| *** rdale has quit IRC | 00:45 | |
| *** toscalix has joined #buildstream | 02:23 | |
| *** toscalix has quit IRC | 02:25 | |
| *** traveltissues has joined #buildstream | 06:59 | |
| *** benschubert has joined #buildstream | 08:14 | |
| *** hasebastian has joined #buildstream | 08:27 | |
| *** tristan_ has joined #buildstream | 08:34 | |
| *** jib has joined #buildstream | 08:57 | |
| *** rdale has joined #buildstream | 09:08 | |
| benschubert | juergbi: https://gitlab.com/BuildStream/bst-plugins-experimental/-/tags we have a 1.93.1.1 tag, why not 1.93.2? Also, it's never been published, so overnight tests have been failing ever since then :( | 09:13 |
|---|---|---|
| coldtom | benschubert: iirc it was so it still matched the buildstream version | 09:14 |
| juergbi | it was just a quick fixup, no code changes | 09:15 |
| benschubert | we don't have a 1.93.1.1 | 09:15 |
| benschubert | ah | 09:15 |
| juergbi | but I failed to bump the version in setup.py :-/ | 09:15 |
| juergbi | (got used to git versioning) | 09:15 |
| benschubert | what was it pulling in? | 09:15 |
| benschubert | Should we just delete it and revert the chang in buildstream tests? | 09:15 |
| juergbi | it was to remove the buildstream dependency due to the recursive pip issue | 09:16 |
| juergbi | or rather, circular dependency issue | 09:16 |
| benschubert | ah yep! Happy to see that removed :) | 09:17 |
| juergbi | i.e., it's actually needed for buildstream CI, iirc | 09:17 |
| benschubert | Wut? | 09:17 |
| benschubert | the version was not pinned so it should not have caused any trouble | 09:17 |
| benschubert | assuming https://gitlab.com/BuildStream/bst-plugins-experimental/-/commit/2dbdb668d3e0a5c788c629f45825150f84fa7e88 is the only change | 09:18 |
| juergbi | we have BST_PLUGINS_EXPERIMENTAL_VERSION in tox.ini | 09:18 |
| benschubert | yep | 09:18 |
| *** tristan_ has quit IRC | 09:19 | |
| benschubert | https://gitlab.com/BuildStream/buildstream/-/jobs/484152873#L598 but it was never published on pypi | 09:20 |
| *** tristan_ has joined #buildstream | 09:21 | |
| benschubert | ok, seems it's the only place not installing from git | 09:21 |
| juergbi | right, but it works for all the other tests because there we install with git+... not pypi | 09:21 |
| benschubert | I'll fix this to install from git and it should pass | 09:21 |
| juergbi | (tox.ini) | 09:21 |
| benschubert | ah gotcha | 09:21 |
| juergbi | ta | 09:22 |
| benschubert | Would be good to always have a release on pypi when we tag though :) I though we already had that in place to be automated | 09:22 |
| benschubert | juergbi: https://gitlab.com/BuildStream/buildstream/-/jobs/484689617#L9852 do you have any idea _where_ that could be coming from? https://gitlab.com/BuildStream/buildstream/-/commit/551e0755e4dcb553ac220c5330a74a7b2b07cb19 is the breaking commit | 09:30 |
| juergbi | benschubert: pytest using a different temp directory? | 09:31 |
| benschubert | both directories still seems to be under cas control or am I reading this wrong? | 09:32 |
| juergbi | hm, yes, both are in /tmp, not just one | 09:32 |
| juergbi | was this the case before as well? I thought we normally use a subdirectory of the current/buildstream directory | 09:33 |
| benschubert | we used to have --basetemp, and I remember fiddling with it | 09:33 |
| benschubert | not sure if I removed it let me check | 09:34 |
| juergbi | and userchroot staging shouldn't be in /tmp | 09:34 |
| juergbi | seems very odd | 09:34 |
| juergbi | benschubert: maybe it's rather this commit? https://gitlab.com/BuildStream/buildstream/-/commit/088581901574d4af46b86bc9ab3f6bf99e694754 | 09:34 |
| juergbi | although, these are different tests, afaict | 09:35 |
| benschubert | https://gitlab.com/BuildStream/buildstream/-/commit/088581901574d4af46b86bc9ab3f6bf99e694754 nope :) | 09:35 |
| benschubert | (yeah same commit, don't know why ...) | 09:36 |
| WSalmon | my tests seem too be very slow and not using much cpu, it that what you two are looking at atm? | 09:36 |
| juergbi | if there is nothing obvious, I'd suggest bisecting the changes inside the commit | 09:36 |
| benschubert | WSalmon: no, I'm trying to update our dependencies :) | 09:36 |
| benschubert | juergbi: yep I'll do that thanks! | 09:36 |
| juergbi | WSalmon: do you mean locally or in CI? | 09:36 |
| juergbi | the whole test suite takes quite a while if you don't parallelize test execution | 09:37 |
| benschubert | juergbi: and the tmpdir was indeed removed. Should I add this back? | 09:37 |
| juergbi | benschubert: what do you mean exactly with tmpdir was removed? | 09:37 |
| WSalmon | ok, i wasnt quite folling but i thought some one had mentioned this here recently, I'm running `tox -e py37` locally i havent done any bst master stuff in ages so i can beleive i have forgoten how to turn more cpus on | 09:37 |
| juergbi | tox -- -n 8 | 09:38 |
| benschubert | the --basetemp was removed so it is in now in /tmp. we could do the same with tox --workdir | 09:38 |
| benschubert | WSalmon: yeah I have the same problem locally, but haven't got time to look at it | 09:38 |
| WSalmon | ta juergbi | 09:38 |
| benschubert | juergbi: so should I add the --workdir in tox to ensure we run in CWD instead of /tmp ? | 09:38 |
| *** santi has joined #buildstream | 09:40 | |
| juergbi | benschubert: where was this removed? I still see it. userchroot has particular requirements as staging happens in a special absolute directory but it needs to be able to hardlink from the CAS directory | 09:42 |
| juergbi | # Set up staging root with permissions required by userchroot, | 09:42 |
| juergbi | # must be on same filesystem as current directory to support hardlinks | 09:42 |
| juergbi | BST_CAS_STAGING_ROOT: "/builds/userchroot" | 09:42 |
| juergbi | it might not be an issue for the other tests because they are integration tests, which use INTEGRATION_CACHE | 09:42 |
| juergbi | which probably still is inside the current directory | 09:43 |
| benschubert | right, they are local | 09:43 |
| benschubert | but it doesn't seem to be using it? Or am I msunderstanding the error message? | 09:44 |
| juergbi | yes but I don't see where we removed it based on the git history | 09:45 |
| juergbi | it's still in tox.ini | 09:45 |
| benschubert | the --basetemp yes | 09:45 |
| benschubert | and it will run in toxworkdir/tmp so $(cwd)/.tox/tmp | 09:45 |
| benschubert | so that should still be fine | 09:45 |
| juergbi | so an update broke basetemp? | 09:46 |
| benschubert | not in the changelog but possible | 09:46 |
| benschubert | Ok I'll start bisecting to make sure :) | 09:46 |
| *** tpollard has joined #buildstream | 09:53 | |
| gitlab-br-bot | BenjaminSchubert opened MR !1839 (bschubert/fix-overnight->master: .gitlab-ci.yml: install requirements from git instead of pipy) on buildstream https://gitlab.com/BuildStream/buildstream/-/merge_requests/1839 | 09:55 |
| *** lachlan has joined #buildstream | 10:05 | |
| *** lachlan has quit IRC | 10:20 | |
| *** lachlan has joined #buildstream | 10:31 | |
| *** tristan_ has quit IRC | 10:48 | |
| gitlab-br-bot | juergbi approved MR !1839 (bschubert/fix-overnight->master: .gitlab-ci.yml: install requirements from git instead of pipy) on buildstream https://gitlab.com/BuildStream/buildstream/-/merge_requests/1839 | 10:58 |
| *** jib has left #buildstream | 11:01 | |
| *** lachlan has quit IRC | 11:30 | |
| *** phildawson has quit IRC | 11:32 | |
| *** lachlan has joined #buildstream | 11:38 | |
| gitlab-br-bot | BenjaminSchubert merged MR !1839 (bschubert/fix-overnight->master: .gitlab-ci.yml: install requirements from git instead of pipy) on buildstream https://gitlab.com/BuildStream/buildstream/-/merge_requests/1839 | 11:40 |
| *** lachlan has quit IRC | 11:44 | |
| *** lachlan has joined #buildstream | 11:46 | |
| benschubert | juergbi: it's indeed the update of pytest that breaks | 11:56 |
| gitlab-br-bot | valentindavid opened issue #1271 (Script element layout should be aware of junctions) on buildstream https://gitlab.com/BuildStream/buildstream/-/issues/1271 | 11:59 |
| tpollard | Thanks for raising that issue valentind | 12:07 |
| *** lachlan has quit IRC | 12:08 | |
| *** phildawson has joined #buildstream | 12:20 | |
| *** lachlan has joined #buildstream | 12:22 | |
| *** traveltissues has quit IRC | 12:23 | |
| *** traveltissues has joined #buildstream | 12:24 | |
| valentind | tpollard, It might need some refactoring. | 12:25 |
| valentind | I think Element.__build_dependencies and Element.__runtime_dependencies will need to store also the local name for the element. | 12:26 |
| *** lachlan has quit IRC | 12:28 | |
| *** traveltissues has quit IRC | 12:40 | |
| *** traveltissues has joined #buildstream | 12:44 | |
| *** traveltissues has quit IRC | 12:46 | |
| *** traveltissues_ has joined #buildstream | 12:46 | |
| *** traveltissues_ has left #buildstream | 12:47 | |
| *** traveltissues has joined #buildstream | 12:48 | |
| *** lachlan has joined #buildstream | 12:50 | |
| tpollard | hmm | 12:50 |
| *** lachlan has quit IRC | 13:06 | |
| *** hasebastian has quit IRC | 13:07 | |
| *** phildawson has quit IRC | 13:23 | |
| *** jib has joined #buildstream | 13:57 | |
| *** phildawson has joined #buildstream | 13:57 | |
| *** santi has quit IRC | 13:58 | |
| benschubert | WSalmon: just saw your comment on https://gitlab.com/BuildStream/buildstream/-/merge_requests/1740#note_311247394. I'm writing a message for the ML and will send it today explaining what my plan would be so we can discuss all of this :) | 13:58 |
| *** lachlan has joined #buildstream | 14:00 | |
| WSalmon | ok, seems like this should be sorted asap. if you have a good idea as to how to do it then lets get it on the ML and if we all like it and you dont have time i am happy to implement | 14:01 |
| *** santi has joined #buildstream | 14:02 | |
| benschubert | sure, thanks for the help and the reminder :) | 14:02 |
| WSalmon | is it worth turning them all off with a flag or do you want to mark them in a way other than skipif? if its complecated then wait for the ML but if its a yes i can get on | 14:18 |
| benschubert | I don't want skipif :) I think we should go with a marker. I believe this is significant enough (though relatively simple) that it should go to the ML | 14:19 |
| WSalmon | np | 14:19 |
| WSalmon | ta | 14:19 |
| benschubert | because we'll need a clear separation to know what it an test requiring externals, as it makes it easier to debug when having failing tests :) | 14:20 |
| benschubert | juergbi: https://gitlab.com/BuildStream/buildstream/-/issues/1042 is that up to date? | 14:30 |
| benschubert | and should we add this to the 2.0 Milestone? | 14:30 |
| juergbi | benschubert: yes, I think it's up-to-date | 14:32 |
| juergbi | not sure, the critical code parts are done | 14:32 |
| benschubert | OK ! | 14:32 |
| juergbi | should probably at least update the architecture docs before 2.0 | 14:32 |
| juergbi | but the remaining optimizations might not be blockers | 14:32 |
| benschubert | fair point :) | 14:33 |
| *** lachlan has quit IRC | 14:37 | |
| *** lachlan has joined #buildstream | 14:42 | |
| *** jib has left #buildstream | 15:04 | |
| *** lachlan has quit IRC | 15:58 | |
| *** lachlan has joined #buildstream | 16:19 | |
| *** m_22[m] has quit IRC | 16:45 | |
| *** krichter[m] has quit IRC | 16:45 | |
| *** doras[m] has quit IRC | 16:45 | |
| *** awacheux[m] has quit IRC | 16:45 | |
| *** DineshBhattarai[m] has quit IRC | 16:45 | |
| *** Demos[m] has quit IRC | 16:45 | |
| *** cgm[m] has quit IRC | 16:45 | |
| *** mattiasb has quit IRC | 16:45 | |
| *** Trevio[m] has quit IRC | 16:45 | |
| *** jjardon[m] has quit IRC | 16:46 | |
| *** tchaik[m] has quit IRC | 16:46 | |
| *** segfault1[m] has quit IRC | 16:46 | |
| *** abderrahim[m] has quit IRC | 16:46 | |
| *** dylan-m has quit IRC | 16:46 | |
| *** reuben640[m] has quit IRC | 16:46 | |
| *** tpollard has quit IRC | 17:10 | |
| *** tpollard has joined #buildstream | 17:10 | |
| *** DineshBhattarai[m] has joined #buildstream | 17:10 | |
| *** Demos[m] has joined #buildstream | 17:12 | |
| *** tpollard has quit IRC | 17:15 | |
| *** tpollard has joined #buildstream | 17:16 | |
| *** tpollard has quit IRC | 17:18 | |
| *** jjardon[m] has joined #buildstream | 17:26 | |
| *** tchaik[m] has joined #buildstream | 17:36 | |
| *** lachlan has quit IRC | 17:36 | |
| benschubert | https://mail.gnome.org/archives/buildstream-list/ gives a 403? Does someone know who to contact? jjardon ? | 17:42 |
| *** tpollard has joined #buildstream | 17:45 | |
| jjardon | av in #syadmin ; let me ping him | 17:46 |
| benschubert | thanks :) | 17:46 |
| *** lachlan has joined #buildstream | 17:48 | |
| jjardon | benschubert: known issue; it should be fixed soonish | 17:50 |
| jjardon | They are moving services around | 17:50 |
| benschubert | great, thanks! Does it affect mails sent during that time? (Just sent one to the ML) | 17:51 |
| jjardon | Not sure, sorry; let's check when it come back to live | 17:52 |
| benschubert | Yep that's fine, I'll have a look and send it back tomorrow if it's not on it :) thanks a lot! | 17:53 |
| benschubert | juergbi: https://gitlab.com/BuildStream/buildstream/-/blob/0591b0c8b038aa9f054b295748f3423384711d50/src/buildstream/_context.py#L289 do we really want this? Would it not hurt if people install both BuildStream versions in parallel? Also, how long would we keep that in? :) | 17:54 |
| juergbi | benschubert: the issue is that they might never be cleaned up otherwise | 17:55 |
| benschubert | fair enough. I was just curious since I was looking at https://gitlab.com/BuildStream/buildstream/-/issues/1261 | 17:56 |
| juergbi | right now we don't really have parallel installability but it's a topic to discuss | 17:56 |
| benschubert | to be frank: I don't care about it :D So if it's a requirement from other people, we should look at it, but otherwise, well let's just document it is not :D | 17:56 |
| juergbi | benschubert: note that the extract directories are just a cache | 17:56 |
| juergbi | hardlinked from cas | 17:56 |
| juergbi | nothing is really lost | 17:57 |
| benschubert | ah fair point so we actually don't care | 17:57 |
| benschubert | Do you think we should add a timer there or is this something that is rare enough that we would not care? | 17:57 |
| benschubert | (I'd rather not slow down and pollute the UI if we don't need to) | 17:57 |
| juergbi | I wouldn't really care about a timer | 17:58 |
| juergbi | although we could print a single info line if we find the directory | 17:58 |
| benschubert | That would be a good compromise, I'll update the ticket | 17:58 |
| *** phildawson has quit IRC | 18:05 | |
| *** m_22[m] has joined #buildstream | 18:06 | |
| *** tpollard has quit IRC | 18:13 | |
| *** awacheux[m] has joined #buildstream | 18:28 | |
| *** santi has quit IRC | 18:33 | |
| *** reuben640[m] has joined #buildstream | 18:36 | |
| *** santi has joined #buildstream | 18:38 | |
| *** santi has quit IRC | 18:41 | |
| *** lachlan has quit IRC | 18:45 | |
| *** Trevio[m] has joined #buildstream | 18:47 | |
| *** lachlan has joined #buildstream | 18:57 | |
| *** abderrahim[m] has joined #buildstream | 18:58 | |
| *** doras[m] has joined #buildstream | 19:07 | |
| *** lachlan has quit IRC | 19:13 | |
| gitlab-br-bot | abderrahimk closed issue #525 ('quit' option doesn't work when interrupting a build) on buildstream https://gitlab.com/BuildStream/buildstream/-/issues/525 | 19:18 |
| *** rdale has quit IRC | 19:23 | |
| *** krichter[m] has joined #buildstream | 19:36 | |
| *** segfault1[m] has joined #buildstream | 19:48 | |
| *** cgm[m] has joined #buildstream | 20:03 | |
| *** mattiasb has joined #buildstream | 20:06 | |
| *** dylan-m has joined #buildstream | 20:08 | |
| *** narispo has quit IRC | 21:27 | |
| *** narispo has joined #buildstream | 21:27 | |
| gitlab-br-bot | sstriker opened issue #1272 (Retire Artifact Cache, use a Remote Asset API based Asset Cache instead) on buildstream https://gitlab.com/BuildStream/buildstream/-/issues/1272 | 21:51 |
| gitlab-br-bot | sstriker opened issue #1273 (Retire Source Cache, use a Remote Asset API based Asset Cache instead) on buildstream https://gitlab.com/BuildStream/buildstream/-/issues/1273 | 21:56 |
| gitlab-br-bot | sstriker opened issue #1274 (Introduce caching of individual sources, using a Remote Asset API based Asset Cache) on buildstream https://gitlab.com/BuildStream/buildstream/-/issues/1274 | 22:02 |
| gitlab-br-bot | sstriker opened issue #1275 (Introduce tracking of individual sources, using the Remote Asset API) on buildstream https://gitlab.com/BuildStream/buildstream/-/issues/1275 | 22:08 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!