| *** tristan has joined #buildstream | 00:53 | |
| *** persia has quit IRC | 04:20 | |
| gitlab-br-bot | push on buildstream@test-build (by Tristan Van Berkom): 15 commits (last: _scheduler module: Create an event loop instead of using the default) https://gitlab.com/BuildStream/buildstream/commit/84c871b8a60f7826392c35a36735ac52df29ef74 | 06:45 |
|---|---|---|
| *** jude has joined #buildstream | 08:05 | |
| gitlab-br-bot | buildstream: Tristan Van Berkom deleted branch test-build | 08:16 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 16 commits (last: _scheduler module: Create an event loop instead of using the default) https://gitlab.com/BuildStream/buildstream/commit/84c871b8a60f7826392c35a36735ac52df29ef74 | 08:17 |
| *** jonathanmaw has joined #buildstream | 08:37 | |
| *** ssam2 has joined #buildstream | 08:52 | |
| *** tiagogomes has joined #buildstream | 09:05 | |
| *** bochecha has joined #buildstream | 09:05 | |
| *** tlater has joined #buildstream | 09:16 | |
| *** tlater has left #buildstream | 09:21 | |
| *** tlater has joined #buildstream | 09:22 | |
| *** bochecha_ has joined #buildstream | 09:33 | |
| *** bochecha has quit IRC | 09:33 | |
| gitlab-br-bot | push on buildstream@cross_platform (by Tristan Maat): 2 commits (last: Fix sandbox read-onlyness as discussed with tristan (GIVE THIS A) https://gitlab.com/BuildStream/buildstream/commit/32ed3ae50ee4b09ef9fa877cab421a49eb4750ad | 11:34 |
| gitlab-br-bot | buildstream: merge request (cross_platform->master: WIP: Cross platform) #81 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/81 | 11:34 |
| *** tlater has quit IRC | 12:13 | |
| *** bochecha has joined #buildstream | 12:18 | |
| *** bochecha_ has quit IRC | 12:20 | |
| *** tlater has joined #buildstream | 12:29 | |
| *** tlater has left #buildstream | 12:35 | |
| *** tlater has joined #buildstream | 12:35 | |
| gitlab-br-bot | push on buildstream@cross_platform (by Tristan Maat): 1 commit (last: Fix sandbox output issues) https://gitlab.com/BuildStream/buildstream/commit/724bbcd3f15bbe13606c4be6314cd326ba1af139 | 13:25 |
| gitlab-br-bot | buildstream: merge request (cross_platform->master: WIP: Cross platform) #81 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/81 | 13:26 |
| *** tlater has quit IRC | 13:55 | |
| *** tlater has joined #buildstream | 13:55 | |
| *** ChanServ sets mode: +o tristan | 14:40 | |
| tristan | tlater, are there any plans yet to run the CI on the platforms we support ? | 14:46 |
| ssam2 | tricky to orchestrate as the only Solaris and AIX machines we have are provided by a 3rd party | 14:48 |
| ssam2 | also GitLab CI on AIX looks like a real headache because the gitlab-ci-runner program is written in Go, and Go doesn't appear to support AIX | 14:48 |
| tristan | :'( | 14:49 |
| tristan | Well, I'm sure gitlab-ci-runner could run somewhere that is not AIX/Solaris | 14:50 |
| tristan | of course it's still 'tricky' | 14:50 |
| ssam2 | yeah it could run as a proxy over SSH | 14:50 |
| tristan | then gitlab-ci-runner could slave it out... but it needs what; docker everywhere I guess | 14:50 |
| * ironfoot remembers email about gitlab-ci limitation | 14:50 | |
| ironfoot | i take that with our own runners we will not have problems? | 14:51 |
| ssam2 | oh, you mean the "CI minutes" limit? yeah I think we're fine there | 14:51 |
| ssam2 | BuildStream uses the GitLab-hosted runners, but I think they are only limited for private projects | 14:52 |
| tlater | tristan: Especially since some unit test things still require ostree support, it would be a bit hard to distinguish between things that should pass/fail | 14:52 |
| tristan | tlater, of course fixing the tests to avoid them where not supported (and report excluded tests in the test output) is part of it | 14:53 |
| tristan | there is a pytest decorator for that iirc | 14:53 |
| tristan | so you can enable/disable tests according to a var | 14:53 |
| * tristan added a lot of tests yesterday | 14:54 | |
| tlater | Hm, well, I can start on that while waiting for builds at any rate | 14:56 |
| tristan | tlater, well hold on... ssam2's concerns need to addressed I think | 14:58 |
| tlater | tristan: Wouldn't tests be orthogonal to that anyway? They would be very useful just to confirm functionality on the platforms. | 14:59 |
| tristan | ssam2, https://gitlab.com/gitlab-org/gitlab-ci-multi-runner/blob/master/docs/executors/virtualbox.md ? | 14:59 |
| tristan | tlater, good point | 15:00 |
| tristan | tlater, you should certainly be able to run ./setup.py test on other platforms | 15:00 |
| tristan | tlater, see @pytest.mark.skipif() in tests/sources/bzr.py for example | 15:01 |
| tristan | I think I have to do similar but I have to figure out how to do it differently | 15:02 |
| tlater | tristan: You can use unittest.skipif, according to the pytest documentation | 15:19 |
| tristan | tlater, not @pytest.skipif() ?? | 15:21 |
| tristan | wait, why would you use unittest.skipif() ? | 15:21 |
| tlater | tristan: It's what the pytest documentation comes up with if you search for skipif - it also has other more verbose helpers for similar things | 15:23 |
| tlater | Such as .skipunless | 15:23 |
| tlater | Nevermind, I'm in the wrong documentation | 15:24 |
| tlater | x) | 15:24 |
| tristan | hmm | 15:36 |
| * tristan has made a couple of tweaks and now tests run again without bzr installed | 15:36 | |
| *** tlater has quit IRC | 15:41 | |
| *** tlater has joined #buildstream | 15:42 | |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 6 commits (last: testutis/site.py: A location for checking the presense of host features) https://gitlab.com/BuildStream/buildstream/commit/0d665475aca1bb847749dce7cc415aefcf238e87 | 15:50 |
| tristan | tlater, ok now there is testutils/site.py | 15:50 |
| tristan | tlater, from any test you can do 'from tests.testsutils.site import HAVE_BZR' for instance | 15:50 |
| tlater | tristan: Ah, so now I can add things like HAVE_OSTREE | 15:52 |
| tlater | or those already exist ;) | 15:53 |
| gitlab-br-bot | push on buildstream@cross_platform (by Tristan Maat): 20 commits (last: Implement tarcache) https://gitlab.com/BuildStream/buildstream/commit/95ae8a23c6dbd5395370e333d0a025407fdf31b4 | 15:56 |
| gitlab-br-bot | buildstream: merge request (cross_platform->master: WIP: Cross platform) #81 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/81 | 15:56 |
| tristan | tlater, Well. | 15:57 |
| tristan | tlater, yes I think that is sane | 15:58 |
| tristan | tlater, probably for consistency we want an entry there; although I suspect you must have checked HAVE_OSTREE at some point in the core anyway | 15:58 |
| tristan | I've added only HAVE_OSTREE_CLI because some tests need a cli but we dont assert for that at install | 15:59 |
| * tristan notes that tests/sources/ostree.py is entirely useless | 15:59 | |
| tristan | and will remove it | 15:59 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 2 commits (last: cachekey.py: Fix broken logic of excluding tests) https://gitlab.com/BuildStream/buildstream/commit/99f7febbcf697ad198af4a493feb21f7f56529dc | 16:00 |
| tlater | tristan: Some tests will require root permissions if bwrap is not available, should I just add something that checks os.geteuid()? | 16:01 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 1 commit (last: tests/sources/ostree.py: Removed test and data.) https://gitlab.com/BuildStream/buildstream/commit/ce675ba46b6745b2625fc21e03a8fbf50b2de78b | 16:04 |
| tristan | tlater, sure | 16:04 |
| tristan | tlater, so what we could do is have an option for force using the fallback backends used for unixy platforms, on linux | 16:06 |
| tristan | tlater, and then we could test those code paths in the integration tests at least | 16:06 |
| tristan | And maybe its a hidden testing only thing | 16:07 |
| tristan | if the user can choose to do that, then we'd probably need to take the sandbox type into account in the cache keys | 16:08 |
| tristan | maybe we should be doing that anyway ? | 16:08 |
| tristan | immaterial right now as long as there is no choice, anyway | 16:08 |
| tlater | tristan: The caches don't currently share artifacts at all, not sure that's necessary - certainly not before tar pushing/pulling is implemented | 16:10 |
| tristan | well if it's on a different platform then the cache key already has to be different | 16:13 |
| tristan | if something is built on another platform | 16:13 |
| tristan | but if it's built on the same platform but with a different Platform backend (cause we forced it), I *think* that we need 'choice of plaform backend' to be taken into account for cache key generation | 16:14 |
| tristan | we dont want to advertise equal keys built by different backends, even if those are (currently) stored by different artifact servers | 16:15 |
| tlater | I guess that would be necessary in the case that another platform supports ostree in the future | 16:15 |
| tristan | Right | 16:15 |
| tlater | Or another cache | 16:15 |
| tristan | We should automatically start using it but have that effect cache key | 16:16 |
| tristan | or smth | 16:16 |
| tlater | Should this block the current MR or be part of something else? | 16:19 |
| tristan | tlater, not block no | 16:19 |
| tristan | tlater, well wait... choice of backend in cache key should be I think | 16:20 |
| tristan | tlater, because it's very inexpensive, we should have it considered | 16:20 |
| tristan | tlater, the tests should also run on your target platforms really | 16:21 |
| tristan | but as far as automating things on gitlab and forcing the fallback platform so that we can automate testing of both... no that shouldnt block | 16:21 |
| tristan | having passing `./setup.py test` shouldnt be too challenging though | 16:21 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 2 commits (last: project.py: Declare main BST_ARTIFACT_VERSION here) https://gitlab.com/BuildStream/buildstream/commit/1d9e7c186dce262842736a8089069fb33121f57c | 16:23 |
| ssam2 | something BuildStream does seems to break 'pdb' ... when I add `pdb.set_trace()` calls they don't always work | 17:12 |
| ssam2 | oh, maybe the error is before my call :-) | 17:13 |
| ssam2 | ignore me | 17:13 |
| *** tlater has quit IRC | 17:17 | |
| *** ssam2 has quit IRC | 17:23 | |
| *** jonathanmaw has quit IRC | 17:31 | |
| tristan | What is the point of `bst workspace reset --no-checkout` ? | 17:55 |
| gitlab-br-bot | buildstream: issue #55 ("Workspace list in machine readable format") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/55 | 19:03 |
| gitlab-br-bot | buildstream: issue #80 ("Awkward to recover after manually deleting workspace dir") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/80 | 19:03 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 7 commits (last: _yaml.py: Make _yaml.dump() optionally dump to stdout) https://gitlab.com/BuildStream/buildstream/commit/32ea8e4f305d5968f50df59798e6eb0342b05bc2 | 19:03 |
| *** tristan has quit IRC | 19:21 | |
| *** tristan has joined #buildstream | 20:38 | |
| *** bochecha has quit IRC | 20:49 | |
| *** bochecha has joined #buildstream | 21:17 | |
| *** bochecha has quit IRC | 22:18 | |
| *** tristan has quit IRC | 22:35 | |
| *** tristan has joined #buildstream | 22:53 | |
| gitlab-br-bot | push on buildstream@test-build (by Tristan Van Berkom): 10 commits (last: Restructuring tests using the Repo and Cli) https://gitlab.com/BuildStream/buildstream/commit/193bc3c195e9d0843d38f4a7e4eecac5b8daa1a2 | 23:33 |
| gitlab-br-bot | push on buildstream@test-build (by Tristan Van Berkom): 2 commits (last: tests/frontend/push.py: Adding test for `bst push`) https://gitlab.com/BuildStream/buildstream/commit/788ceecb07b42897b20241738d791725962aacd2 | 23:38 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!