Log for #openttdcoop.devzone on 10th November 2014:
Times are UTC Toggle Colours
09:04:18  <planetmaker> wtf, 6 hours?
09:04:47  <Supercheese> yetis were on strike
09:06:06  <V453000> new industry :)
09:09:41  <Sylf> well
09:09:50  <Sylf> there are things I'm starting to notice
09:10:18  <Sylf> like this message
09:10:20  <Sylf> ERROR: Workspace reports paths.default as" target="_blank"> which looks different than$JOB_NAME
09:11:07  <Sylf> or how posting the bundle pushes the source zip, which is about 2GB, taking up almost 3 hours
09:11:44  <Sylf> maybe the first error may fix the issue of jenkins not using the cache
09:24:26  <planetmaker> Sylf, yes, that message is normal and currently not a bug
09:24:38  <Sylf> hm, ok
09:25:37  <Sylf> anyway, it's the lack of cache and 2GB transfer that contributes to 6 hour job
09:26:21  <planetmaker> in the mid future I want to get rid of it, but that but it basically requires to change the jenkins config
09:26:49  <planetmaker> not a big deal, but then I can't use the template. Which means I need a script which generates the config and writes it via REST api to jenkins
09:26:54  <planetmaker> that's the plan. but not yet done
09:27:23  <planetmaker> and yes, yeti would profit a lot from not complete re-build. I guess... I could change that there...
09:28:34  <planetmaker> Sylf, on the other hand: the CF is supposed to test building from scratch with cold caches everywhere
09:30:50  <planetmaker> so... let's skip the complete re-clone everytime by manually configuring the SCM URL
09:30:58  <planetmaker> for now
09:32:27  <planetmaker> that will save 15 minutes at most :P
09:32:39  <Sylf> :/
09:33:03  <planetmaker> I can disable cleaning the working dir, if you want
09:33:11  <planetmaker> thus have nml use the caches
09:33:17  <planetmaker> should we try?
09:33:41  <Sylf> we should
09:34:26  <Sylf> so yeti won't take up a worker thread 6 hours every day as more translations come in
09:34:44  <planetmaker> hm... that wasn't even set. But yeah, it cleaned anew, so they were gone
09:35:12  <planetmaker> well, I'm not concerned about the cpu usage really. If it were used 24 hours a day... fine :)
09:35:17  <planetmaker> it's its purpose after all
09:36:09  <planetmaker> I would really only be concerned when both cpus assigned to the CF were used 24/7 :)
09:36:18  <V453000> btw brewing? :D
09:36:23  <V453000> wtf
09:36:24  <V453000> *
09:36:25  <planetmaker> hu?
09:37:21  <Sylf> anyway.  take a look at too
09:37:36  <Sylf> it's missing the actual grf file
09:39:59  <V453000> :D
09:45:15  <planetmaker> hm... why is it missing that? :S
09:46:41  <planetmaker> well, there shouldn't be a grf file but a zip file
09:46:45  <planetmaker> and the doc files
09:47:25  <planetmaker> <-- basically that should be there...
09:49:13  <planetmaker> wait, that's the *nightly* dir. It's all in the push dirs
09:49:24  <planetmaker> so... something's going wrong when moving latest push to nightly
09:49:34  <planetmaker>
09:50:00  <planetmaker> Sylf, 'nightly' is not build specifically. It's only copying from push to nightly folder. Good catch!
09:50:02  <Sylf> oh
09:50:46  <planetmaker> I changed that somewhen... as nightlies became pointless when we build every push anyway
09:50:48  <V453000> btw I think eints pushes should not make it compile, no real reason
09:51:11  <planetmaker> V453000, yes, there's a real reason: it's a change, it's a push. And translators will want to see their work and test they did it right
09:51:23  <planetmaker> I won't disable that
09:51:35  <V453000> meh k
09:52:07  <planetmaker> also it's a hell to figure out where a push comes from. Not going to dive into trying to figure out that
09:52:23  <V453000> sure was just a suggestion :)
09:53:31  <planetmaker> if you don't want translations automatically, you can disable them. But if you want them, you get a built for translation updates
09:53:48  <planetmaker> that's fine-grain enough to control imho
09:54:25  <planetmaker> and as said: I don't mind the CF being used
10:00:27  <V453000> sure sure :)
10:01:35  <planetmaker> we'll see how long it takes next push or 2nd next. I *think* it should now use the nml caches
10:01:44  <planetmaker> and not clone completely from scratch either
10:02:00  <planetmaker> not sure whether caches are still there... so either next or 2nd next
12:37:34  *** Supercheese has quit IRC
15:29:48  *** frosch123 has joined #openttdcoop.devzone
16:04:10  *** ODM has joined #openttdcoop.devzone
16:51:32  *** Alberth has joined #openttdcoop.devzone
16:52:13  <DevZone> Project World Airliner Set build #586-push: SUCCESS in 4 min 2 sec:
16:52:42  <DevZone> Project FIRS Industry Replacement Set build #195-push: SUCCESS in 4 min 39 sec:
16:54:16  <DevZone> Project 2ccts build #352-push: SUCCESS in 2 min 2 sec:
17:30:40  *** Alberth has left #openttdcoop.devzone
17:31:29  *** Alberth has joined #openttdcoop.devzone
17:50:06  *** oskari89 has joined #openttdcoop.devzone
18:08:56  *** LSky` has joined #openttdcoop.devzone
18:40:32  <DevZone> Project road-hog build #471-nightlies: SUCCESS in 31 sec:
18:58:23  <DevZone> Project World Airliner Set build #587-push: SUCCESS in 3 min 19 sec:
19:11:19  <DevZone> Project World Airliner Set build #588-push: SUCCESS in 3 min 22 sec:
19:14:51  <Sylf> 31 second build time... now, I'm jealous
19:20:08  <Alberth> :)
19:23:38  <V453000> XD
19:36:06  *** Supercheese has joined #openttdcoop.devzone
19:41:27  <planetmaker> Sylf, will you commit today anything to yeti?
19:41:44  <Sylf> I can - I have some changes in my mind
19:42:11  <planetmaker> otherwise I'd trigger a build. But if you're about to commit something, then I won't
19:45:13  <Sylf> Let me commit this one small change
19:46:35  <planetmaker> no rush... in case of doubt I won't notice anyway before I wake up again tomorrow :P
19:48:33  <Sylf> making yeti fat again...
19:48:42  <Sylf> change yeti weight from 80kg to 1t
19:59:57  <planetmaker> :D
20:46:35  *** Alberth has left #openttdcoop.devzone
21:04:22  <DevZone> Project xussrset - Trains from Russia build #541-push: SUCCESS in 4 min 23 sec:
21:09:35  *** ODM has quit IRC
21:19:39  <DevZone> Project xussrset - Trains from Russia build #542-push: SUCCESS in 4 min 53 sec:
21:32:19  <DevZone> Project CHIPS Station Set build #16: FAILURE in 2.2 sec:
21:39:08  <frosch123> oh my, i must be terrible at programming :p
21:39:16  <planetmaker> hu?
21:39:34  <frosch123> i wrote a c module for nml which does the GRF compression in C
21:39:51  <frosch123> i profiled it, there is no time lost in overhead of passing data etc
21:40:10  <frosch123> yet, the c implementation is way slower than nml :p
21:40:23  <planetmaker> :D
21:40:43  <frosch123> s/than nml/plain python/
21:40:44  <planetmaker> standard python is cpython...
21:40:52  <planetmaker> so some parts are already in c
21:41:29  <frosch123> the compression algorithm is some nasty almost quadratic time thing
21:42:03  <frosch123> all the time is used in a single loop resp. function, which i ported to C
21:42:05  <Sylf> hm
21:42:14  <Sylf> jenkins compiled yeti in 34 seconds
21:42:25  <frosch123> now the C function takes all the time, though the function is really plain C, no fancy functions called etc
21:42:47  <planetmaker> sounds quick, Sylf :)
21:42:53  <planetmaker> check whether it's correct?
21:43:05  <Sylf> not yet
21:43:30  <planetmaker> publishing?
21:43:38  <Sylf> yup.
21:43:46  <frosch123> my machine takes 16 seconds if the cache is ready
21:43:51  <frosch123> so, 34 sounds kind of slow
21:43:55  <planetmaker> that sounds correct then :)
21:44:18  <frosch123> oh, i have a two weeks old version though
21:44:24  <frosch123> so, maybe you added stuff :p
21:45:00  <Sylf> 1 industry, yes.
21:46:18  <DevZone> Yippee, build fixed!
21:46:19  <DevZone> Project CHIPS Station Set build #17-push: FIXED in 26 sec:
21:46:20  <DevZone> Yippee, build fixed!
21:46:21  <DevZone> Project CHIPS Station Set build #17-push: FIXED in 26 sec:
21:47:13  <planetmaker> seems chips in jenkins didn't get any attention for ages :D
21:52:31  <frosch123> yay, 43% faster
21:53:10  <frosch123> somewhere around that area was also grfcodec
21:55:16  <planetmaker> :) So there was room for improvement?
21:55:50  <frosch123> apparently memcmp was not inlined contrary to my expectations
21:56:23  <DevZone> Project NML for CETS build #11-push: STILL FAILING in 8.3 sec:
21:56:54  <planetmaker> wtf?
23:12:16  *** oskari89 has quit IRC
23:48:16  *** LSky` has quit IRC
23:58:02  *** frosch123 has quit IRC

Powered by YARRSTE version: svn-trunk