Config
Log for #openttdcoop.devzone on 19th June 2014:
Times are UTC Toggle Colours
00:02:17  *** yorick has quit IRC
08:13:31  <Ammler> planetmaker: good morning, another package maintainer added grfcodec to opengfx, is that needed now?
08:14:45  <Ammler> grfid, hmm.
08:15:16  <Ammler> my issue is that grfcodec is not part of factory (suse main repo)
08:20:38  *** Supercheese has quit IRC
08:27:15  <planetmaker> yes, grfid is essential, I'm afraid
08:27:30  <planetmaker> good morning also :)
08:27:46  <planetmaker> thus the dep on grfcodec package is a hard one
08:28:05  <planetmaker> gentoo(?) obviously failed to do so and provide(d) broken OpenGFX
09:53:08  *** frosch123 has joined #openttdcoop.devzone
11:28:41  *** gelignite has joined #openttdcoop.devzone
12:35:08  *** yorick has joined #openttdcoop.devzone
14:42:28  *** LSky` has joined #openttdcoop.devzone
14:54:41  *** Alberth has joined #openttdcoop.devzone
15:00:00  <Ammler> I wonder what would be easier, adding grfid to nml or submitting grfcodec to suse main repo
15:03:50  <planetmaker> it doesn't make sense to add grfcodec repo to nml
15:04:00  <Rubidium> Ammler: do it the archlinux way
15:04:59  <Rubidium> https://projects.archlinux.org/svntogit/community.git/tree/trunk/PKGBUILD?h=packages/openttd-opengfx
15:12:26  <Ammler> no, no binaries if possible
15:12:49  <Rubidium> then you already needed grfcodec for openttd
15:12:57  <Ammler> no, why?
15:13:35  <Ammler> openttd.grf is not necessary
15:13:38  <Rubidium> because openttd.grf is binary and you should build that from source if you can
15:14:09  <Ammler> but it might be an additional reason to submit grfcodec to factory
15:17:32  *** Zuu has joined #openttdcoop.devzone
15:27:51  <Ammler> meh, I have no clue anymore how to submit a package to factory, I need to start from scratch :-(
15:37:15  <Ammler> sr created: https://build.opensuse.org/request/show/238075
15:58:16  <Alberth> it's amazing how fast you for get things
15:59:06  <planetmaker> it gets scary when you forget what you forgot ;)
15:59:32  <Alberth> you are not affected by it, by then :)
15:59:42  <planetmaker> :D
16:00:05  <Alberth> but indeed, let's hope that doesn't happen
16:00:36  <Alberth> hmm, why did I add a space between "for" and "get"?
16:00:54  <planetmaker> just forget it ;)
16:01:12  <Alberth> I have no idea why :(
16:01:30  <planetmaker> foreign language. And both, for and get are more common words
16:01:32  <Alberth> ah well, let's do some dinner instead
16:01:44  <planetmaker> bon appetit
16:01:56  * planetmaker just had nicely fresh break and cottage cheese
16:02:04  <planetmaker> break? bread!
16:02:28  <Alberth> most of my uses of those words are  for (int i =0; ...    and  foo.getBlah...   :p
16:02:36  <planetmaker> see :)
16:02:56  <planetmaker> and I use 'break' also more often than 'bread' :)
16:03:07  <Alberth> break;
16:03:09  <Alberth> :)
16:03:13  <planetmaker> ^^
16:03:33  <Alberth> you can use "goto" instead :p
16:04:06  <planetmaker> not sure IDL supports it.
16:04:16  <planetmaker> probably does, though
16:04:38  <planetmaker> but I hate it :)
16:22:01  <DevZone> Project Japanese Buildings build #236-nightlies: SUCCESS in 21 sec: https://jenkins.openttdcoop.org/job/jpbuild/236/
16:34:05  <DevZone> Project Finnish Rail Infrastructure - Rails build #350-nightlies: SUCCESS in 8 min 25 sec: https://jenkins.openttdcoop.org/job/frissrails/350/
16:41:26  <DevZone> Project road-hog build #236-nightlies: SUCCESS in 45 sec: https://jenkins.openttdcoop.org/job/road-hog/236/
16:48:15  <DevZone> Project Iron Horse build #915-nightlies: SUCCESS in 1 min 34 sec: https://jenkins.openttdcoop.org/job/iron-horse/915/
16:49:37  <DevZone> Project Dutch Trainset build #190-push: SUCCESS in 1 min 22 sec: https://jenkins.openttdcoop.org/job/dutchtrains/190/
16:50:01  <DevZone> Project termite build #32-push: SUCCESS in 23 sec: https://jenkins.openttdcoop.org/job/termite/32/
16:51:05  <DevZone> Project xussrset - Trains from Russia build #324-push: SUCCESS in 3 min 37 sec: https://jenkins.openttdcoop.org/job/xussrset/324/
16:51:33  <DevZone> Project 2ccts build #160-push: SUCCESS in 1 min 32 sec: https://jenkins.openttdcoop.org/job/2ccts/160/
17:12:32  <Alberth> planetmaker: I guess I should a few lexer bugs?
17:12:40  <Alberth> +fix
17:13:44  <planetmaker> should is a strong word :)
17:14:08  <Alberth> s/should/could/
17:14:21  <Alberth> :)
17:14:29  <planetmaker> I'm sure you can, yes :) And it would be nice :)
17:20:13  <Alberth> hmm, t_error is indeed, silly
17:20:24  <Alberth> any other problems you found?
17:37:38  <planetmaker> no(t yet). The t_error seems to trigger for a couple of grfs though
17:42:11  <Alberth> it's weird, t_error is for reporting scanner errors, while the nml source is supposed to be correct
17:47:11  <Alberth> :o   cpp line hints are not understood :)
17:47:44  <Alberth> yeah, that should trigger quite a few projects ;)
17:54:31  <planetmaker> oh :)
18:03:28  *** ODM has joined #openttdcoop.devzone
18:33:17  *** Supercheese has joined #openttdcoop.devzone
18:47:32  <Alberth>  http://devs.openttd.org/~alberth/diffs/1_c_lex.patch    new patch
18:47:49  <Alberth> firs results are not so good, just 10 seconds gain :(
18:57:42  <Alberth> 1:27 -> 1:17
19:26:22  <planetmaker> hm... is that really the new patch?
19:28:14  <planetmaker> Alberth, when I use(d) that patch the t_error persists?
19:29:35  <Alberth> that's interesting, as I don't have "lexer" in that line:  print("Illegal character '{}' (character code 0x{:02X}) at {}, column {:d}".format(t.value[0], ord(t.value[0]), t.pos, t.column))
19:30:11  <planetmaker> Location: File "/home/planetmaker/ottd/grfdev/nml/nml/tokens.py", line 179, in t_error
19:30:48  <Alberth> what is wrong?
19:31:42  <planetmaker> https://paste.openttdcoop.org/show/3448/ <-- Am I right that this should have been fixed?
19:32:08  <planetmaker> hm... wait
19:33:20  <planetmaker> ah, yeah. wget doesn't overwrite files
19:33:35  <planetmaker> sorry, my fault :)
19:58:13  <Alberth> hmm, fair enough, it used to be 2:00  :)
19:59:23  <Alberth> http://paste.openttdcoop.org/show/3447/  new timing of the profile
20:05:08  *** ODM has quit IRC
20:07:23  <planetmaker> hm :)
20:07:46  <planetmaker> 1:17h to 2:00 is quite a good gain imho
20:09:09  <Alberth> the first 30+ seconds was adding a "+"  :p
20:09:28  <planetmaker> :D
20:12:14  <Alberth> scanner is down to 8 seconds, at 1.6E6 calls
20:12:30  <Alberth> parser is 16 seconds
20:12:40  <Alberth> and then a lot of 2-ish seconds
20:14:55  <planetmaker> ah, you look at the per-call time
20:15:03  <planetmaker> was looking at the cumulative one :)
20:15:44  <planetmaker>      1080    1.116    0.001   39.644    0.037 action2layout.py:372(get_layout_action2s) could bring much with small improvements
20:16:22  <planetmaker> but those are different places, totally. I guess
20:18:42  <Alberth> I fear you have to do something drastic like converting the Expression classes
20:19:26  <Alberth> but even then, together with removing the entire parser + scanner
20:19:38  <Alberth> you still barely within a minute
20:21:34  <Alberth> ever tried pushing the cpp output directly into nml?
20:22:17  <planetmaker> you mean like gcc file.pnml | nmlc --grf file.grf ?
20:22:48  <Alberth> yes
20:23:05  <Alberth> my implementation doesn't support that atm though
20:23:44  <Alberth> but I wonder whether it helps not writing 13MB data, and having overlap in CPU time between nml and cpp
20:24:46  <planetmaker> I didn't test that yet. At least not for timing purposes
20:25:25  <planetmaker> I did not go there in order to have an nml file to look at. But if it helps, that is of course an option
20:26:40  <frosch123> you could put the cpp output onto a tmpfs
20:26:44  <frosch123> just for testing the impact
20:27:14  <frosch123> when compiling ottd i can set -j = number of cores when using tmpfs
20:27:19  <frosch123> for the object files
20:31:25  <Alberth> adding a tee in the pipeline is easy :p
20:32:18  <planetmaker> yeah :)
20:38:20  <Alberth> patch updated
20:38:55  <Alberth> filename -   should open stdin
20:39:24  <Alberth> no idea if nml accepts that though :p
20:42:17  <Alberth> even if you reduce scanning + parsing to 0, you're still left with 50 seconds or so
20:42:21  <planetmaker> https://paste.openttdcoop.org/show/3449/ with the not-updated patch
20:43:40  <Alberth> hmm, not sure I check whether you can open the file :)
20:44:32  <DevZone> Project FIRS Industry Replacement Set build #126-push: SUCCESS in 5 min 21 sec: https://jenkins.openttdcoop.org/job/firs/126/
20:44:38  <planetmaker> hm, but same with new patch :)
20:44:58  <Alberth> add a -s  so you get a useful stack dump
20:46:28  *** frosch123 has quit IRC
20:47:02  <DevZone> Project opengfx-mars build #129-push: SUCCESS in 2 min 29 sec: https://jenkins.openttdcoop.org/job/opengfx-mars/129/
20:47:12  <planetmaker> haha :) https://paste.openttdcoop.org/show/3450/
20:47:20  <DevZone> Project Japanese Stations build #63-push: SUCCESS in 17 sec: https://jenkins.openttdcoop.org/job/jpstations/63/
20:47:21  <planetmaker> that kinda gives a hint in the last line :D
20:47:30  <DevZone> Project Townnames - Swiss build #29-push: SUCCESS in 10 sec: https://jenkins.openttdcoop.org/job/swisstowns/29/
20:47:45  <DevZone> Project skye-rail build #71-push: SUCCESS in 15 sec: https://jenkins.openttdcoop.org/job/skye-rail/71/
20:47:46  <DevZone> Project DevZone script update build #31-push: SUCCESS in 8 min 57 sec: https://jenkins.openttdcoop.org/job/misc/31/
20:48:02  <Alberth> :D
20:50:26  <Alberth> planetmaker:  change that to   if input_filename is None: input_filename = "-"
20:51:10  <Alberth> time to go to bed again
20:52:03  <Alberth> gn
20:52:06  <planetmaker> yeah, soonish. Sleep well
20:52:32  *** Alberth has left #openttdcoop.devzone
22:18:54  *** LSky` has quit IRC
23:22:40  *** Zuu has quit IRC

Powered by YARRSTE version: svn-trunk