| *** jude has quit IRC | 02:24 | |
| *** tristan has quit IRC | 02:38 | |
| *** tristan has joined #buildstream | 02:47 | |
| *** ChanServ sets mode: +o tristan | 02:47 | |
| *** jude has joined #buildstream | 04:19 | |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 6 commits (last: _pipeline.py: Fixed bad except statement) https://gitlab.com/BuildStream/buildstream/commit/59150f5cc70c3f4334e2404787dbee8165cdd99c | 06:32 |
|---|---|---|
| *** tristan has quit IRC | 07:31 | |
| *** tlater has joined #buildstream | 08:02 | |
| tlater | Does tristan happen to be here? | 08:04 |
| tlater | No :/ | 08:04 |
| gitlab-br-bot | buildstream: merge request (tristan/test_tests->master: WIP: Add integration tests) #40 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/40 | 08:20 |
| gitlab-br-bot | buildstream: merge request (tristan/test_tests->master: WIP: Add integration tests) #40 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/40 | 08:21 |
| *** jonathanmaw has joined #buildstream | 08:40 | |
| *** ssam2 has joined #buildstream | 09:13 | |
| tlater | Alright, here's my current attempt: https://gitlab.com/BuildStream/buildstream-tests/blob/6539eaf2ce41e8e71da172bee4cce0f737c74e53/compose-test/elements/dependencies/amhello-run.bst | 09:38 |
| tlater | I've tried applying the same changes to the project.conf, but the test domain doesn't seem to be usable in a compose file. | 09:39 |
| ssam2 | https://gitlab.com/BuildStream/buildstream/blob/master/buildstream/data/projectconfig.yaml says ... | 09:40 |
| ssam2 | # Elements may extend these rules by adding entries to the # various split domains either in a plugin definition or # in a project element declaration. | 09:40 |
| ssam2 | not sure what that means though | 09:40 |
| tlater | I thought that's what I attempted... Well, thanks, I'll have to look through the source I guess. | 09:41 |
| ssam2 | try adding the `split-rules: ` dict to a project.conf file perhaps ? | 09:41 |
| ssam2 | maybe that's what a "project element declaration" is.. | 09:41 |
| tlater | Yeah, I tried | 09:41 |
| ssam2 | ah | 09:41 |
| tlater | Unless this is an actual bug, that doesn't work. | 09:42 |
| ssam2 | also possible :-) | 09:42 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 1 commit (last: utils.py: Add missing signal import) https://gitlab.com/BuildStream/buildstream/commit/1204b690ce5a05759947b7db44f1d87e7adb2c1a | 10:33 |
| gitlab-br-bot | buildstream: merge request (utils-call-fix->master: utils.py: Add missing signal import) #43 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/43 | 10:33 |
| tlater | Some of the tests take quite long; Especially 'qmake' and 'distutils' elements have a ton of large dependencies. | 12:41 |
| tlater | There isn't much of a better way to run these, is there? | 12:41 |
| ssam2 | would artifact caching help here ? | 12:43 |
| tlater | Not sure what it does, does it deliver ready-built "packages"? | 12:44 |
| ssam2 | yeah | 12:45 |
| ssam2 | you build them once, push them to the cache, then second time you don't need to build them again | 12:45 |
| tlater | Ah, I don't think that will help much, the CI will have to run on each commit. | 12:45 |
| tlater | If that takes ~2 hours for each commit it won't be very useful. | 12:45 |
| ssam2 | what makes it so slow ? | 12:46 |
| tlater | Building everything in an image up till python. | 12:46 |
| tlater | Or qt, for that matter. | 12:46 |
| ssam2 | right, so surely if you could reuse prebuilt versions of all those artifacts, it would help ? | 12:46 |
| tlater | Yeah, but that won't work on the CI server, will it? | 12:47 |
| ssam2 | we could make it work | 12:47 |
| ssam2 | the current GitLab CI runners that jjardon[m] provides don't make this easy | 12:47 |
| ssam2 | but we can provide runners in some other way, or provide a persistant cache somewhere that the runners can access easily | 12:47 |
| tlater | That might be necessary | 12:48 |
| ssam2 | Baserock has some OpenStack capacity that Codethink sponsors... so we could set up runners there with a persistant cache perhaps | 12:50 |
| tlater | Not all builds should be cached though. Some of them actually need to run for testing purposes. Is that possible? | 12:53 |
| ssam2 | it could be | 12:56 |
| ssam2 | there are a few possible approaches .. | 12:56 |
| ssam2 | add a --force-rebuild option to BuildStream | 12:56 |
| ssam2 | manually delete the thing we want to rebuild from our cache | 12:57 |
| ssam2 | or include some kind of UUID in the build instructions to force a rebuild | 12:57 |
| tlater | Yeah, that should be fairly easy to automate using test scripts actually. Probably should do that anyway. | 12:59 |
| tlater | I guess we'll have to wait a bit before we can set something like this up though? I can avoid committing the big tests for now, so that we can merge this and not break gitlab. | 13:01 |
| ssam2 | it's stuff we need to look at soon | 13:02 |
| ssam2 | but probably not ASAP | 13:02 |
| tlater | Hmmm. Would it make sense to use mocks instead? i.e., small bash scripts that pretend to be python, and just write to a file what they were called with? | 13:23 |
| *** jude has quit IRC | 13:35 | |
| ssam2 | i'm not sure.. best to discuss with tristan | 13:43 |
| ssam2 | there are arguments both ways. In general though I feel it's best to test what you're actually going to use, rather than a separate fake version | 13:43 |
| tlater | Probably. If it's going to be thrown out for a more thorough test next week I'd also rather not waste the time to come up with an elaborate mocking system. | 13:45 |
| ssam2 | yeah.. it maybe that you've done enough with this task for now | 13:46 |
| tlater | Soon, anyway. I'm done with the element tests, I just need to go through source tests and fix the compose test. | 13:47 |
| tlater | That's ignoring the tests with large dependencies because I'm not sure how to tackle them yet. | 13:47 |
| *** violeta has joined #buildstream | 13:48 | |
| *** violeta has quit IRC | 14:08 | |
| *** violeta has joined #buildstream | 14:11 | |
| *** jude has joined #buildstream | 14:20 | |
| *** tlater has quit IRC | 16:20 | |
| gitlab-br-bot | buildstream: merge request (sam/track-default-none->master: Change the default behaviour of `bst track` to --deps=none) #44 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/44 | 16:29 |
| *** jude has quit IRC | 16:59 | |
| *** jude has joined #buildstream | 17:06 | |
| *** ssam2 has quit IRC | 17:08 | |
| *** jonathanmaw has quit IRC | 17:52 | |
| *** violeta has quit IRC | 18:56 | |
| *** tristan has joined #buildstream | 19:25 | |
| *** tristan has quit IRC | 20:45 | |
| *** juergbi has quit IRC | 20:45 | |
| *** persia has quit IRC | 20:45 | |
| *** jjardon[m] has quit IRC | 20:45 | |
| *** dabukalam has quit IRC | 20:45 | |
| *** tristan has joined #buildstream | 20:46 | |
| *** juergbi has joined #buildstream | 20:46 | |
| *** persia has joined #buildstream | 20:46 | |
| *** jjardon[m] has joined #buildstream | 20:46 | |
| *** dabukalam has joined #buildstream | 20:46 | |
| *** jude has quit IRC | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!