < Pommers> Turtles
5753 14:19 <@matches> I think fglrx must have a memory leak
5754 14:20 <@matches> Even after the program quits everything dies
5755 14:20 <@matches> And stays dead
5756 14:21 <@matches> Apparently I only have 80MB of memory left, yet the most anything in top is using is 100MB
5757 14:21 <@matches> ...
5758 14:21 <@matches> It's still decreasing
5759 14:21 <@matches> I have 10 MB left ?
5760 14:22 <@matches> Farewell
5761 14:26 < Pommers> glhfdd
5762 15:23 <@matches> http://szmoore.net/ipdf/code/src/svg-tests/turtle.svg
5763 15:23 <@matches> Can't be bothered colouring it in
5764 15:23 <@matches> It occurs to me that if we had recursive SVG support we would be Winning (TM)
5765 15:24 <@matches> I wonder how hard it would be to hack that on...
5766 15:24 <@matches> Need to map SVG "id" attributes to a Path
5767 15:24 <@matches> Need to allow Path to include Path
5768 15:24 <@matches> Hmm...
5769 15:24 <@matches> It's probably going to be as hard as shading
5770 15:25 <@matches> ie: Significantly harder than it should be due to the somewhat flawed design of our Object representations
5771 15:25 <@matches> We really only need Path and none of these other shaders
5772 15:26 <@matches> Maybe one Path shader per path too
5773 15:26 <@matches> As opposed to one Bezier shader for every single Bezier
5774 15:26 <@matches> Well, same shader but with different VBOs
5775 15:26 <@matches> Or IBOs
5776 15:26 <@matches> whatever
5777 15:28 <@matches> The problem is I can't implement Turtles all the Way Down in such a way that it shows all the turtles without either a) Making the DebugScript turing complete, or b) Implementing recursive SVGs
5778 15:29 <@matches> And option b) has the advantage of the document still being interactive whilst option a) restricts the view to what the script is doing
5779 15:30 <@matches> Also if the quad tree was working, turtles all the way down would be a good test of it, because the Arbitrary Precision bounds gets really slow since it has to always transform every path even those that are off screen
5780 15:31 <@matches> I should probably start writing a thesis but I really want to actually have something impressive to write about
5781 15:31 <@matches> I mean I still have to write about all the other stuff as well
5782 15:31 < Pommers> Yay off by one errors!
5783 15:32 <@matches> What, with the Log10 ?
5784 15:32 <@matches> It shouldn't be off by that much
5785 15:32 <@matches> Wait how are you running this
5786 15:32 <@matches> Wait you're talking about something else -_-
5787 15:32 < Pommers> Yes
5788 15:32 < Pommers> I've spent the past day looking for this error
5789 15:34 <@matches> Good job?
5790 15:34 < Pommers> Nope. So annoying. THought I just had it. Nope.
5791 15:35 <@matches> Sounds like we could make you an honarary IPDF developer
5792 15:35 < Pommers> I thought I already was :P
5793 15:38 < Pommers> Nope. Now my numbers are more off :/
5794 16:07 <@matches> Try ParanoidNumber
5795 16:07 <@matches> Guaranteed to give semi-accurate segfaults
5796 16:10 <@matches> Trying to look up some paper to justify my Logarithm approximation (because doing maths myself is hard)
5797 16:10 <@matches> "Computing discrete logarithms with the parallelized kangaroo method"
5798 16:10 <@matches> I... don't... even...
5799 16:11 <@matches> Why isn't there a paper on this
5800 16:11 <@matches> It shouldn't be hard, ergo, someone should have written a paper
5801 16:14 * matches really regrets only getting 60% in applied maths
5802 16:14 <@matches> I'm fairly sure this sort of thing was in it
5803 16:14 <@matches> Series approximations and stuff
5804 16:45 <@matches> Argahsdgf
5805 16:45 <@matches> It's actually hard to define what we *want* from our arbitrary precision document
5806 16:46 <@matches> Hmm
5807 16:46 <@matches> You have a view port, and you want primitives that are scaled to the view port to appear identical, regardless of where the view port is
5808 16:46 <@matches> Does that work
5809 16:46 <@matches> sulix are you alive
5810 16:46 <@matches> Is the quad tree going to be ok
5811 16:48 < sulix> Alive might be overstating it slightly. The quadtree is still kind-of functional.
5812 16:48 < sulix> (None of your commits have broken anything obvious, which is nice)
5813 16:49 * Pommers stabs floating point ints
5814 16:49 < Pommers> 0.0000000001 off
5815 16:50 <@matches> Most of my commits affect stuff that's wrapped in a bunch of #ifdef QUADTREE_DISABLED
5816 16:50 <@matches> I added some typedef's but they are all just equivelant to Real if QUADTREE_DISABLED
5817 16:51 <@matches> Pommers: Are you using floats or doubles? And if so, change to long doubles
5818 16:51 < sulix> That was very kind of you. I'm not sure how many drastic changes the quadtree can take at this point.
5819 16:52 < sulix> Was is SPARC that supported "quads"?
5820 16:52 <@matches> I'm kind of curious as to whether anything actually supports Base 10 floats
5821 16:52 < Pommers> Doubles
5822 16:52 < Pommers> I could try long doubles
5823 16:52 <@matches> Because it's in the standard
5824 16:53 <@matches> long doubles will probably help
5825 16:53 <@matches> I haven't run out software with long doubles yet because I'm really expecting it to do amazingly better than Gmp rationals
5826 16:54 <@matches> In theory, with enough memory, GMP rationals will win
5827 16:56 * matches suddenly realises just how much stuff there is that needs to be done
5828 16:57 * matches cries
5829 16:57 < Pommers> My great commit log... https://github.com/mpomery/CITS3402/commits/master
5830 16:57 * sulix sobs sympathetically
5831 16:58 <@matches> That has nothing on the IPDF commit logs I'm sorry
5832 16:59 <@matches> I should really stop deliberately making the commit messages over dramatic
5833 16:59 < Pommers> Probably not
5834 16:59 < Pommers> Definately
5835 17:01 < sulix> The more I think about the quadtree, the more certain I am that it should just be crashing all the time.
5836 17:10 * sulix quietly wonders if he can steal some code from that xkcd comic.
5837 17:18 <@matches> So, it only works if you give it a starting SVG because it loads the whole thing into the root node and then builds the other nodes as you zoom
5838 17:18 <@matches> And inserting things at some arbitrary point is hard
5839 17:18 <@matches> Could you... brute force... recreate the entire thing from the root node
5840 17:19 < sulix> The plan (after the great minus-sign move of '14) is to simply create a new document from wherever you add something, and "overlay" it.
5842 17:20 <@matches> sounds clever
5843 17:20 <@matches> Hmm
5844 17:20 <@matches> There's a whole bunch of definitions and obvious assumptions that I need to state
5845 17:20 <@matches> How do I even explain this
5846 17:21 <@matches> So, bresenham lets you draw a line exactly given the end points => You only care about rounding the end points
5847 17:21 <@matches> Bezier = decompose into Bresenham => You care about rounding in the Bezier decomposition
5848 17:21 < sulix> Well, bresenham doesn't let you draw a line exactly at all.
5849 17:21 <@matches> Pfft
5850 17:21 <@matches> Exactly shchmexactly
5851 17:22 <@matches> To pixel accuracy I guess
5852 17:22 <@matches> If the Bezier is in screen space and the buffer is small enough, then the bezier decomposition is accurate-enough to within the nearest pixel
5853 17:22 <@matches> So you care about transforming the bezier control points
5854 17:22 <@matches> They can be written relative to the bounds so you care about transforming the bounds
5855 17:23 <@matches> The bounds can be written relative to the path ergo you only need to transform the path bounds
5856 17:23 <@matches> Ergasdfadfg
5857 17:23 <@matches> It's all relative
5858 17:24 <@matches> There's no point being able to render something in bounds 1e-100000 if you don't also want to store something up at width 1
5859 17:24 <@matches> Otherwise you may as well just use bounds with width 1 and then do a multiplication before you printf them
5860 17:24 <@matches> Problem solved!
5861 17:29 <@matches> What I am trying to do makes perfect sense in my head and absolutely no sense translated to english
5862 17:29 <@matches> And translating it to maths is just asking for trouble
5863 17:29 < Pommers> #pragma omp single means that I'll start up a single thread to do something right?
5864 17:30 <@matches> Well, it won't create a new thread, it means a single thread out of the current pool will do it
5865 17:30 < Pommers> Okay. How can I get OpenMP to create a thread to do one of several for loops?
5866 17:31 <@matches> #pragma omp parallel for
5867 17:31 <@matches> Divides the loops up evenly amongst the threads
5868 17:31 <@matches> Is that what you want?
5869 17:31 < Pommers> Nah, multiple for loops, not the one
5870 17:32 <@matches> Is the loop inside a parallel secion already?
5871 17:32 < Pommers> Unless I am misunderstanding what parallel for does
5872 17:32 < Pommers> Nope, not inside a loop yet
5873 17:32 <@matches> Unless you explictly use a #pragma omp parallel, the code will be executed by the main thread
5874 17:33 < Pommers> So if I go:
5875 17:33 < Pommers> #pragma omp parallel
5876 17:33 < Pommers> #progma omp single
5877 17:33 < Pommers> for loop
5878 17:33 < Pommers> #progma omp single
5879 17:33 < Pommers> for loop
5880 17:34 < Pommers> #pragma omp barrier
5881 17:34 < Pommers> The for loops with get executed by potentially different threads?
5882 17:34 <@matches> Yes
5883 17:34 <@matches> And they only get executed once
5884 17:34 <@matches> But you can't predict the order
5885 17:34 < Pommers> Cool
5886 17:34 <@matches> Anything outside the "single" gets done by all threads
5887 17:35 <@matches> I don't think you need the barrier
5888 17:35 <@matches> There are implicit barriers at the end of each parallel section
5889 17:35 <@matches> Oh, also make sure the scope is correct
5890 17:35 <@matches> Put {} around things
5891 17:35 < Pommers> Okay
5892 17:35 < Pommers> Imma come back to it in a moment
5893 17:35 < Pommers> Forgot to go pick up my housemate :/
5894 17:36 < Pommers> ty matches
5895 17:36 <@matches> (You don't need the {} if you have a loop immediately after the parallel directive though)
5896 17:37 <@matches> Using OMP with certain text editors is a nightmare because they want to automatically put preprocessor directives in the first column
5897 17:37 <@matches> Which totally screws up indentation
5898 17:37 <@matches> Certain uses of OMP don't actually use indentation in the first place though, which I can't understand at all
5899 17:37 <@matches> They indent the for loop but not the parallel directive that applies to it
5900 17:38 * matches shudders
5901 17:38 <@matches> These are the same lecturers that discouraged us from using functions because function calls slow the program down
5902 17:38 <@matches> Whole program should be in main()
5903 17:38 <@matches> Goto if necessary
5904 17:42 <@matches> (I compiled something with and without a function call and saw that it was the exact same assembly and then proceeded to totally ignore any programming advice they gave us from then on)
5905 17:43 <@matches> I guess it would be true if you had an ancient compiler
5906 17:46 < sulix> (Secret: function calls _are_ slow. So is goto, though, so the point is moot.)
5907 18:07 <@matches> Baha
5908 18:08 <@matches> There is some kind of deep maths behind why transforming things relative to Path bounds works
5909 18:08 <@matches> I guess
5910 18:08 <@matches> It is both obvious and not obvious
5911 18:08 <@matches> It is very frustrating
5912 18:09 <@matches> I think this project is somewhat more appropriate for a maths major than mechatronics engineers...
5913 18:09 <@matches> I guess I don't have to go into the maths, but if I don't then I'm basically just saying "I tried this because it seemed like it would work, and it did!"
5914 18:10 <@matches> Even the Logarithm I implemented has nice maths in it
5915 18:11 <@matches> Which is just slightly beyond my ability to explain satisfactorily
5916 18:14 < sulix> I don't think maths-majoring has helped me much. It gives me snarky things to say about floats vs rationals, but doesn't stop the fact that I've now spent over a week trying to work out where the -ve signs need to go in the quadtree.
5918 18:18 < sulix> On the bright side, by the power of trial, error, cribbing from the xkcd comic, and forcibly recreating all of the GPU buffers 4 times a frame, panning around the quadtree now seems to work.
5919 18:18 <@matches> Yay!
5920 18:18 <@matches> Can you zoom yet
5921 18:19 < sulix> Almost*
5922 18:19 <@matches> Zooming is just panning but with different numbers
5923 18:19 <@matches> Or is panning just zooming but with different numbers
5924 18:19 <@matches> Something
5925 18:19 <@matches> Some of the numbers are zero
5926 18:19 <@matches> Or possibly 1
5927 18:20 <@matches> Let me know when eye_of_the_rabbit works and we will have a performance show down!
5928 18:20 < sulix> Zooming in works about as well as it did before. Zooming out works as long as you don't try to zoom out beyond the original view.
5929 18:20 < sulix> Panning around the document works, though might have some problems if you try to pan outside (0,0)-(1,1) from the original view.
5930 18:20 <@matches> Who needs to zoom out anyway
5931 18:22 <@matches> I'm going to make the debug font more dramatic to answer some criticism
5932 18:22 <@matches> Apparently I need to indicate the width of the view in SI units
5933 18:22 <@matches> (And in really big text)
5934 18:25 < sulix> But there are no units for the view!
5935 18:25 <@matches> On my monitor the view is 22mm
5936 18:25 <@matches> window rather
5937 18:25 <@matches> So just multiply the view width by that
5938 18:25 < sulix> Just short of an attoparsec...
5939 18:33 < sulix> Okay, fixed the "you have to zoom in on non-binary-fraction boundaries for infinite precision issue that's plagued the quadtree for months.
5940 18:33 < sulix> Now we can see: discontinuities at quadtree node boundaries!
5941 18:33 < sulix> Fortunately, I have an excuse for leaving them there!
5942 18:34 < sulix> "The object wasn't added at that resolution."
5943 18:34 <@matches> Heh
5944 18:34 < sulix> (Also: we totally don't solve cubics analytically, so they're worse than they need to be)
5945 18:35 * sulix smells a "further work" section.
5946 18:35 <@matches> You have a "current work" section? Jealous
5947 18:36 * sulix didn't say that.
5948 18:36 <@matches> Apparently as long as I make something pretty people will be impressed
5949 18:36 <@matches> I wonder if it's too late to implement shading (again)...
5950 18:36 < sulix> Maybe different-coloured outlines?
5951 18:36 <@matches> Haha
5952 18:36 < sulix> More demo svgs?
5953 18:36 <@matches> I quite like the turtles demo
5954 18:37 <@matches> Unfortunately for reasons mentioned earlier it can't actually add turtles forever
5955 18:37 <@matches> Well it could
5956 18:37 <@matches> But you couldn't stop it and go look at them
5957 18:37 <@matches> You'll only ever see the bottom one
5958 18:37 <@matches> The demo that draws 20 and then zooms back out and then in again is probably good enough
5959 18:38 < sulix> Hmm... eye_of_the_rabbit is segfaulting the quadtree.
5960 18:38 <@matches> It clears the document every so often
5961 18:38 <@matches> Try commenting that out first?
5962 18:38 < sulix> That'd probably do it.
5963 18:38 <@matches> I thought we probably wanted to test zooming without altering the number of primitives, and that was the only way I could think of doing it
5964 18:38 < sulix> It also made the quadtree slow down, though that's a side effect of something I like to call "shitty coding".
5965 18:39 < sulix> The quadtree is full of it.
5966 18:45 * sulix is merging now. I love the #pragma message that happens for every file!
5967 18:48 < sulix> Code is up for your enjoyment/horror.
5968 18:48 * sulix breaks for dinner.
5969 21:01 <@matches> Yeah the #pragma message probably wasn't that good an idea
5970 21:02 <@matches> The problem was there are 3 different "Reals" now :S
5971 21:02 <@matches> So I wanted to make sure I knew which ones it was using
5972 21:08 < sulix> Fair enough.
5973 21:11 < sulix> Oh dear god, what have you done to the debug text?
5974 21:13 * sulix shudders.
5975 21:14 <@matches> It's readable!
5976 21:14 <@matches> Through the automerge seems to have forgotten the #ifdef 0
5977 21:15 <@matches> Oh that stuff is printed somewhere else...
5978 21:16 <@matches> Anyway, adding stuff seems to work!
5979 21:16 <@matches> But not from the script for some reason
5980 21:16 <@matches> Oh right, the clearing
5981 21:17 < sulix> It shouldn't work totally in the quadtree, I've just made it break more quickly.
5982 21:17 < sulix> Basically, you can (a) only add to the root node and (b) nodes are not yet updated with new objects if they already exist.
5983 21:17 <@matches> It doesn't work when things overlap nodes?
5984 21:17 <@matches> Right
5985 21:17 <@matches> I was getting excited for a minute
5986 21:18 < sulix> On the bright side, it no longer creates an entirely new quadtree for every individual object and it actually adds new things to new child nodes.
5987 21:20 <@matches> Running eye of the rabbit I see a LOT of "Recreate buffers" messages but no actual rabbits
5988 21:21 < sulix> It now recreates the buffers every frame because that was easier than getting it to recreate the buffer in arcane parts of quadtree code.
5989 21:29 <@matches> Well, it does work when you add something to the root node I guess
5990 21:30 < sulix> Almost.
5991 21:31 < sulix> If you zoom in, zoom out, add something to the root node and then zoom in again it isn't there.
5992 21:31 < sulix> But that will (fingers crossed) soon be resolved through the power of recursive functions.
5993 21:31 <@matches> Hmm, that's almost what happens when you use IEEE singles!
5994 21:34 < sulix> I'm still trying to work out how I can delay the generation of new child overlays until after all new objects have been added.
5995 21:34 < sulix> If I do things the obvious way, it'll generate thousands of overlays and totally kill performance.
5996 21:35 <@matches> Damn
5997 21:35 <@matches> I think I just need to write lots of stuff and fast
5998 21:35 <@matches> Well
5999 21:35 <@matches> Some form of result would be nice
6000 21:35 <@matches> A graph
6001 21:41 < sulix> I think I've managed to make the quadtree even slower than ParanoidNumber<GmpRat> would be.
6002 21:42 <@matches> How though? It seemed like it would be efficient
6003 21:43 <@matches> It doesn't have to render the whole document
6004 21:43 <@matches> s/render/transform
6005 21:44 < sulix> Well, it has to regenerate the entire quadtree everytime there is a new bezier, and there are lots of O(n) linked lists and recursive functions there.
6006 21:44 < sulix> And doing this the naïve way generates a pathalogically-worst case situation.
6007 21:45 < sulix> (I'm still waiting for my first "add something test" to complete.
6009 21:45 <@matches> Hmm
6010 21:46 <@matches> I guess once you already have a document with lots of detail in it, a quadtree is good
6011 21:46 <@matches> But actualy making the document with lots of detail in it...
6012 21:46 * sulix suspects it might just be infinite looping.
6013 21:46 <@matches> Why does it have to rebuild the entire quadtree? Can't it just modify the current node + nearest neighbours?
6014 21:47 <@matches> Or go up to the parent and add all the objects in the parent, then split them up for the children
6015 21:47 <@matches> Hmm
6016 21:48 <@matches> You need to be able to add objects to the ancestors as you zoom out I guess
6017 21:48 <@matches> But you should be able to do it lazily?
6018 21:48 <@matches> In theory
6019 21:48 < sulix> In theory.
6020 21:48 * matches is glad he is not writing the quadtree after all
6021 21:48 < sulix> The big problem is that adding things to the quadtree needs to be done in bulk to be efficient.
6022 21:49 < sulix> i.e. If you add a whole bunch of things to one node, and then a whole bunch of things to another, it's okay.
6023 21:49 < sulix> But if you alternate (add something to one), (add something to the other) repeatedly, you fragment the quadtree and it treats each bézier as it's own overlay from then on.
6024 21:49 < sulix> The easy way to do things is the latter way.
6025 21:50 < sulix> Because it's hard to get all of the info you need about what things have changed in bulk at the end of an entire operation.
6026 21:51 < sulix> But clearly doing things curve by curve is not going to cut it if even the simplest examples take over 5min/frame to render.
6027 21:52 <@matches> Well if it's any condolance, ParanoidNumbers are still worse than that
6028 21:53 * sulix should say "5min and counting"/frame.
6029 21:53 < sulix> I've not yet had a frame render.
6030 21:53 <@matches> Hey, at least you can still type into a terminal
6031 21:54 <@matches> When I run ParanoidNumbers I can't even kill the process for 5 minutes...
6032 21:54 < sulix> I am thankful for these small mercies.
6033 21:55 <@matches> I'd hoped they'd be a bit better with the more efficient transformation model but apparently not enough
6034 21:55 <@matches> In fact they are segfaulting again
6035 21:56 <@matches> I thought I stopped that
6036 21:56 <@matches> The plot thickens!
6037 21:57 <@matches> They segfault when std=c++0x
6038 21:57 <@matches> But not when std=c++11
6039 21:57 < sulix> Oh dear god.
6040 21:57 <@matches> I suspect it is related to the use of "auto"
6041 21:57 <@matches> And it is probably why google told you not to use it
6042 22:00 <@matches> Actually it's more likely because I forgot to make clean
6043 22:01 <@matches> Well I guess I will come back to them once I actually have something resembling a thesis draft
6044 22:01 <@matches> Hahaha
6045 22:02 <@matches> The turtles script is way cooler than the rabbit one by the way, if you haven't run it yet
6046 22:02 <@matches> (But it breaks the quad tree even more)
6047 22:04 <@matches> And PN do work if you restrict the size to like, 3
6048 22:05 <@matches> News in 5 minutes: Is it actually better than just using doubles
6049 22:06 <@matches> It seems to speed up every so often and then slow down again
6050 22:06 <@matches> Aaamd
6051 22:06 <@matches> Yeah that turtle rounded to the wrong spot
6052 22:07 <@matches> I should see how well our custom Rational implementation does too
6053 22:12 <@matches> I am confused by PN
6054 22:12 <@matches> A size limit of >= 5 causes apocalyptic memory usage
6055 22:12 <@matches> A size limit of 4 is totally fine and actually not much slower than Gmprat
6056 22:13 <@matches> Also I have a sudden urge to compile this with MinGW so that the Engineers who mark it can actually run it
6057 22:13 <@matches> I mean, they probably won't run it
6058 22:13 <@matches> But the last thing I want is for them to try and not be able to
6059 22:13 <@matches> And then just assume I fudged the results
6060 22:14 <@matches> I mean, a rabbit scaled to 1e-6 and then viewed under 1e6 magnification looks the same as the original rabbit...
6061 22:14 <@matches> It's not really even worth including both images
6062 23:56 <@matches> One does not simply cross compile for MinGW
6063 23:57 <@matches> Although I must say I was bracing for it to be worse
6064 23:58 * matches stares in shocked amazement at NO ERROR MESSAGES
6065 23:58 <@matches> COULD IT BE
6066 23:58 <@matches> COULD I HAVE SUCCEEDED?
6067 23:59 < Pommers> Nope.
6068 23:59 <@matches> -_-
6069 23:59 <@matches> If we didn't need spectators I'd totally /kick you
6070 23:59 < Pommers> 99 commits in my assignments repo >.>
6071 23:59 <@matches> Pfft
6072 23:59 <@matches> Commit them all to IPDF
6073 --- Day changed Sun Oct 05 2014
6074 00:00 < Pommers> I can kill your percentage of lines here. That's what you asked me to do
6075 00:00 < Pommers> Also
6076 00:00 < Pommers> First
6077 00:00 <@matches> err:module:import_dll Library SDL2.dll (which is needed by L"Z:\\home\\sam\\Documents\\University\\2014\\ipdf\\code\\src\\ipdf") not found
6078 00:00 <@matches> Hmm
6079 00:03 <@matches> If I try link with static SDL2 everything goes to hell though :(
6080 00:04 <@matches> I guess I can just add SDL2.dll
6081 00:05 <@matches> It runs in Wine
6082 00:05 <@matches> Does that make it more or less likely to run under windows
6083 00:05 <@matches> Ah crap the Qt stuff isn't compiled in though
6084 00:05 <@matches> Qt is annoying
6085 00:05 <@matches> They appear to ask you for money if you try download their source
6086 00:06 <@matches> Also they don't like you using Qt4 and I really can't remember why I picked Qt4 instead of Qt5 now
6087 00:07 <@matches> I guess I assumed it would be better, after experiences with GNOME 2 -> GNOME 3
6088 00:07 <@matches> New = terrible
6089 00:08 <@matches> Oh dur this is what apt-get source is for
6090 00:10 <@matches> I will just feel a lot safer if I know that the marker can actually run the software without needing to know linux
6091 00:10 <@matches> Call me a Paranoid Number...
6092 00:10 <@matches> (Ho ho ho)
6093 00:11 <@matches> (This is probably a sign that it is bed time)
6094 00:13 < Pommers> Yeah, you should probably go to bed matches
6095 00:17 <@matches> What about you?
6096 00:18 < Pommers> I'm already in bed
6097 00:18 <@matches> SULIX
6098 00:18 <@matches> THERE IS A GOTO
6099 00:18 * Pommers grabs out his pitchfork
6100 00:20 <@matches> I only saw it because it is also outside an #ifdef and raises an "unused label" warning
6101 00:20 <@matches> How many others are there...
6102 00:24 <@matches> There is also a goto in stb_truetype.h but I'll let that one slide since sulix didn't write it
6103 10:01 < sulix> Gah: runs fine, no warnings in valgrind, crashed with memory corruption errors outside of valgrind.
6104 10:02 < sulix> This is clearly karmic retribution for the goto.
6106 10:25 <@matches> Are you compiling with or without the quadtree?
6107 10:25 <@matches> Because I set the VReal and PReal typedef to ParanoidNumber
6108 10:26 <@matches> Running in valgrind ParanoidNumber is equivelant to a double since the FE_INEXACT exception isn't raised
6109 10:26 <@matches> Running outside of valgrind... anything could happen
6110 11:31 <@matches> One does not simply compile with MinGW and have things actually work
6111 11:32 <@matches> Wine likes it
6112 11:32 <@matches> Actual windows seems to have some difficulty with OpenGL functions
6113 11:32 <@matches> Also thank goodness for gdb compiled for windows
6114 11:33 <@matches> Because "ipdf.exe has experienced an error, would you like to send Microsoft a report?"
6115 11:33 <@matches> Is really not helpful
6116 11:33 <@matches> Command prompt won't even tell you "Segmentation fault"
6117 11:39 <@matches> Hmm, needs more "-lopengl32" ?
6118 11:39 <@matches> But why weren't there "undefined reference" errors
6119 11:39 <@matches> So confuzled
6120 11:40 <@matches> I bet even if it runs it won't work due to geometry shaders or something
6121 11:49 <@matches> Oh dear
6122 11:49 <@matches> This is to do with the several thousand lines of autogenerated stuff in gl_core44.*
6123 11:50 <@matches> Isn't this what GLEW was supposed to be for
6124 11:50 <@matches> Except GLEW is ancient so nobody uses it
6125 11:50 <@matches> Shouldn't someone reinvent a less shitty GLEW then
6126 11:50 < sulix> Apparently glew works again.
6128 11:51 < sulix> I did get glew working under mingw32 yesterday, as it happens.
6129 11:51 <@matches> Commit please?
6130 11:51 <@matches> Oh right for something else :P
6131 11:51 <@matches> What magical incantations are required
6132 11:51 < sulix> Yeah. And by working, I mean compiling, not running.
6133 11:51 <@matches> Baha
6134 11:52 < sulix> (Though I think it was another bit of the code that segfaulted)
6135 11:52 <@matches> Why is getting OpenGL working so horrifying
6136 11:52 <@matches> It's supposed to be a standard
6137 11:52 < sulix> So the gl_core44.* file was supposed to be basically glew, but hooked into SDL.
6138 11:52 <@matches> It's not working on Windows though
6139 11:53 < sulix> Do you have OpenGL drivers on windows?
6140 11:53 < sulix> They're not installed by default.
6141 11:53 <@matches> Yeah on my desktop
6142 11:53 <@matches> Although I did switch to Maaxen
6143 11:53 < sulix> Hmmm...
6144 11:53 <@matches> There were other things wrong
6145 11:53 <@matches> I wasn't liking with -lopengl32
6146 11:54 <@matches> Although wine didn't seem to care
6147 11:54 < sulix> In theory, the glcore44 and SDL stuff should load opengl32 at runtime if needed.
6148 11:54 < sulix> So you shouldn't, in theory, need to link to them.
6149 11:54 < sulix> But that's very much "in theory".
6150 11:55 <@matches> I will investigate the driver situation on Maaxen
6151 11:55 <@matches> Urgh why can't windows just have `apt-cache search`
6152 11:55 <@matches> And `lspci | grep vga`
6153 11:55 <@matches> Lets see... "Device Manager"
6154 11:55 < sulix> Maaxen will almost certainly not have working GL 3+ drivers.
6155 11:56 <@matches> Oh it's a VM
6156 11:56 * matches backs away very quickly
6157 11:56 <@matches> Well I can't keep rebooting every time I try and fix a segfault
6158 11:56 <@matches> Bloody scope broke my other laptop
6159 11:57 <@matches> I used to have Windows XP...
6160 11:57 <@matches> Oh wait I smashed the hard drive open so I could draw it for an assignment
6161 11:58 < sulix> You should subscribe to the "if it works under wine, it's good enough" philosophy.
6162 11:58 * sulix has shipped things that way.
6163 12:00 < sulix> Anyway, the quadtree is still segfaulting when not run under valgrind by (I think) getting the video card to somehow corrupt unrelated system memory or something.
6165 12:01 < sulix> It always crashes when trying to glDeleteBuffers() which is really disturbing.
6166 12:02 <@matches> I've noticed that when there are a lot of objects, and you kill the program, you never get the memory back ever
6167 12:02 <@matches> And your system stays slow
6168 12:02 <@matches> Which is a bit of a problem
6169 12:02 <@matches> Need to reboot every hour or so of testing...
6170 12:02 < sulix> Okay, I haven't had that happen.
6171 12:02 <@matches> I guess that's just fglrx amazingness
6172 12:03 < sulix> Guess so.
6173 12:03 <@matches> For example, I can run turtles_all_the_way_down.script without anything crashing but then I am left with like 20MB and when I open Iceweasel everything goes to hell
6174 12:03 < sulix> On the bright side, the fglrx developers are planning to get rid of the fglrx kernel module and just have their closed-source GL driver talk to the open-source kernel driver.
6175 12:04 <@matches> Is that... good
6176 12:04 < sulix> We'll see, but I can't imagine it making fglrx _worse_
6177 12:04 <@matches> The open source driver doesn't work that well either
6178 12:05 <@matches> I sense another few days trying to get X working again after an `apt-get upgrade` at some point in the future then :(
6179 12:08 < sulix> Don't worry, they're still in the "have meetings" phase.
6180 12:37 <@matches> I think the glDebug stuff is giving windows issues
6181 12:37 <@matches> At least running in actual windows not a VM
6182 12:37 <@matches> Running in a VM is not conductive to working
6183 12:37 <@matches> Sanity tells me to give up
6184 12:38 <@matches> It can do glCreateProgram though!
6185 12:38 <@matches> glObjectLabel seems broken
6186 12:39 < Pommers> http://th01.deviantart.net/fs70/PRE/i/2012/076/1/2/oh_my_glob__by_tsubasaangel-d4t0g68.jpg Oh my glOb
6187 12:39 <@matches> Yes
6188 12:40 <@matches> It would be nice if the extension manager (if you could call gl_core44 that?) used stubs instead of NULL
6189 12:41 <@matches> I mean for debugging
6190 12:41 <@matches> Then you can have it print "This function doesn't actually exist guys"
6191 12:41 <@matches> gdb is nicer I supose
6192 12:41 <@matches> In fact if I weren't trying to cross compile on linux and then reboot and test in windows this would be a lot easier
6193 12:58 <@matches> Ok screw this
6194 12:58 <@matches> I need to write a thesis
6196 12:58 <@matches> But after lunch
6197 13:51 <@matches> http://thecodelesscode.com/case/57
6198 14:12 < Pommers> Lunch sounds good
6199 14:23 <@matches> I'm currently reading codelesscode using the excuse that the turtles tester takes a very long time to sigfpe
6200 14:23 <@matches> (But it sigfpe's nonetheless)
6201 14:24 <@matches> Can we disable sigfpe for overflow/underflow casts from double to float
6202 14:24 <@matches> That would make things easier
6203 14:24 <@matches> Well
6204 14:24 <@matches> Assuming that the float casts to FLT_MAX or FLT_MIN if it would underflow/overflow
6205 16:09 <@matches> sulix: I can no longer load anything into the document
6206 16:09 <@matches> Congratulations though, this means it always renders accurately...
6207 16:20 <@matches> Ok...
6208 16:20 <@matches> We now have
6209 16:20 <@matches> #ifdef
6210 16:20 <@matches> ...
6211 16:21 <@matches> m_count++;
6212 16:21 <@matches> return m_count;
6213 16:21 <@matches> #else
6214 16:21 <@matches> return m_count++;
6215 16:21 <@matches> #endif
6216 16:22 <@matches> Because I don't know if the quadtree *needs* to return an invalid array index (it doesn't even use the same array...) but over in monotree land we like our array indices to be valid
6217 16:24 <@matches> Now what was I doing
6218 16:24 <@matches> Ah, seeing if OpenGL really could do Wu lines
6219 16:25 <@matches> So when I say it is in my report I am not lying
6220 16:25 <@matches> Report...
6221 16:25 <@matches> The thing I'm writing
6222 16:25 <@matches> Write
6223 16:25 < Pommers> How long is the report going to be matches?
6224 16:25 < Pommers> And when are you going to finish it?
6225 16:25 <@matches> It has to be 6000 words
6226 16:26 <@matches> My LitReview is ~8000 ...
6227 16:26 <@matches> Although that might be including the SVG and postscript examples
6228 16:26 <@matches> I think I should take those out?
6229 16:26 <@matches> You'd think having a huge pre-written document would make it easier but no
6230 16:27 <@matches> Enabling LINE_SMOOTH leads to, of all things, SIGFPE
6231 18:25 <@matches> I am making a whole lot of no progress here
6232 18:25 <@matches> Draft due on Friday...
6233 18:26 <@matches> I need to produce a Design Process
6234 18:26 <@matches> The software wasn't so much designed as congealed
6235 18:27 <@matches> I forgot this is an Engineering Project not CS and therefore needs to be done in such a way that future Engineers can improve upon the work...
6236 18:27 * matches sobs some more
6237 18:27 <@matches> Rectangles and Circles don't even work anymore
6238 18:28 <@matches> I sacrificed them for the glory of the Bezier To Path transformations
6239 18:29 <@matches> Also, instead of a transformation per path, it would make more sense to use one per SVG
6240 18:30 <@matches> It's basically the localised coordinate approach but not as fancy as a quadtree
6241 18:30 <@matches> If you were building up a document one bezier at a time by drawing them, you'd have to do something fancy to make your paths
6242 18:31 <@matches> (Un)Fortunately we don't have an editor anyway
6243 18:47 <@matches> Hmm, that really needs reverse parenthesis to indicate that the word is "Unfortunately" but the "Un" is removed
6244 18:47 <@matches> )Un(Fortunately
6245 18:47 <@matches> I have invented a new gramatical construct-thing
6246 18:47 * matches adds it to the list of results
6247 --- Day changed Mon Oct 06 2014
6248 18:02 <@matches> So the good news is I worked out the transform, the bad news is it's yet another horrible hack
6249 18:03 <@matches> Which will undoubtably be extremely slow if I use Gmprat
6250 19:01 <@matches> It's amazing how useful a "grid.svg" that is just lines which should in theory be one pixel apart is
6251 19:01 <@matches> If we could draw it in a different colour it would be even more useful
6252 19:01 <@matches> But the clock of doom is counting down to the epsilon of time
6253 19:17 <@matches> A quick bit of <commenting out large parts of the shading algorithm and enabling shading anyway> later: http://szmoore.net/ipdf/fox-vector+grid.png
6254 19:20 <@matches> Wow that took 16 minutes
6255 19:20 <@matches> ... 19 if you count the last 3...
6256 19:20 <@matches> I'm running out of epsilons
6257 19:21 <@matches> If I represent the amount of time I have left on the project as a floating point number, then a single second is below the epsilon for floats, therefore each second will not affect the amount of time I have left, therefore I will never have to finish
6258 19:21 * matches resists the urge to verify if a single second is actually below epsilon or not
6259 --- Day changed Tue Oct 07 2014
6260 13:52 <@matches> http://ieeexplore.ieee.org.ezproxy.library.uwa.edu.au/stamp/stamp.jsp?tp=&arnumber=5577902
6261 13:52 <@matches> Uh oh
6262 13:57 < sulix> We did look at that briefly, IIRC, but ignored it because it was only integers, I think.
6263 16:05 <@matches> Yeah, although a rational is just two integers
6264 16:05 <@matches> But too late now
6265 16:05 <@matches> I am not making progress
6266 16:05 <@matches> I have deleted words
6267 16:05 <@matches> That is not progress
6268 16:05 <@matches> (I had 8000 words, I require 6000 words...)
6269 16:06 <@matches> I should probably get some performance tests done, but every time I try there ends up being something I have to code first
6270 17:25 <@matches> I have a somewhat less awful explanation of number representations that still manages to carefully not actually define a number
6271 17:26 <@matches> I wonder if I need a citation though
6272 17:28 <@matches> Does it classify as "blindingly obvious" or "known since the dawn of time" that you can write numbers as "\sum_{i=-\infty}^{\infty} d_i B^i" where "d_i" and B are also numbers (it's so meta)
6273 17:28 <@matches> You're the maths guy, surely there is a paper
6274 17:29 <@matches> If it's blindingly obvious, I probably shouldn't state it, so I guess it's not blindingly obvious...
6275 17:29 <@matches> But on the other hand it's just a fancy way of talking about what you learn in primary school...
6276 17:43 <@matches> http://thecodelesscode.com/case/137
6277 17:50 <@matches> Dear goodness it's hideous
6278 20:08 < sulix> matches: It's actually kind-of more complicated than that.
6279 20:08 < sulix> You might find it interesting to look up Cantor's diagonal argument.
6280 20:09 < sulix> (Which is technically a proof that the size of the set of rationals is different to the size of the set of reals, but it uses the idea of infinite sequences of digits)
6281 20:09 < sulix> For example \sum_{i=-\infty}^{\infty} d_i B^i actually gives you the reals, not the rationals.
6282 20:10 < sulix> But you have to be super-careful about what "infinity" is.
6283 21:11 <@matches> Did I say Rational
6284 21:11 <@matches> I thought I said real
6285 21:11 <@matches> No I said "number"
6286 21:11 <@matches> Wel
6287 21:12 <@matches> I will look up Cantor's diagonal bishop
6288 21:12 <@matches> I mean argument
6289 21:13 <@matches> But basically anything you are going to represent on a computer is either written as that sum or based on components that are
6290 21:13 <@matches> A float is like changing where the "0" is
6291 21:13 <@matches> An integer is starting the sum from zero
6292 21:14 <@matches> An arbitrary integer uses the same sum, but the digits are smaller integers
6293 21:14 <@matches> (In another base, and then the base of the arbitrary integer is 2^{N})
6294 21:14 <@matches> I don't know how much I am expected to explain this sort of thing
6295 21:15 <@matches> And of course a rational is two integers written using the sum
6296 21:15 <@matches> Infinity is...
6297 21:16 <@matches> Yeah
6298 21:17 <@matches> Screw that, if people don't like me writing a number as an infinite sequence of digits I will point out the name of my major (and then hope they don't ask why I am doing a project about numbers considering my major)
6299 21:17 <@matches> Other things I don't really want to include in my report are the 4 operations for floats
6300 21:18 <@matches> But I feel like some sort of explanation of why "(x - view.x) / view.w" totally dies
6301 21:18 <@matches> Is needed
6302 21:18 <@matches> I'll commit "grid.svg", you might find it useful
6303 21:18 < sulix> "If view.w is zero, you're in trouble."
6304 21:18 <@matches> Yeah
6305 21:19 < sulix> I'm sure grid.svg will lead me to much confusion!
6306 21:19 <@matches> No it's really cool
6307 21:19 <@matches> It is a 1px by 1px grid
6308 21:19 < sulix> But how will I know where I am?
6309 21:19 <@matches> So, because we add SVGs scaled to the view, whenever you insert it the view should be black
6311 21:20 <@matches> But if you zoom in, you can resolve the lines that are 1px apart in the original view
6312 21:20 * sulix really needs to get on that "adding SVGs scaled to the view" thing.
6313 21:21 <@matches> So, if you add it scaled to the view at some point and you *don't* see total blackness, it's showing you the locations in the document that can be represented when scaled to the view
6314 21:21 <@matches> ... If that makes any sense
6315 21:22 < sulix> Ah.... that's cool.
6316 21:22 <@matches> For example using single precision, at {5,5,1e-6,1e-6} you only get 4 lines (and also, the distance between the lines is the amount you need to move the mouse in order to translate)
6317 21:22 <@matches> Yeah, and hopefully it will be good for my bounds measurement test
6318 21:23 <@matches> I know it's a shitty measure but it's better than just saying "This picture looks funny!"
6319 21:23 <@matches> "This graph indicates how funny a picture will look"
6320 21:23 <@matches> I feel like graphs make things *look* more legitimate even if they aren't actually
6321 --- Day changed Wed Oct 08 2014
6322 12:48 <@matches> Writing up these "results" it really doesn't feel like I've achieved much of consequence...
6323 12:49 <@matches> It feels like I spent 9.5 months not understanding anything and then stumbled accross the blindingly obvious solution that should have taken a week.
6324 12:50 <@matches> Then again, we were talking about local vs global coordinate systems right at the start...
6325 12:51 <@matches> Anyway all I've done is introduce a local coordinate system which is expressed with arbitrary precision relative to the global coordinate system.
6326 12:51 <@matches> The only advantage I can see with the quadtree is that it clips things
6327 12:51 <@matches> Which is admittedly a really huge advantage
6328 12:51 <@matches> Since GMP rationals grow exponentially...
6329 12:51 <@matches> It's a very big time constant thanks to GCD though
6330 12:52 <@matches> I wish shading worked because these figures would look so much more awesome with colour
6331 12:57 <@matches> I really hate that I can't just include a pixel perfect copy of the png figures
6332 12:58 <@matches> There's a massive amount of antialiasing going on
6333 13:01 <@matches> ...
6334 13:01 <@matches> I've been writing too much LaTeX when I start using '%' as comment in our debug script
6335 22:57 <@matches> I must say, the one place where PDF's magical blurring of things makes them look better is the figures from back when shading almost worked
6336 22:57 < Pommers> LaTeX <3
6337 22:57 <@matches> It's not even something you can control in LaTeX
6338 22:58 <@matches> The PDF viewer will do whatever the hell it wants
6339 22:58 <@matches> Although I've realised that the screenshot tool in debian is also adding antialiasing of its own
6340 22:59 <@matches> So at some point between the draft thesis and the final thesis I guess I need to go through IPDF's own screenshot function
6341 22:59 <@matches> szmoore.net/ipdf/sam/presentation/presentation.pdf
6342 23:00 <@matches> The presentation with a big gap where I in theory explain Rational numbers, arbitrary precision integers, the GMP library, Why we only care about the bounds of beziers, the different ways you can apply bounds transforms to get better results, and all the graphs that I am going to make that don't exist yet
6343 23:00 < Pommers> Oh yeah, did you see the Github pack that I posted in #UCC last night?
6344 23:00 <@matches> Well no I was too busy crying myself to sleep
6345 23:01 <@matches> Thesis' are harder than regular assignments mainly because there isn't a set goal
6346 23:01 <@matches> You have to set your own goal
6347 23:01 < Pommers> https://education.github.com/pack
6348 23:02 <@matches> And it tends to end up as not actually thinking about the goal until the end and then realising that you didn't meet it
6349 23:02 <@matches> Huh what's that
6350 23:03 <@matches> I see many "cloud" and "crowdsourcing" references and "$" signs
6351 23:03 < Pommers> All free stuff for students
6352 23:04 <@matches> Cool
6353 23:04 <@matches> Goodnight
6354 23:04 < Pommers> Night
6355 23:06 <@matches> Hmm, I should probably have removed the Western Australian Marine Science Institution logo from that presentation template I stole
6356 23:07 <@matches> Or it will haunt my dreams
6357 23:07 < Pommers> Probably. Whena re you presenting?
6358 23:10 < sulix> I'm just disappointed that your presentation isn't "CONFIDENTIAL" like some of th others on the list.
6359 23:17 < Pommers> Nice presentation
6360 --- Day changed Thu Oct 09 2014
6361 09:22 <@matches> I'm kind of worried that my live demo takes about 10 minutes to run
6362 09:23 <@matches> Also there are so many things that don't work
6363 09:23 <@matches> Like actually loading an SVG with arbitrary precision
6364 09:24 <@matches> The bottleneck in the live demo seems to be the shear number of objects in it though
6365 09:24 <@matches> Maybe I just need fewer turtles
6366 09:25 <@matches> Also is Rowan in Perth
6367 09:25 <@matches> I think he might be supposed to mark it
6368 09:26 <@matches> Although they listed Tim and not Rowan on the supervisors list
6369 09:26 <@matches> :S(
6370 10:43 < sulix> Rowan is here.
6371 10:50 <@matches> Excellent
6372 10:50 <@matches> "Record yourself presenting as practice" they said "It will make you feel more confident" they said
6373 10:51 <@matches> They never mentioned that it will make you realise just how stupid your voice sounds, just how boring you are at presenting, and just how many things you don't have time to cover
6374 10:51 <@matches> Well that last one I guess they did mention
6375 10:51 <@matches> They didn't mention the crushing despair...
6376 10:59 * sulix is in desperate need for a suitably punny title for his seminar.
6377 11:17 <@matches> I didn't use a pun
6378 11:17 <@matches> I used a very mundane title
6379 11:17 <@matches> Which is probably why I am on last...
6380 11:17 <@matches> Whoops I'm not supposed to give Pommers hints as to when it is, or he might show up
6381 11:22 <@matches> I make such a leap in my presentation
6382 11:22 <@matches> I go from "Look this PDF viewer has a zoom limit of 800%" to "The standards specify IEEE-754 floats"
6383 11:22 <@matches> Quietly ignoring that you really should be able to go further than 800% with IEEE-754 floats
6384 11:22 < sulix> Sounds good!
6385 11:23 <@matches> I guess if someone calls me out I have the backup "Look, I can zoom further but then X crashes"
6386 11:23 <@matches> "What a shame, looks like I can't do the rest of the presentation!"
6387 11:23 <@matches> Also
6388 11:23 <@matches> Where am I supposed to mention the collaborative aspect
6389 11:24 < sulix> Just point at me and go: "the bugs are all his fault".
6390 11:24 <@matches> Currently it is in "References"
6391 11:24 <@matches> - David did things
6392 11:24 <@matches> - Go see David talk about Quadtrees
6393 11:25 < sulix> - He used a goto, get the pitchforks.
6394 11:27 <@matches> "So I'm going to do some live demos now"
6395 11:27 <@matches> "And as you can see, if I zoom in
6396 11:27 <@matches> "Oh fuck SIGFPE"
6397 11:29 <@matches> Also, if you have any time to set up some of the performance tests that would be kind of awesome
6398 11:29 <@matches> Since we both need them, but I appear to need them sooner :P
6399 11:30 <@matches> I guess I can leave them out and just have the spots in the draft where I want them to go, but I'm getting exponentially nervous about them
6400 11:30 <@matches> We can compile with MPFR C++ mpreal now
6401 11:30 <@matches> You can set the precision, but it appears to be fixed
6402 11:30 < sulix> Oooh...
6403 11:31 < sulix> Is it faster that gmprat?
6404 11:31 <@matches> I don't know yet
6405 11:31 <@matches> That's one of the performance tests I want to do
6406 11:32 <@matches> I think mpreal with a fancy algorithm for how you increase precision *might* beat GMP rat
6407 11:32 < sulix> So do I.
6408 11:32 <@matches> The problem is, if you use a fixed precision then you can always choose some view bounds for which things aren't rendered correctly
6409 11:32 <@matches> Also I do not have time to implement a fancy algorithm for how you increase precision...
6410 11:33 <@matches> I feel like I have achieved half of the project
6411 11:33 <@matches> Rather than the whole thing
6413 11:34 <@matches> I do suspect that if you just go "I will use a 1000 bits" for mpreal, then for ranges where you don't need that precision you're wasting a lot of time
6414 11:36 < sulix> You can't do "I'll use n*zoom bits" or something?
6415 11:37 <@matches> I think I would be better off analysing the implementation that works rather than implementing the implementation that might work better...
6416 11:37 <@matches> I guess there's still some time *after* the presentation
6417 11:44 <@matches> sulix: People are getting lunch at the tav because it's Pommer's birthday, you should come
6418 11:45 <@matches> I went in the tav once and didn't die
6419 11:45 <@matches> (Not counting the quiz night, at which I'm pretty sure I did die)
6420 11:45 < sulix> I guess I could eat some lunch.
6421 18:32 <@matches> I have pushed the horrible python/debugscript combo
6422 18:32 <@matches> Featuring totally inconsistent naming conventions in the python scripts
6423 18:36 <@matches> I do recommend you learn how to use python for horrible hacky scripts like this because it is about to save my life
6424 18:38 <@matches> Maybe you can write one that demonstrates the flaw by zooming in on an intersection :P
6425 18:39 <@matches> Hmm
6426 18:39 <@matches> I wonder if the use of "cin" and "stdout" for input to the DebugScript and output from the DebugScript does anything wierd
6427 18:40 <@matches> Conshmistency
6428 18:40 * matches goes into super draft mode
6429 --- Log closed Thu Oct 09 18:40:43 2014
6430 --- Log opened Thu Oct 09 18:41:03 2014
6432 18:41 -!- Irssi: #ipdf: Total of 3 nicks [0 ops, 0 halfops, 0 voices, 3 normal]
6433 18:41 -!- Irssi: Join to #ipdf was synced in 0 secs
6434 18:41 < matches> (You might say something important that I overlooked I guess?)
6435 19:47 < sulix> Okay, so the grid is totally breaking the quadtree.
6436 19:47 < sulix> Because it has lines that coincide _exactly_ with the edge of the quadtree.
6437 19:48 < sulix> So the quadtree doesn't know what nodes to put them in.
6438 19:48 < sulix> Will there be two copies of them? Will there be none?
6439 19:48 < sulix> (Also you broke debugscript reading from a file, but I've fixed it and will push shortly)
6440 19:49 < sulix> (The power of iostreams!)
6441 20:00 < sulix> Okay, so not solving both the x and y cubics for vertical/horizontal lines didn't help.
6442 20:00 < sulix> (On the bright side, it doesn't seem to have broken anything too significant)
6443 20:19 < sulix> Interesting things from Knuth.
6444 20:20 < sulix> - "The origins of floating-point notation can be traced back to Babylonian mathematicians ~1800 BC" (who used bas 60, but didn't explicitly state the exponent)
6445 20:30 < sulix> - Bigints have O(N) addition/subtraction, multiplication/division O(N^2)
6446 20:32 < sulix> - Knuth writes the idea of "positional number systems" as (...a_3 a_2 a_1 a_0 . a_{-1} a_{-2} ...)_b = ... + a_3 b^3 + a_2 b^2 + a_1 b^1 + a_0 + a_{-1} b^{-1} + a_{-2} b^{-2} +...
6447 20:32 < sulix> i.e. neatly using "..." to avoid ever writing infinity.
6448 20:33 < sulix> He also uses the term "radix point" to mean decimal/binary/whatever point.
6449 20:33 < sulix> (and then gives about four pages of history of ancient civilisations counting)
6450 20:35 < sulix> There is then some discussion of negative and complex bases and the different ways of representing negatives.
6451 20:45 < sulix> - The "ulp" is a unit of floating-point error equal to one unit in the last place of the fraction (mantissa) part of a float.
6452 20:45 < sulix> Floating point operations are correct to within half an ulp.
6453 20:46 < sulix> The associative law for multiplication is holds within about 2 ulps.
6454 20:47 < sulix> (Provided no exponent overflow or underflow occurs)
6455 20:47 < matches> Cool
6456 20:47 < matches> Well, I've got a graph showing that mpfr reals with 6 bits of precision are (somewhat unsurprisingly) worse than either doubles or singles
6457 20:47 < matches> Also FPS
6458 20:47 < matches> FPS goes all over the place with everything
6459 20:48 < sulix> ulps are only valid for normalized floats.
6460 20:48 < sulix> Yeah, FPS is a totally terrible unit.
6461 20:49 < matches> SPF might be better
6462 20:49 < sulix> You should try for ms/frame, but even that will probably be a bit broken.
6463 20:50 < matches> FPS is in the code now, it can change later
6464 20:50 < matches> "This wobbly line is slightly lower than this one, indicating WORSE"
6465 20:50 < matches> Ok, what other performance tests did I need
6466 20:51 < matches> Oh yeah, ones that actually alter the view with things in it, as opposed to ones that import an svg at a fixed scale
6467 20:53 < matches> Man explaining how all these tests are actually valid will be a pain
6468 22:34 < matches> I am too tired to work out how to skip the tests that are essentially pointless for various transformation settings
6469 22:35 < matches> I guess they'll just be in the figures as straight lines thus proving my point
6470 22:35 < matches> (Or confusing the figure)
6471 22:35 < matches> The fact that there is a graph should be good enough for a draft
6472 22:36 < matches> Yes, they are straight lines, thus proving my point
6473 22:36 < matches> (Actually they just make the figure confusing)
6474 22:38 < matches> I think what I've actually invented is just a very convoluted way to demonstrate things that everyone has known since Knuth
6475 22:39 < matches> And less of a useful SVG viewer
6476 22:40 < matches> Either that or an epileptic siezure machine
6477 22:41 < matches> So many flashes
6478 22:42 < matches> Hmm it looks like the grid cumulated error is really not as noticable with the grid
6479 22:42 < matches> s/grid//
6480 22:42 < matches> s//grid/
6481 22:42 < matches> skdfjhasdf
6482 22:50 * matches prepares the "hairy fox" hypothesis
6483 22:58 < sulix> No relation to the hairy ball theorem? http://en.wikipedia.org/wiki/Hairy_ball_theorem
6484 23:07 < matches> That is... an unfortunate name
6485 23:08 < matches> Oh, it turns out to produce the hairy fox you need to do a lot of cumulative transforms (dur)
6486 23:08 < matches> Not just two really big transforms
6487 23:08 < matches> So the ulp thing is useful
6488 23:08 < matches> For understanding the rounding errors for small widths
6489 23:10 < matches> The subtraction of x - v_x gives you up to 1 ulp and then the division by 1e-6 increases it to 1e6 ulp and if 1 ulp \approx 5e-5 or more, suddenly it is half your view width
6490 23:10 < matches> Increasing the exponent of (v_x, v_y) will double the ulp
6491 23:10 < matches> So for the same width you get half the precision
6492 23:20 < matches> Hairy fox syndrome appears to work best if you use pxzoom
6493 23:20 < matches> I won't question it...
6494 23:41 < matches> Yeah you have to try really hard to get the hairy fox
6495 23:42 < matches> My greatest fear is that someone will realise to achieve these "errors" by using the mouse (which is what it simulates) it would take you several days of determined scrolling
6496 23:42 < matches> Wow it makes interesting patterns in the grid though
6497 23:43 < matches> It affects both the position and the dimensions of the lines so your grid gradually disappears
6498 23:55 < sulix> pxzoom will have more error because it randomly calls expf() on its input, so you've got some dodgy approximation of 'e' messing things up.
6499 --- Day changed Fri Oct 10 2014
6500 00:16 < matches> Well that would have been nice to know before I ran a 20 minute test
6501 00:16 < matches> I thought it was equivelant to using the mouse
6502 00:17 < matches> Oh well
6503 00:17 < matches> The test still proves the point
6504 00:17 < matches> Even though there's the dodgy approximation of e going on
6505 00:17 < matches> GMP rationals give a very small error
6506 00:17 < matches> Cumulative transforms give a very large error
6507 00:17 < matches> Direct transforms give an increasing error and also have the issue with precision
6508 00:18 < matches> I should probably distinguish between error and precision a bit more
6509 00:18 < matches> We want precision and we don't want error
6510 00:19 < matches> So the GMP rationals applied to Path bounds pass both tests
6511 00:19 < matches> Whereas the other two approaches pass one test and fail the other
6512 00:19 < matches> Or fail both
6513 00:19 < matches> s/two/three/however many I had I lost count
6514 00:45 < matches> New draft is pushed
6515 00:45 < matches> Enjoy the graphs
6516 00:46 < matches> Try not to have too much fun questioning the validity of the results, they are all that's keeping me going
6517 00:46 < matches> I wish path-rat gave no error
6518 00:46 < Pommers> When do you need iPDF working on windows/
6519 00:47 < matches> It's not really a priority
6520 00:48 < matches> I'll probably try and get it working after the final report is due
6521 00:48 < matches> I will link to it anyway see
6522 00:49 < matches> The odds of anyone actually following the link are slim
6523 00:49 < matches> If it doesn't work I'll tell people to run it in wine
6524 09:33 < sulix> Those are damn nice graphs!
6525 09:34 < sulix> I take it there's no 'rat' measurement because it's too slow.
6526 09:35 < sulix> pxzoom is actually the same as the mouse, the mouse still does random expf() calls to work out the zoom amount.
6527 09:35 < sulix> (I thought I'd changed that, but apparently not)
6528 09:37 < sulix> Also, thank god we don't use 'sin': http://randomascii.wordpress.com/2014/10/09/intel-underestimates-error-bounds-by-1-3-quintillion/
6529 09:38 < matches> There should be a rat measurement?
6530 09:38 < matches> In the loss of precision graph, the path-rat program loses no precision
6531 09:39 < matches> But it's underneath all the other implementations that also lose no precision
6532 09:39 < matches> They are all on the same line...
6533 09:39 < matches> In the accumulated error graph, it is down near 10^{-2}
6534 09:39 < matches> I'm not sure why it isn't at zero...
6535 09:39 < matches> I'm just going to point vaguely at the expf function
6536 09:41 < matches> I'll probably add MPFR with 10 bits (half floats...) and X bits where X is a lot
6537 09:41 < matches> But not today
6538 09:41 < matches> My draft fleshes out a lot more when I uncomment things...
6539 09:41 < matches> I mean the idea of commenting them was that they'd be removed
6540 09:42 < matches> Also there are two sources of error in the Direct transformation
6541 09:42 < matches> I need to think about whether they are both equally bad or one is worse
6542 09:42 < matches> You need to apply the inverse view transformation to express the svg in document space
6543 09:43 < matches> *Then* you apply the view transformation to render it
6544 09:43 < matches> Either of those could be the one that breaks it, but I suspect the rendering transformation is worse
6545 09:46 < matches> Maybe some of the ipython figures don't have path-rat in them, I forget
6546 11:40 * sulix sighs.
6547 11:40 < sulix> cssubmit requires my abstract to be in ASCII, so the é in Bézier means cssubmit rejects it.
6548 11:51 < matches> Haha
6549 11:51 < matches> I got excited when Rowan said "Looks great to me" and then I realised he was talking about your abstract not my draft... :P
6550 11:52 < matches> That would have been impressive speed reading
6551 11:52 < matches> My abstract is sort of really terrible
6552 11:52 < matches> It clings to the whole "document formats" thing that the project was alledgedly about but has moved well away from
6553 11:53 < matches> It also promises that I'll compare arbitrary precision floats as well, so I guess I better hack something together with those...
6554 11:53 < matches> Maybe a graph showing how you can increase the precision but then the FPS goes down
6555 11:54 < sulix> That'd ve really good, actually.
6556 11:55 < matches> Yeah I think I did something similar for CQM once
6557 12:06 < matches> Hmm, I can probably hack together a CPU renderer only binary for windows
6558 12:06 < matches> He says, ignoring the tendency for the CPU renderer to shit itself and die
6559 12:07 < matches> It'd be a Cool Feature for it to fall back to the CPU renderer instead of just segfaulting
6560 12:08 < matches> Falling back to the backup renderer always sounds impressive
6561 12:08 < matches> Everyone loved it when HTR did that
6562 12:08 < matches> (And then I realised the "backup" renderer was actually better)
6563 12:17 < sulix> I've fixed (hopefully) the OpenGL version detection. It should now Fatal() when the driver doesn't have needed extensions.
6564 13:43 < sulix> Excellent: the é has gone through!
6565 --- Day changed Sat Oct 11 2014
6566 10:02 * sulix -> Perth Games Festival
6567 14:52 < Pommers> Is it good there?
6568 18:22 < matches> The "fix" to OpenGL version detection now means I don't have a valid version of OpenGL.
6569 18:23 < matches> It was working perfectly fine before
6570 18:24 * matches -> s/Fatal/Error/ and hopes it still works anyway
6571 18:25 < matches> Should you be setting the GL minor version to 1 and then failing if it isn't 3
6572 18:26 < matches> I am confused how this even works
6573 18:26 < matches> My OpenGL version is apparently 3.1 and the if statement apparently checks "IsVersion" 3.2
6574 18:26 < matches> And then says "We want 3.3"
6575 18:28 < matches> I get no error if I set the minor version attribute to 3
6576 18:28 < matches> So I take it we required 3.3 functions but were specifying that we only needed 3.1 or something
6577 18:28 < matches> And somehow the 3.3 functions still worked?
6578 18:29 < matches> Anyway, I'm in favour of the "Instead of terminating the program, report an error and then see whether it will crash anyway" approach
6579 18:29 < matches> If it *does* crash, then at least we warned them
6580 18:29 < matches> If it doesn't crash, hey it's running! Everyone wins!
6581 18:43 < matches> Using integer indices for different types of Real seemed like *such* a good idea at the time
6582 18:43 < matches> And then there was the ability to have up to three different types of Real...
6583 18:45 < matches> -std=c++11 -g -Wall -Werror -Wshadow -pedantic -rdynamic -I../contrib/include/SDL2 -I`pwd` -I/usr/share/qt4/mkspecs/linux-g++-64 -I. -I/usr/include/qt4/QtCore -I/usr/include/qt4/QtGui -I/usr/include/qt4 -I. -Itests -I. -DQT_NO_DEBUG -DQT_GUI_LIB -DQT_CORE_LIB -DREALTYPE=1 -DQUADTREE_DISABLED -DQT_NO_DEBUG -DQT_GUI_LIB -DQT_CORE_LIB -DTRANSFORM_OBJECTS_NOT_VIEW -DTRANSFORM_BEZIERS_TO_PATH -DPATHREAL=Gmprat -c -MMD
6584 18:45 < matches> Truly our Makefile is a thing of beauty
6585 18:47 < matches> It's so beautiful that there is now a Python function to invoke it with different arguments
6586 18:57 * matches stabs indentation errors
6587 18:59 < matches> "Unindent does not match any outer indentation level" What the fuck
6588 18:59 < matches> What the hell is an undindent
6589 18:59 < matches> It is pointing at a ':' character
6591 19:00 < matches> It's because I copied it from ipython which magically converts your tabs to spaces
6592 19:00 < matches> And then tried to add things to it indenting with tabs
6593 19:01 < matches> I think it's a conspiracy to make the tabs vs spaces war even worse
6594 19:02 < matches> Now people have no option to compromise
6595 19:02 < matches> Or, in my case, use tabs where we bloody well typed tabs in the first place
6596 19:35 < matches> My clock says it's 3am, that's about how I feel...
6597 19:42 < matches> "Just use MPFR" is looking more and more likely to be the conclusion
6598 19:42 < matches> Wait...
6599 19:42 * matches eats hat
6600 19:43 < matches> Well, GMP rationals applied to path bounds is faster than mpfr with 1024 bits applied directly
6601 19:43 < matches> That's probably not a fair test though
6603 --- Day changed Sun Oct 12 2014
6604 14:53 < matches> Setting up the performance tests is a real pain if you want to actually include the really slow methods
6605 14:53 < matches> Well, restarting them can be painful
6606 14:54 < matches> Which means if you screw them up...
6607 14:58 < matches> Placing a limit of 30s per frame was perhaps a bit too generous
6608 15:05 < Pommers> THis is why you should use a build server
6609 15:05 < matches> Yeah except that's more effort to set up than it would save
6610 15:05 < matches> We even have one
6611 15:06 < matches> Motsugo could potentially run all these tests
6612 15:06 < matches> But it has to use a CPU renderer and is therefore shit
6613 15:06 < matches> Seeing the results in real time is a huge advantage
6614 15:06 * Pommers really needs to set up a proper Jenkins set up for UCC
6615 15:06 < matches> Oh, I'm not allowed to use my own laptop in the conference
6616 15:07 < matches> So if you want to fix the MinGW build for me... :P
6617 15:07 < matches> I'll see if sulix's magical GL version changes fixed anything later
6619 15:08 < Pommers> I can give it a shot tonight. If you proof read some stuff for me right now
6620 15:08 < matches> I'm querying the FPS once the view has stopped being scaled
6622 15:09 < matches> (I should record performance in the python script I think)
6623 15:09 < Pommers> tonight means after I finish work that is
6624 15:09 < matches> So 3am?
6625 15:09 < Pommers> Nah, like 6.30ish
6626 15:09 < matches> Honestly I'm probably just going to run a video
6627 15:10 < matches> But it'd be nice to say "Here's the binary with which you can reproduce the results"
6628 15:18 < matches> It's probably going to need to run on a Windows 95 presentation PC or something
6629 15:19 < matches> It has geometry shaders so it's probably beyond hope that it would work
6630 15:19 < matches> Hmm, do I need to prepare a response do a "Why didn't you just use <insert name of a language that isn't C++ here>"
6631 15:20 < matches> I seem to get that question every time I use C++ for anything
6632 15:20 < matches> Or alternately "Why didn't you use boost"
6633 16:19 < matches> sulix: What black magic should I do if I want to include the debug font in a screenshot...
6634 16:45 * matches resorts to ffmpeg instead
6635 16:45 < matches> With the disadvantage that 1 frame in a screen grap != 1 frame in the demo :(
6636 16:45 < matches> grap
6637 16:45 < matches> grape
6638 16:45 < matches> grab
6639 --- Day changed Tue Oct 14 2014
6642 00:17 -!- bobgeorge33 is now known as Pommers
6645 13:34 < matches> My presentation is tonight
6646 13:34 < matches> Looks like Tim is back just in time to watch it!
6647 14:50 < sulix> In order to break things at exactly the wrong moment, IPDF has a profiler now.
6648 14:50 < sulix> It dumps the time spent in various functions at the end of each frame.
6649 15:10 < matches> That's pretty awesome
6650 15:11 < matches> Ok, time to test this pdf in adobe reader...
6651 15:11 < matches> And scream if it doesn't work
6652 15:19 * matches screams
6653 15:20 < sulix> Oh dear...
6654 15:20 < sulix> The power of adobe!
6655 15:24 < matches> I bet the videos don't work either
6656 15:26 < sulix> What format are they in?
6657 15:26 < matches> mkv
6658 15:26 < matches> They work
6659 15:26 < matches> I guess it was too much to expect pdflatex to produce a valid pdf
6660 15:35 < Pommers> Latex PDF? Got it working yet?
6662 15:56 < matches> I have narrowed it down to
6663 15:56 < matches> "Cannot include graphics in slides"
6664 15:56 < matches> Which will make for an amazing presentation
6666 15:58 < matches> Adobe doesn't support including pdfs within pdfs
6667 15:58 < sulix> What!?
6668 15:59 < sulix> Man, I've included pdfs within pdfs in pretty much every assignment I've submitted ever.
6669 15:59 < matches> I don't know
6670 15:59 < matches> That's the first place it shits itself
6671 15:59 * sulix ponders whether or not this may have affected marks.
6672 15:59 < matches> Debugging by repeatedly commentint out bits of TeX and recompiling is not very efficient and/or fun
6673 16:00 < sulix> Worst comes to the worst, take screenshots of all of your slides, I guess...
6674 16:00 < sulix> (Or download a windows binary for evince/etc and put that on the thumb drive)
6675 16:02 < matches> Yeah I'm supposed to be registering now
6676 16:25 < Pommers> matches: Send me the tex source and I'll try
6677 --- Day changed Wed Oct 15 2014
6678 10:38 < matches> Frames compiled it for me on Windows, so I survived
6679 10:40 < Pommers> Yeah. That's what I was going to do
6680 10:40 * Pommers should really set up that build server properly
6681 13:18 < matches> I might have promised a Windows version in the talk...
6682 13:18 < matches> Also saving, loading
6683 13:18 < matches> And...
6684 13:18 < matches> Shading
6685 16:05 < matches> I wonder if I can hack in a "draw" mode...
6686 16:05 < matches> Now that the conference is over, ANYTHING IS POSSIBLE
6687 16:06 < matches> (Except there's only two weeks left)
6688 16:06 < matches> ... < 2 weeks left
6689 --- Day changed Sun Oct 19 2014
6690 14:48 < sulix> Far out: I'm looking up some IEEE GPU stuff, and Google keeps pointing me to your lit review.
6692 17:02 < matches> I need to muster the courage to write more of my report
6693 17:02 < matches> So tired
6694 --- Day changed Mon Oct 20 2014
6695 22:45 < sulix> So it turns out that the crazy multithreadedness of the control panel totally bollockses up the profiler.
6696 22:46 < sulix> Which explains the random crashing due to "impossible things happening" like "functions returning before their parents do".
6697 22:49 < sulix> The everything gets bollocksed when adding text bug is also fixed.
6698 22:50 < sulix> The magical invisible turtles bug remains.
6699 22:51 * sulix wonders if he should see the teenage mutant ninja turtles movie for an explaination.
6700 23:48 < sulix> For reasons unknown, I'm generating a lot of quadtree overlays with the exact same objects in them.
6701 23:48 < sulix> By which I mean the same {object_begin, object_end} pairs.
6702 23:49 < sulix> So lots of things are being rendered ~40 times, and many things aren't being rendered at all.
6703 23:51 < sulix> Actually, never mind, I'm just printing out the wrong thing in my debug statement.
6704 --- Day changed Tue Oct 21 2014
6705 00:48 * sulix has a lot of writing to do tommorrow.
6706 00:48 < sulix> By which I mean today.
6707 09:02 * sulix ->sainity--
6708 09:30 < Pommers> Morning
6710 11:11 * sulix has engaged "panic mode"
6711 11:12 < Pommers> Can you try and shake it off? https://www.youtube.com/watch?v=nfWlot6h_JM
6712 11:12 < sulix> Also "holy shit, I don't have any results other than SIGSEGV" mode.
6713 11:20 < matches> I wish I hadn't read the comments on that video
6714 11:24 < matches> I'm having trouble understanding the metaphor
6715 11:24 < matches> Is Taylor Swift meant to be the thesis or me
6716 11:25 < matches> And how is the illuminati involved
6717 11:28 < Pommers> Never read YT comments
6718 11:28 < Pommers> Metaphor? Th[4~esis? Illuminati?
6719 11:34 < matches> "Haters gonna hate" would be me hating the thesis, but the thesis being an inanimate and currently purely digital object would be unable to shake itself
6720 11:35 < matches> According to the comments Taylor Swift is in the illuminati
6721 11:35 < sulix> The only shaking I can imagine is the solemn shaking of heads.
6724 11:36 < Pommers> Does your thesis hatre you, hence it's not happening
6725 11:36 < Pommers> and therefore you need to shake it off?
6726 11:36 < matches> Perhaps the clue is in the name, in that we need to Swiftly apply Taylor series to things
6727 11:38 < sulix> matches: That is the only consistent interpretation.
6728 11:38 < matches> I regret catching up on 6 weeks of lectures instead of working on the thesis
6729 11:40 < matches> http://www.irregularwebcomic.net/374.html
6730 11:40 < sulix> http://phdcomics.com/comics/archive.php?comicid=1749
6731 11:42 < matches> Back to it I guess
6732 11:42 < matches> By which I mean back to spending 30 minutes mentally preparing myself to go back to it
6733 11:43 < Pommers> While you relax
6734 11:43 < Pommers> You could fix NTP on your laptop
6735 11:46 < sulix> better still, set the time to three days ago so that you have more time to work on the thesis.
6736 11:47 < Pommers> Thats's a good idea! Just avoid anything that relies on time
6737 11:49 < sulix> The number of sudden thoughts about much better ways to do this project is slightly terrifying.
6738 12:44 < sulix> /me suspects he's deleted more than he's written.
6739 13:02 < sulix> Someone's looked at quadtrees for "infinite bitmaps": http://programmers.stackexchange.com/questions/104609/infinite-bitmap
6740 13:44 < sulix> Knuth apparently does not like the term "mantissa". He has quite a rant about it and now I'm feeling guilty for using it.
6741 13:47 < matches> Let's face it, Knuth won't read our work, and if he does, he won't like it
6742 13:48 < sulix> I had a nightmare where Kahan was berating me for not understanding floats.
6743 13:49 < matches> You don't even need to write about floats? Do you?
6744 13:49 < matches> Also, should I be urgently producing graphs of how slow rationals are
6745 13:50 < matches> I have 6662 words, but this is really not good
6746 13:51 < sulix> I'm sort-of leaving the "results" section blank at the moment.
6747 13:51 < matches> Wait but what else do you need?
6748 13:51 < matches> Lit Review should be done*
6749 13:52 < matches> There is a "Methods" chapter which I'm a bit embarressed about
6750 13:52 < sulix> Also, I should push the "speed up everything because it was dodgy renderrange code that was taking >90% the time" fix at some point.
6751 13:52 < matches> How badly would pulling break everything...
6752 13:52 < sulix> I've been writing up how the quadtree works, and fixing up the lit review, intro, gpu float bits, etc.
6753 13:53 < sulix> Well, I don't think I've been brave enough to push yet, but probably it will break everything.
6754 13:53 < sulix> Though, with some small amount of luck, it should still compile.
6755 13:56 * sulix has pushed. Time to warm up your "git revert" commands.
6756 13:57 < sulix> Also I need to write up about how rationals can represent more things than the quadtree, which can only represent binary fractions.
6757 14:18 < matches> "A naive approach would be to simply replace all floating point operations with arbitrary precision operations, and this is in fact what we did before we realised how stupid that was"
6758 14:21 < sulix> I was going for "A naïve approach would be to simply replace all floating point operations with arbitrary precision operations, and unlike the quatreethulu, that does actually work."
6759 14:21 < matches> How do you get the ï in LaTeX
6761 14:22 < matches> Thankyou
6762 14:25 < matches> Is there a good reference I can cite for how shaders work...
6763 14:25 < matches> I really don't want to or have the space to explain the OpenGL API
6764 14:25 < sulix> The OpenGL spec?
6765 14:25 < matches> I guess
6766 14:25 < sulix> Or ARB_vertex_program / ARB_shader_program
6767 14:25 < sulix> s/shader_program/fragment_program
6768 14:47 < matches> You know I am linking to the git repository in my report
6769 14:47 < matches> The one with all the "My eyes, they burn!" commit messages...
6770 14:48 < matches> I figured that the markers assuming there wasn't version control was marginally worse than them reading my commit messages
6771 14:48 < matches> But only marginally
6772 14:49 * sulix wipes some nervous sweat from his brow.
6773 14:50 < Pommers> What did you do for commit messages?
6774 14:50 < matches> Usually stuff to the effect of "This code is awful"
6775 14:50 < matches> "I didn't even design this!"
6776 14:50 < matches> Etc
6777 14:51 < sulix> Pommers: cover your eyes: http://git.ucc.asn.au/?p=ipdf/code.git;a=shortlog
6778 14:52 < sulix> My favourite is "A Song of Floodfills and Segfaults"
6779 14:52 < Pommers> haha
6780 15:11 < sulix> matches: Any idea why "bezier.h:61:20: error: call of overloaded ‘Abs(Gmprat)’ is ambiguous"
6781 15:13 < sulix> Also: "rect.h:21:13: error: no match for ‘operator<<’ (operand types are ‘std::basic_ostream<char>’ and ‘const IPDF::Rational<IPDF::Arbint>’)" if I try Rational<T>
6782 15:37 < matches> Aaaah
6783 15:37 < matches> It wasn't ambiguous when I last used it!
6784 16:03 * sulix squirms nervously.
6785 16:06 < matches> I finally made the performance test graphs
6786 16:06 < matches> They are kind of dodgy
6787 16:07 < sulix> Ooh... show me.
6788 16:13 < matches> http://szmoore.net/ipdf/sam/figures/time.pdf
6789 16:13 < matches> http://szmoore.net/ipdf/sam/figures/memory.pdf
6790 16:13 < matches> It's using eye_of_the_rabbit.script
6791 16:14 < matches> It keeps track of how much memory is used for the View width
6792 16:14 < matches> And how long it takes to do each cycle that involves adding 3 rabbits seperated by 1e-3 or whatever it was
6793 16:14 < matches> A totally arbitrary test but hey, the graph is linear
6794 16:15 < matches> I'm pretty fed up with this project at this point
6795 16:15 < matches> Words are on the page
6796 16:16 < matches> It's amazing how much less you have to write when you remove the "TODO: I really should explain this properly"
6797 16:19 < sulix> Those are with the Gmprat for path bounds, doubles? for beziers?
6798 16:21 < matches> Gmprat for path bounds
6799 16:21 < matches> Not really much point doing it for anything else
6800 16:21 < matches> Since anything else won't render the images correctly
6801 16:21 < matches> I guess I should do it for doubles as well
6802 16:22 < matches> Actually doubles would be able to render that particular script correctly
6803 16:22 < matches> For a long time
6804 16:23 < sulix> The quadtree can render it correctly until there is more than one rabbit.
6805 16:23 < sulix> Also eye_of_the_rabbit is cpu rendered by default, no?
6806 16:24 < matches> Yeah I changed it to make that test
6807 16:24 < sulix> Okay.
6808 16:24 < matches> I'll run it for mpfr-1024 path
6809 16:24 < matches> I've had enough of having to tweak things to recompile the software...
6810 16:25 < matches> So I'm limited to the binaries I already have and they have a habit of sigfpe'ing
6811 16:26 < matches> Oh, it was also done with the window hidden
6812 16:26 < sulix> By choice or by fglrx?
6813 16:26 < matches> By choice
6814 16:31 < matches> Urgh this project is dumb
6815 16:31 < matches> I really don't want to have to repeat my final year project :S
6816 17:17 < matches> How do you like being referred to as "Gow"
6817 17:17 < matches> Also it's been an hour or so without me saying this project is silly
6818 17:17 < matches> So I'll say it again
6819 17:18 < matches> I ended up putting the Circles on CPU/GPU in my "Lit Review" part...
6820 17:18 < matches> Because it just fit a lot better underneath the section that was actually talking about it
6821 17:18 < matches> Rather than in the results
6822 17:19 < matches> "By the way we did some circles too"
6823 17:19 < matches> I feel like I should say something snarky about the naive method not being worthy of publication...
6824 17:19 < matches> That might be a bad idea
6825 17:29 < matches> "Hi all, No doubt you are busy preparing your final report..."
6826 17:29 < matches> AAAAAHHH
6827 17:30 < matches> I don't know, there's only so much you can polish a turd..
6828 17:30 < matches> Holy shit how do I have 40 references
6829 17:30 < matches> I guess there was that "Look I read about GPUs [10 references]" paragraph
6830 17:31 < matches> Should I mention the arbitrary precision on GPU stuff
6831 17:31 < matches> Blargh
6832 17:41 < matches> "The main body of the report should be between 6,000 and 8,000
6833 17:41 < matches> words excluding appendices. (i.e. from the beginning of the
6834 17:41 < matches> Introduction to the end of the Conclusions/Future work section)."
6835 17:41 < matches> Finally found the word limit
6836 17:41 < matches> Dammit
6837 17:41 < matches> Here I was aiming for 6000 maximum...
6838 17:44 < matches> I can stick in the pointless sections about "Pixels or Perish" I guess
6839 17:44 < matches> Distract them with pretty pictures
6840 17:47 < matches> Takes me up to 46 references...
6841 17:47 < matches> And 6972 words
6842 17:47 < matches> Done
6843 17:48 < matches> Submit
6844 18:21 < sulix> I think your thesis is significantly better than mine.
6845 18:22 < sulix> Mostly because it has more results and less "TODO"
6846 18:26 * sulix seeps into a 20 minute come.
6847 18:26 < sulix> s/come/coma
6848 18:26 < sulix> (The state of my brain is becoming more and more evident)
6849 18:55 < matches> (The trick is to delete the TODOs)
6850 --- Day changed Wed Oct 22 2014
6851 12:50 < matches> ECM sent me an email about 2015 Final Year Project preferences...
6853 12:50 < matches> I already did this!
6854 12:50 < matches> They also sent me an email about submitting vacation work reports
6855 12:50 < matches> I didn't do that yet!
6856 12:51 < matches> This is what you get for totally ignoring the recommended degree progression
6857 17:17 < Pommers> Guess who has card access through the main door for CS xD
6858 17:18 * sulix is filled with a jealous rage.
6859 17:20 < Pommers> hehehehe
6860 17:20 < Pommers> Pretty sure I now have more Access to CS than both of you :D
6861 17:54 < matches> They'll probably be glad they get to revoke our access at the end of the year
6862 17:54 < matches> Doing other assignments is a nice distraction from the project although in terms of time investment it is not wise
6863 17:54 < matches> In terms of time investment the correct approach is to fail everything else
6864 17:55 < matches> Although that didn't work so well last semester
6865 17:55 < matches> (I didn't actually fail but my marks got consistently lower :(
6866 17:55 < matches> I don't know how to bracket that
6867 17:55 < matches> :()
6868 --- Day changed Thu Oct 23 2014
6869 18:12 < matches> My report scores a Flesch Reading Ease of 42.30
6870 18:12 < matches> Which is higher than "university graduate" and lower than "13 to 15 year old students"
6871 18:13 < matches> It scores a Flesch-Kincaid Grade of 12.44
6872 18:13 < matches> I guess that means it is high school level...
6873 18:13 < matches> That isn't as reassuring as I'd hoped...
6874 18:14 < matches> At least I'm beating "an average 6th grade assignment"
6875 18:17 < matches> My physics thesis is 39.90 and 12.99 indicating my writing has gotten dumber, as expected
6876 18:19 < matches> Yours is at 46.06 and 11.39
6877 18:19 < matches> The question is, what should we be aiming for...
6878 18:21 < matches> That was an interesting but not particularly productive diversion into readability of documents
6879 18:21 < matches> (I think the most this test can conclusively say is that we did indeed write some words)
6880 18:21 < matches> (Many of which had more than one syllable)
6881 19:40 * sulix shall have words with "an arbitrary number of syllables" in his thesis.
6882 19:41 < sulix> Also, it turns out printperf has a big #ifdef QUADTREE_DISABLED around it.
6883 19:41 < sulix> Which explains much.
6884 19:59 < sulix> Did you know: when rendering with GMPrat, sometimes the debug text displays the bounds as being totally wrong for a frame, and then they're back to normal.
6885 20:00 * sulix has found something even more fun than watching paint dry: watching gmprat render.
6886 20:16 * sulix now has a graph: http://davidgow.net/stuff/ipdf-gmpqttest1.pdf
6887 20:54 < matches> Cool
6888 21:16 < matches> That looks very similar to the turtles graph I just got
6889 21:16 < matches> Except qtree isn't on it
6890 21:16 < matches> Path-gmprat is my favourite
6891 21:29 < matches> Ok, I know it's only 9:28 but my clock says 5:28
6892 21:29 < matches> So that counts as bed time
6893 21:30 * matches tries not to think about the asignment due tomorrow or the 3 weeks of assignments that need to be marked before students start to hate me
6894 21:30 < matches> Or, for that matter, how he is going to explain Quantum Mechanics...
6895 21:31 < matches> Or CoderDojo...
6896 21:31 < matches> I must be getting old
6897 21:31 < matches> Back in 2012 I could work on my thesis until 5am and still be awake enough to do my other assignments :S
6898 21:32 < matches> That might be a slight exagerration, I think there was a large amount of feeling like death involved
6899 21:36 < matches> I used to know the quantum number selection rules off by heart too...
6900 21:41 < matches> Ahh, the amazing HTML formatting of the quantum mechanics webpage
6901 21:41 < matches> The memories
6902 --- Day changed Sat Oct 25 2014
6903 11:26 * sulix will not be able to make coderdojo today, for obvious reasons.
6904 11:26 < matches> Ok, good luck
6906 --- Log opened Sun Oct 26 14:54:49 2014
6908 14:54 -!- Irssi: #ipdf: Total of 4 nicks [1 ops, 0 halfops, 0 voices, 3 normal]
6909 14:54 -!- Irssi: Join to #ipdf was synced in 3 secs
6910 14:55 < matches> Very much in the "The referencing doesn't produce ?? submit it" phase
6911 14:57 < matches> Ah crap the referencing still produces ??
6912 14:57 < matches> But only in the appendix...
6913 15:50 < matches> I get to submit this to a "Printerface"
6914 15:50 < matches> Oh ho ho
6915 15:53 * sulix has had the joy of having to redo his data with vblank off.
6916 15:54 < sulix> Also, the random flickering bug is actually showing up in the graphs
6917 16:02 < matches> Yuk
6918 16:03 < matches> So you have loads of data now though?
6919 16:03 < matches> I had vague plans of redoing graphs and then apathy happened
6920 16:03 < sulix> Yeah, lots and lots of graphs and numbers.
6921 16:03 < matches> Cool
6922 16:04 < matches> I just have like, 3...
6923 16:04 < sulix> I extended the profiler to also output the number of objects in the document and rendered
6924 16:04 < sulix> (Well, I only have 3 interesting graphs)
6925 16:05 < sulix> The quadtree spens 23.7% of its time generating new nodes.
6926 16:05 < sulix> (In case you cared)
6927 16:05 < matches> Only 23.7%?
6929 16:06 < sulix> It generates at total (during the zoom-in-on-grid test for 800 frames) of 219 nodes.
6930 16:06 < sulix> And clips a total of 6559 objects
6931 16:07 < matches> I'll email what I have.
6932 16:07 < matches> Prepare to be whelmed.
6933 16:07 < matches> Oh dear I just quoted *that* movie
6934 16:09 < sulix> I'll email mine when I've finished regenerating all of the graphs and stats with non-vblanked data.
6935 16:10 < sulix> (The quadtree is now officially 2127 times faster than naïve use of GMP rationals)
6936 16:12 < matches> I think naive use of GMP rationals is a slightly unfair benchmark... :P
6937 16:12 < matches> But it'll make for impressive graphs
6938 16:12 < sulix> The requested URL /sam/ipdf/thesis.pdf was not found on this server.
6939 16:13 < matches> /ipdf/sam/thesis.pdf
6940 16:13 < matches> Also attached it to the email
6941 16:13 -!- Irssi: #ipdf: Total of 4 nicks [1 ops, 0 halfops, 0 voices, 3 normal]
6942 16:13 < sulix> Also, "Wintrhop Professor".
6943 16:13 < matches> Oh boy
6944 16:14 < matches> Wait where does it even say that...
6946 16:14 < matches> The damn letter
6947 16:19 < matches> Yeah going by the rubric I think I'll have a solid 1/4 average
6948 17:05 < sulix> Again, the closer this project is to being due, the more interesting mathematical things I'm finding that I wish I'd seen earlier.
6949 17:06 * sulix slips a small reference to the Riemann Zeta function into a footnote.
6950 17:23 < matches> I'm concerned about the lack of interesting mathematical things in my thesis, footnote or otherwise
6951 17:24 < matches> I think I'm allowed to call it a thesis since that's what I was supposed to put on the "letter"
6952 17:25 < matches> You can't diss this ertation
6953 17:25 < matches> That worked in my head
6954 17:25 < matches> synonyms..
6955 17:25 < matches> monograph
6956 17:26 < matches> disquisition
6957 17:26 < matches> disquietening
6958 17:26 < matches> composition
6959 17:26 <@bobgeorge33> When does Tim officailly start work again?
6960 17:26 < matches> in the key of F# minor
6962 17:26 < matches> Next year?
6963 17:27 < matches> Someone has to write a "supervisor report"
6964 17:29 < matches> Well, I at least have lots of words
6965 17:29 < matches> Too many to proof read
6966 18:07 < matches> It has been submitted to the Printer Face
6967 18:07 < matches> Good luck sulix
6968 18:10 * sulix solemnly nods: "Good luck to us all!"
6969 23:34 < sulix> It is done: http://davidgow.net/stuff/DavidFYPThesis.pdf
6970 23:38 <@bobgeorge33> When are they due?
6971 23:42 <@bobgeorge33> Lastly, as a double degree student I need to
6972 23:42 <@bobgeorge33> express a double degree of thanks to my friends and family for their continued patience whilst I
6973 23:42 <@bobgeorge33> completed my second “final year” project in what is technically a penultimate year.
6974 23:42 <@bobgeorge33> Very nice
6975 --- Day changed Mon Oct 27 2014
6976 08:46 < matches> They're due at 12:00pm
6977 09:00 < matches> Hmm, there are only a few highly wrong statements in my thesis
6978 09:01 < matches> To resubmit or not to resubmit
6979 09:03 < matches> I keep thinking I really need more detail and then I realise it's 9000 words...
6980 09:03 < matches> What did I even say in those 9000 words I don't even
6981 09:04 < matches> In other news it's 9:04 and no one is at my tutorial...
6982 09:05 < matches> I was going to show them CQM animations
6983 09:09 < matches> Curses a student
6984 09:09 < matches> (I don't know what I want)
6985 10:15 * sulix resubmits to fix the typo in the latin reference that Tim disconvered.
6986 10:16 < sulix> I've also added the original proposal as an Appendix because apparently I'm meant to, though it does sort-of highlight how little of it I've actually done.
6987 10:19 * sulix proceeds to collapse from exhaustion.
6988 10:44 -!- Netsplit mussel.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @bobgeorge33
6989 10:45 -!- Netsplit over, joins: @bobgeorge33
6990 11:11 -!- Netsplit mussel.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @bobgeorge33
6991 11:12 < matches> Aaaand
6992 11:12 < matches> LMS is being LMS
6993 11:13 -!- Netsplit over, joins: @bobgeorge33
6994 11:13 < matches> UWA is having problems?
6995 11:13 < matches> Trying to resubmit to Printer Face
6996 11:13 < matches> Not sure if I am now going to end up with two copies...
6997 11:14 < matches> The second copy is slightly less vague
6998 11:14 < matches> I guess people can work out that they need to read IEEE to get the full gory details of the encodings and stuff
6999 11:15 < matches> Also people can probably work out the legends on my graphs if they think
7000 11:15 < matches> Also there is like, wierd functional programming notation in Tim and Rowan's emails
7001 11:15 < matches> Is it a secret code?
7002 11:23 < matches> AAAH SHIT
7003 11:23 < matches> My last commit is going to be at 7:23pm
7004 11:23 < matches> I'll get in trouble for using a time machine to submit it before 12:00pm
7005 11:40 -!- Netsplit mussel.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @bobgeorge33
7006 11:42 -!- Netsplit over, joins: @bobgeorge33
7007 11:42 < matches> Well, that as anticlimactic
7008 11:43 < matches> The whole having 3 other assignments due this week thing sort of ruins it
7009 11:51 -!- Netsplit mussel.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @bobgeorge33
7010 11:51 -!- Netsplit over, joins: @bobgeorge33
7011 11:53 < matches> I actually really don't trust someone else to print my assignment :S
7012 11:53 < matches> Who is this Printer and why does it have a Face
7013 11:54 < matches> What if the inline postscript gets interpreted by the driver as actual postscript and it just prints 40 pages of shape.eps
7014 11:54 < matches> Aaaah
7015 11:56 <@bobgeorge33> What happens if it decides to print 100 copies instead of 1 and they aren't collated!
7016 11:59 * matches stabs
7017 12:01 < matches> sulix: "Programmme"
7018 12:01 < matches> Does it have three mmm like that
7019 12:01 < sulix> Um... it does now.
7020 12:02 < matches> I approve
7021 12:02 < sulix> I am a programmmmmmer
7022 12:03 -!- Netsplit mussel.ucc.au.uniirc.com <-> arctic.uniirc.com quits: @bobgeorge33
7024 18:38 -!- bobgeorge33 is now known as Pommers
7025 --- Day changed Thu Oct 30 2014
7026 13:45 < matches> sulix: Is there going to be a meeting about your presentation? Should I come, or stay away?
7027 14:16 < sulix> I don't think there's going to be a meeting... I've got another research project due at 4:00, so I'll be doing that.
7028 14:16 < sulix> The seminar schedule is out now, though: http://undergraduate.csse.uwa.edu.au/year4/Current/Project/seminarTimetable.html
7029 14:17 < sulix> It's hysterically empty.
7030 14:28 < matches> Baha
7031 14:29 < matches> 25 minutes, ew
7032 14:29 < matches> Good luck
7033 14:29 < sulix> You're welcome to come along to the talk if you hate sleep.
7034 14:29 < matches> I will try and come
7035 14:30 < matches> "Time travel debugging"...
7036 14:30 < sulix> Yeah, Ben's a genius.
7037 14:30 < matches> I do that all the time though
7038 14:30 < matches> (With printf statements)
7039 14:30 < sulix> But you live in the future.
7040 14:30 < matches> (And scrolling through the output with less)
7041 14:30 < matches> That too
7042 14:31 < matches> Don't worry, your abstract sounds really cool too
7043 --- Day changed Sat Nov 01 2014
7044 10:27 < matches> I guess we're done here, good job
7045 --- Day changed Tue Nov 04 2014
7046 14:30 < matches> We will leave this channel here until the marks come out
7047 --- Day changed Sat Nov 08 2014
7048 23:00 -!- Pommers is now known as bobgeorge33
7049 23:00 -!- bobgeorge33 is now known as Pommers
7050 --- Day changed Tue Nov 11 2014
7052 --- Day changed Wed Nov 12 2014