| *** narispo has quit IRC | 04:39 | |
| *** narispo has joined #buildstream | 04:39 | |
| *** slaf has joined #buildstream | 05:34 | |
| *** slaf has joined #buildstream | 05:34 | |
| *** samkirkham has joined #buildstream | 07:14 | |
| *** ikerperez has joined #buildstream | 07:46 | |
| *** phil has joined #buildstream | 07:46 | |
| *** narispo has quit IRC | 08:13 | |
| *** narispo has joined #buildstream | 08:13 | |
| *** phildawson_ has joined #buildstream | 08:18 | |
| *** phil has quit IRC | 08:18 | |
| gitlab-br-bot | juergbi opened MR !1566 (juerg/cache-quota->master: Cache quota configuration fixes) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1566 | 08:26 |
|---|---|---|
| gitlab-br-bot | coldtom closed MR !1556 (coldtom/update-plugin-tests->master: Remove test dependency on bst-plugins-experimental) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1556 | 08:30 |
| *** Becky has joined #buildstream | 08:30 | |
| *** rdale has joined #buildstream | 08:32 | |
| gitlab-br-bot | BenjaminSchubert approved MR !1566 (juerg/cache-quota->master: Cache quota configuration fixes) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1566 | 08:36 |
| *** rdale has quit IRC | 08:40 | |
| *** rdale has joined #buildstream | 08:41 | |
| benschubert | I struggle to understand why !1552 is needed, a redirection would do the job in _every_ case I can think off. am I missing something? Kinnison ? | 08:42 |
| gitlab-br-bot | MR !1552: Addition of --file option to bst artifact log https://gitlab.com/BuildStream/buildstream/merge_requests/1552 | 08:42 |
| Kinnison | benschubert: It might be nicer to have each log in a separate file, which is the only value I see in it | 08:43 |
| benschubert | Ah, the MR description is not accurate then | 08:45 |
| Kinnison | Indeed | 08:45 |
| Kinnison | Hence my last comment on it | 08:45 |
| benschubert | also, it's not a file anymore, but a dir | 08:45 |
| Kinnison | indeed | 08:45 |
| benschubert | ok, I'll review it then, thanks for the update | 08:45 |
| Kinnison | np | 08:45 |
| Kinnison | thanks | 08:45 |
| *** tme5 has joined #buildstream | 08:59 | |
| tme5 | ready for someone to merge !1557 if you're happy with it | 09:07 |
| gitlab-br-bot | MR !1557: Add in_subprocess pytest mark and modify tests which run in another process to use it https://gitlab.com/BuildStream/buildstream/merge_requests/1557 | 09:07 |
| Kinnison | tme5: it needs a rebase | 09:10 |
| benschubert | tme5: any reason why we ended up copying pytest-forked instead of importing it, since it's already one of our transitive dependencies? (not a blocker, just curious) | 09:11 |
| tme5 | Kinnison, ah sorry, i thought marge did that | 09:13 |
| Kinnison | She does, though I tend to find it's more sensible to rebase it first in case the pipeline fails | 09:13 |
| juergbi | it seems like there is a (new?) flaky test test_fetch_build_checkout[bzr-strict] | 09:15 |
| juergbi | FAILURE import element at build-test-bzr.bst [line 1 column 0]: No files were found inside directory '/' | 09:15 |
| juergbi | anyone familiar with this? | 09:15 |
| Kinnison | Bazaar seems confused at times | 09:15 |
| benschubert | juergbi: definitely hit it a few times, not familiar with it though | 09:15 |
| juergbi | ok, will just retry for now | 09:16 |
| Kinnison | I think the test is leaking stuff into ~/.bazaar | 09:16 |
| tme5 | benschubert, I did consider that, but I thought it didn't really seem right to call internal, undocumented function from the plugins implementation | 09:16 |
| benschubert | tme5: fair point :) | 09:17 |
| tme5 | like, externally all pytest-forked does it offer the --forked option | 09:17 |
| benschubert | tme5: so you can confirm that we get nice tracebacks again? :) | 09:17 |
| tme5 | i thought it would be cleaner to control the implementation ourselves, seeing as it's not many lines | 09:17 |
| tme5 | benschubert, I tested it by adding `assert False` in the tests and the report was indistinguishable from a normal test | 09:18 |
| tme5 | juergbi, I saw that yesterday a few times. I also saw an error from bzr itself right at the bottom of the error log, something about .bazaar not existing | 09:19 |
| benschubert | tme5: cheers, this change is awesome! | 09:19 |
| gitlab-br-bot | BenjaminSchubert approved MR !1557 (tmewett/test-in-subprocess->master: Add in_subprocess pytest mark and modify tests which run in another process to use it) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1557 | 09:20 |
| tme5 | thanks :) | 09:20 |
| *** jonathanmaw has joined #buildstream | 09:25 | |
| *** lachlan has joined #buildstream | 09:30 | |
| *** lachlan has quit IRC | 09:40 | |
| tlater[m] | juergbi: I have two test failures that I think you might be able to help with a little | 09:44 |
| tlater[m] | They're both in this pipeline: https://gitlab.com/BuildStream/buildstream/-/jobs/282460275 | 09:44 |
| tlater[m] | The first is a new check to utils._is_main_process that I'm now somehow triggering though it wasn't before | 09:44 |
| tlater[m] | And the latter seems to be a missing property on a proto | 09:44 |
| tlater[m] | Any idea? | 09:45 |
| *** lachlan has joined #buildstream | 09:46 | |
| juergbi | tlater[m]: the latter seems like a typo in the error handling path | 09:46 |
| juergbi | I can quickly push a commit to fix this | 09:47 |
| tlater[m] | o\ I noticed that yesterday | 09:47 |
| tlater[m] | I'll need to check why we go down that path, that's probably the issue, not the typo | 09:47 |
| juergbi | good point | 09:47 |
| tlater[m] | As for the theck to utils._is_main_process - jennis introduced that recently and said you thought it wouldn't cause any issues for that particular case | 09:48 |
| tlater[m] | Could you explain to me why exactly? I thought any grpc calls on the main thread are a big no-no | 09:48 |
| juergbi | you might need to disable fork for bst artifact show | 09:48 |
| juergbi | they are, if there is any chance of buildstream forking after the grpc call | 09:49 |
| juergbi | if we know we're not forking (anymore), we can call context.disable_fork() to allow grpc calls | 09:49 |
| tlater[m] | Right | 09:49 |
| juergbi | more convenient than having to move everything to a subprocess | 09:49 |
| tlater[m] | Any chance you could really quickly summarize why? :) | 09:50 |
| * tlater[m] is curious | 09:50 | |
| juergbi | why we can't mix gRPC calls and forking? | 09:50 |
| tlater[m] | Yup | 09:50 |
| juergbi | gRPC opens connections and creates threads in the background. neither which works with the fork subprocessing model | 09:50 |
| juergbi | non-main threads are not copied on fork() (and locks held would break the main thread) | 09:51 |
| juergbi | and stream sockets can't be used by multiple processes at the same time | 09:51 |
| tlater[m] | Right | 09:51 |
| Kinnison | tlater[m]: It's entirely because the grpc library doesn't properly interact with python's multiprocessing model and thus cannot cope with the forks | 09:51 |
| tlater[m] | This sounds like it should be ok as long as we don't call more grpc methods in the subprocess | 09:52 |
| Kinnison | tlater[m]: It would be possible to fix, in theory, but noone has been bothered to upstream | 09:52 |
| juergbi | it's generally a Python-independent issue. it's specific to us using fork without exec | 09:52 |
| juergbi | gRPC is supposed to have some 'atfork' support to untangle the forked child process before the actual fork/clone syscall | 09:53 |
| juergbi | however, that seems to be limited or maybe that part doesn't work with Python | 09:53 |
| tlater[m] | Hm, fair enough. I'll need to take a closer look to understand this properly at this at some point. | 09:54 |
| juergbi | tlater[m]: artifact_show() already does disable_fork() if there are fetch remotes | 09:54 |
| juergbi | wondering why that doesn't work | 09:54 |
| tlater[m] | Hm, that might be something I broke then | 09:55 |
| * tlater[m] checks | 09:55 | |
| tlater[m] | Who knows, maybe I forget to register artifact remotes as fetch remotes | 09:55 |
| juergbi | tlater[m]: ah, no, I think the issue is that remote.init() doesn't have the 'fork disabled' exception | 09:56 |
| juergbi | I thought jennis changed that | 09:56 |
| tlater[m] | Ah, that probably got lost in the rebase | 09:56 |
| jennis | No, I just removed the assertion in ArtifactRemote.init() | 09:57 |
| tlater[m] | Oh | 09:57 |
| tlater[m] | juergbi: This should just be a check for context.fork_allowed, right? | 09:57 |
| jennis | Reason is in the commit comment: https://gitlab.com/BuildStream/buildstream/commit/6c7a2eeec7b648870ffb8384a794208b05661c9a | 09:58 |
| juergbi | tlater[m]: eh, is this actually a failure on your latest branch? | 09:58 |
| juergbi | tlater[m]: ah, rebasing issue | 09:58 |
| tlater[m] | Yep | 09:58 |
| juergbi | you moved things to _remote.py but not jennis's fix | 09:58 |
| juergbi | so simply reapply jennis's change | 09:59 |
| tlater[m] | Yep | 09:59 |
| tlater[m] | Should I not bother and just remove the assertion? | 09:59 |
| tlater[m] | Not sure I like that much | 09:59 |
| juergbi | have you read the commit message? | 09:59 |
| tlater[m] | Yes | 10:00 |
| juergbi | is it no longer calling get_local_cas() in all cases? | 10:00 |
| tlater[m] | We don't use get_local_cas() in init(), and probably not in artifact/source caches | 10:00 |
| tlater[m] | We never did the former, I don't think we do anymore in the latter cases. | 10:00 |
| juergbi | right, there will be pure artifact/source remote queries which don't involve CAS, at least not early on | 10:03 |
| juergbi | in that case we'll need to replicate the fork check, I suppose | 10:03 |
| tlater[m] | But that should just be passing a reference to context and checking context.fork_allowed | 10:04 |
| tlater[m] | So it's not that bad :) | 10:04 |
| juergbi | for cas remote context is not acceptable (at least right now) due to bst-artifact-server | 10:05 |
| tlater[m] | Oh, right | 10:05 |
| tlater[m] | Darnit. Well, can always pass in whether forks are allowed as a kwarg | 10:06 |
| juergbi | that's why cascache has its own variable for it | 10:06 |
| tlater[m] | It does? | 10:06 |
| tlater[m] | Is that casd? | 10:06 |
| juergbi | CASCache._fork_disabled | 10:07 |
| juergbi | I don't understand the second question | 10:07 |
| * tlater[m] can't find that variable | 10:08 | |
| tlater[m] | In either case, I'll introduce it, and get all remotes to listen to it. | 10:10 |
| *** lachlan has quit IRC | 10:18 | |
| tme5 | my MR is rebased and passing | 10:28 |
| tlater[m] | \o/ | 10:29 |
| tlater[m] | tme5: mccabe isn't as happy with you | 10:33 |
| tme5 | wh-who? | 10:37 |
| gitlab-br-bot | tlater approved MR !1557 (tmewett/test-in-subprocess->master: Add in_subprocess pytest mark and modify tests which run in another process to use it) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1557 | 10:39 |
| tlater[m] | tme5: Our code quality linter | 10:39 |
| tlater[m] | It's unhappy about the cyclomatic complexity of some of your functions :) | 10:39 |
| Kinnison | One day I'll start a punk rock band called 'Cyclomatic Complexity' | 10:39 |
| tlater[m] | I couldn't spot better ways of doing those things though, so whatever. | 10:39 |
| Kinnison | Sometimes the artificial separations that the 'code quality checker' encourages can reduce readability | 10:40 |
| Kinnison | Also it's unhappy with imported code | 10:41 |
| Kinnison | no point worrying about that | 10:41 |
| tlater[m] | Kinnison: Your first album's song titles will need to reference previous songs. | 10:42 |
| Kinnison | tlater[m]: Nope, album one references album two's songs | 10:42 |
| Kinnison | The first track will be called 'from future import success' | 10:42 |
| tlater[m] | Hehe | 10:42 |
| *** lachlan has joined #buildstream | 10:47 | |
| gitlab-br-bot | juergbi opened MR !1567 (juerg/cas->master: Remove code duplication in casremote.py) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1567 | 11:00 |
| *** lachlan has quit IRC | 11:00 | |
| juergbi | tlater[m]: ^^ fix for attribute error | 11:01 |
| tme5 | do we have a strict min. version dependency on pytest? | 11:04 |
| tlater[m] | juergbi: ta! Shame I just included that in my MR as well ;p | 11:06 |
| tlater[m] | Tsk "No changes to code quality" although you obviously removed code duplication | 11:07 |
| gitlab-br-bot | tlater approved MR !1567 (juerg/cas->master: Remove code duplication in casremote.py) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1567 | 11:07 |
| tlater[m] | tme5: We don't currently. | 11:08 |
| tlater[m] | We could introduce one | 11:08 |
| tlater[m] | tme5: I'll wait for your branch to land btw, so that I can rewrite my tests in the new style. | 11:08 |
| *** julien has joined #buildstream | 11:10 | |
| *** lachlan has joined #buildstream | 11:10 | |
| tme5 | tlater[m], alright, I'll just fix up the last few things | 11:14 |
| *** narispo has quit IRC | 11:16 | |
| *** narispo has joined #buildstream | 11:16 | |
| *** pmilo has joined #buildstream | 11:18 | |
| *** traveltissues has joined #buildstream | 11:20 | |
| tme5 | tlater[m], I don't even know about cognitive complexity thing, i mean the code is a bit complex, but it's not easy to know when the checker will be happy | 11:23 |
| tme5 | i could try and tidy it a bit? | 11:24 |
| tlater[m] | This isn't cognitive complexity, it's cyclomatic for the most part | 11:24 |
| tme5 | oh, well the web thing says cognitive | 11:24 |
| tlater[m] | But as I said, I looked through it and couldn't really spot a much better way to do it. | 11:24 |
| * tlater[m] might have misread | 11:24 | |
| tlater[m] | In any case, I don't think it's grounds for delaying the MR much. I just wanted to point out to you that mccabe exists :) | 11:25 |
| tme5 | ahh ok, thanks haha | 11:25 |
| *** lachlan has quit IRC | 11:28 | |
| *** pmilo has quit IRC | 11:28 | |
| gitlab-br-bot | tlater opened (was WIP) MR !1540 (tlater/cache-endpoints->master: Support separate end points for artifact caches) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1540 | 11:29 |
| tlater[m] | qinusty, jjardon: Do you guys think I could ask you for review in that MR? | 11:37 |
| tlater[m] | Just a quick skim to see if you think the functionality is right, not an in-depth code review :) | 11:37 |
| *** lachlan has joined #buildstream | 11:39 | |
| *** narispo has quit IRC | 11:42 | |
| gitlab-br-bot | jennis opened MR !1568 (jennis/update_push_pull->master: Update artifact push and pull so that they can handle artifact refs) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1568 | 11:47 |
| gitlab-br-bot | traveltissues opened issue #1112 (Follow-up from "Addition of --out option to bst artifact log" - allow artifact log overwriting) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1112 | 11:51 |
| gitlab-br-bot | traveltissues approved MR !1552 (becky/artifact_log_file_option->master: Addition of --out option to bst artifact log) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1552 | 11:54 |
| *** bochecha has joined #buildstream | 12:10 | |
| gitlab-br-bot | marge-bot123 merged MR !1567 (juerg/cas->master: Remove code duplication in casremote.py) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1567 | 12:15 |
| jjardon | tlater[m]: sure, let me take a look | 12:36 |
| *** mohan43u has joined #buildstream | 12:42 | |
| jjardon | tlater[m]: done | 12:44 |
| *** tristan has joined #buildstream | 12:49 | |
| *** phoenix has joined #buildstream | 12:53 | |
| *** phoenix has quit IRC | 12:58 | |
| gitlab-br-bot | traveltissues approved MR !1568 (jennis/update_push_pull->master: Update artifact push and pull so that they can handle artifact refs) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1568 | 13:04 |
| *** tchaik[m] has quit IRC | 13:07 | |
| *** skullone[m] has quit IRC | 13:07 | |
| *** krichter[m] has quit IRC | 13:07 | |
| *** theawless[m] has quit IRC | 13:07 | |
| *** reuben640[m] has quit IRC | 13:07 | |
| *** jjardon[m] has quit IRC | 13:07 | |
| *** segfault3[m] has quit IRC | 13:07 | |
| *** Trevinho[m] has quit IRC | 13:07 | |
| *** rafaelff[m] has quit IRC | 13:07 | |
| *** asingh_[m] has quit IRC | 13:07 | |
| *** mrmcq2u[m] has quit IRC | 13:08 | |
| *** abderrahim[m] has quit IRC | 13:08 | |
| *** dineshdb[m] has quit IRC | 13:08 | |
| *** nielsdg has quit IRC | 13:08 | |
| *** m_22[m] has quit IRC | 13:08 | |
| *** mattiasb has quit IRC | 13:08 | |
| *** dbuch has quit IRC | 13:08 | |
| *** tlater[m] has quit IRC | 13:08 | |
| *** kailueke[m] has quit IRC | 13:08 | |
| *** connorshea[m] has quit IRC | 13:08 | |
| *** pro[m] has quit IRC | 13:08 | |
| *** awacheux[m] has quit IRC | 13:08 | |
| *** albfan[m] has quit IRC | 13:08 | |
| *** gimpnet-irc[m] has quit IRC | 13:08 | |
| *** Demos[m] has quit IRC | 13:08 | |
| *** ssssam[m] has quit IRC | 13:08 | |
| *** doras has quit IRC | 13:08 | |
| *** waltervargas[m] has quit IRC | 13:08 | |
| *** verdre[m] has quit IRC | 13:08 | |
| *** tlater[m] has joined #buildstream | 13:10 | |
| *** julien has left #buildstream | 13:12 | |
| *** julien has joined #buildstream | 13:16 | |
| *** bochecha has quit IRC | 13:20 | |
| gitlab-br-bot | traveltissues approved MR !1552 (becky/artifact_log_file_option->master: Addition of --out option to bst artifact log) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1552 | 13:22 |
| jennis | Thanks for the review traveltissues | 13:23 |
| jennis | Would appreciate one more review/approval before landing :) | 13:23 |
| jennis | all ^ | 13:23 |
| gitlab-br-bot | traveltissues opened issue #1113 (Use fixture for creating artifact shares in testing) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1113 | 13:27 |
| traveltissues | np | 13:27 |
| jennis | I have some failing tests I'm unable to reproduce (even though I'm using buildbox): https://gitlab.com/BuildStream/buildstream/-/jobs/282931933 which fail when we try to checkout an ArtifactElement because: "Error while staging dependencies into a sandbox: 'The BuildBox Sandbox does not support real directories.'". I suspect this is because I've no | 13:28 |
| jennis | t done a very good job configuring a sandbox for an ArtifactElement (https://gitlab.com/BuildStream/buildstream/merge_requests/1565/diffs?commit_id=faf675b13994838792f71f3f3f8677548e8c4a5e#73e0b3efdde524069a4a1d2c1f446e0fae2d6461_122_129), but I'm unsure how I should fix this | 13:28 |
| jennis | wondered if anyone has any ideas? | 13:28 |
| *** lachlan has quit IRC | 13:34 | |
| gitlab-br-bot | tpollard approved MR !1552 (becky/artifact_log_file_option->master: Addition of --out option to bst artifact log) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1552 | 13:40 |
| tme5 | tlater[m], shall I assign marge now? | 13:42 |
| tme5 | to my MR | 13:42 |
| gitlab-br-bot | traveltissues approved MR !1552 (becky/artifact_log_file_option->master: Addition of --out option to bst artifact log) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1552 | 13:44 |
| gitlab-br-bot | traveltissues approved MR !1566 (juerg/cache-quota->master: Cache quota configuration fixes) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1566 | 13:50 |
| *** lachlan has joined #buildstream | 13:56 | |
| gitlab-br-bot | coldtom opened MR !1569 (coldtom/validate-patch->master: plugins/sources/patch.py: Validate keys) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1569 | 13:59 |
| gitlab-br-bot | tpollard approved MR !1569 (coldtom/validate-patch->master: plugins/sources/patch.py: Validate keys) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1569 | 14:13 |
| tme5 | everyone happy with !1557 now? | 14:16 |
| gitlab-br-bot | MR !1557: Add in_subprocess pytest mark and modify tests which run in another process to use it https://gitlab.com/BuildStream/buildstream/merge_requests/1557 | 14:16 |
| gitlab-br-bot | beckyella16 opened MR !1570 (becky/cleanup_artifact_tests->master: Cleanup: splitting artifact.py into separate files) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1570 | 14:33 |
| gitlab-br-bot | traveltissues approved MR !1552 (becky/artifact_log_file_option->master: Addition of --out option to bst artifact log) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1552 | 14:47 |
| *** lachlan has quit IRC | 14:51 | |
| *** tristan has quit IRC | 14:54 | |
| gitlab-br-bot | traveltissues approved MR !1566 (juerg/cache-quota->master: Cache quota configuration fixes) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1566 | 14:56 |
| *** narispo has joined #buildstream | 15:05 | |
| *** mohan43u has quit IRC | 15:12 | |
| *** mohan43u has joined #buildstream | 15:12 | |
| *** lachlan has joined #buildstream | 15:17 | |
| *** lachlan has quit IRC | 15:45 | |
| *** phoenix has joined #buildstream | 15:52 | |
| *** slaf has quit IRC | 15:52 | |
| *** slaf has joined #buildstream | 15:52 | |
| *** phoenix has quit IRC | 15:53 | |
| gitlab-br-bot | jennis approved MR !1570 (becky/cleanup_artifact_tests->master: Cleanup: splitting artifact.py into separate files) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1570 | 16:07 |
| *** tristan has joined #buildstream | 16:15 | |
| juergbi | bzr flaky tests are really painful | 16:15 |
| coldtom | +1 | 16:15 |
| juergbi | why is this suddenly happening? | 16:15 |
| juergbi | can we revert whatever did this or disable bzr? | 16:16 |
| juergbi | benschubert: is there any chance the bzr test issue is related to !1537 ? not sure when it started happening, but I don't see what other change might have had an impact on source plugin tests | 16:17 |
| gitlab-br-bot | MR !1537: testing/sources: Automatically register plugin sources https://gitlab.com/BuildStream/buildstream/merge_requests/1537 | 16:17 |
| benschubert | juergbi: I don't see why that would be the case | 16:18 |
| benschubert | unless tests were actually broken | 16:18 |
| juergbi | bzr is still in the core repo, though, so it shouldn't have had an impact, right? | 16:19 |
| *** lachlan has joined #buildstream | 16:19 | |
| benschubert | correct | 16:19 |
| juergbi | wondering whether it might have always been flaky but maybe that MR changed test order which triggers it much more frequently? | 16:20 |
| benschubert | possible | 16:20 |
| gitlab-br-bot | tpollard opened (was WIP) MR !1550 (tpollard/notificationhandler->master: Stream - Scheduler notification handler) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1550 | 16:22 |
| tpollard | would really appreciate eyes on that MR please ^ :) | 16:23 |
| juergbi | I consider CI unusable right now | 16:37 |
| juergbi | what's the best way to disable bzr tests? or ideally, xfail them? | 16:38 |
| juergbi | and then open an issue to get this fixed properly | 16:38 |
| *** traveltissues has quit IRC | 16:38 | |
| *** tme5 has quit IRC | 16:43 | |
| tpollard | do we know if it's always the same target docker/dockers that is failing it? | 16:43 |
| tpollard | I wonder if it's a specific host bzr issue | 16:43 |
| tpollard | We've had different versions of git break tests in the past | 16:43 |
| juergbi | I see it mainly on Fedora-based docker images | 16:44 |
| juergbi | but I think I also saw it at least once on Ubuntu 18.04 | 16:44 |
| juergbi | ah, I think I found the bug in bzr | 16:50 |
| juergbi | but given that bzr seems unmaintained, we have to work around it, I suppose | 16:50 |
| tpollard | Lovely.... | 16:51 |
| *** lachlan has quit IRC | 16:52 | |
| *** lachlan has joined #buildstream | 16:59 | |
| benschubert | juergbi: it's a bzr bug? Oo | 17:02 |
| *** lachlan has quit IRC | 17:02 | |
| juergbi | benschubert: yes, it seems that way. bzr has a function ensure_config_dir_exists() which does | 17:03 |
| juergbi | if not os.path.isdir(path): | 17:03 |
| juergbi | os.mkdir(path) | 17:03 |
| juergbi | i.e., race condition | 17:03 |
| juergbi | either we need to ensure we use per-test (or at least per python instance) config directory | 17:04 |
| benschubert | that ought to be in ~/ right? is it respecting XDG_*? Can we add this in the tests to ensure they are isolated? | 17:04 |
| juergbi | or create .bazaar in our code (race free) before calling bzr the first time | 17:04 |
| juergbi | it is supposed to be respecting XDG_CONFIG_HOME | 17:04 |
| juergbi | however, I'm not sure why this doesn't work because it seems we already set that for per-test directories | 17:05 |
| juergbi | or is it possible that aspect is broken for source plugin tests? | 17:05 |
| benschubert | That's possible | 17:08 |
| juergbi | ah, it's using XDG_CONFIG_HOME only if $XDG_CONFIG_HOME/bazaar exists | 17:21 |
| juergbi | otherwise it uses ~/.bazaar | 17:21 |
| juergbi | :-/ | 17:21 |
| *** tpollard has quit IRC | 17:24 | |
| benschubert | oh well, fixture(autouse=True) ? :'D | 17:25 |
| benschubert | to create that directory | 17:25 |
| *** rdale has quit IRC | 17:26 | |
| juergbi | benschubert: I'm currently thinking along the lines of setting HOME to {envtmpdir}/home to cover other potential uses of the home directory | 17:38 |
| juergbi | or do you see an issue with that? | 17:38 |
| juergbi | (in tox.ini) | 17:38 |
| benschubert | that seems good a-priori :) | 17:39 |
| juergbi | tests seem to pass locally, so let's check CI | 17:41 |
| gitlab-br-bot | juergbi opened MR !1571 (juerg/tox-home->master: tox.ini: Set HOME environment variable) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1571 | 17:47 |
| gitlab-br-bot | marge-bot123 closed issue #1106 (Addition of '--file' option to 'bst artifact log') on buildstream https://gitlab.com/BuildStream/buildstream/issues/1106 | 17:52 |
| gitlab-br-bot | marge-bot123 merged MR !1552 (becky/artifact_log_file_option->master: Addition of --out option to bst artifact log) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1552 | 17:52 |
| *** jonathanmaw has quit IRC | 17:54 | |
| *** cgmcintyre[m] has joined #buildstream | 18:12 | |
| *** connorshea[m] has joined #buildstream | 18:13 | |
| *** asingh_[m] has joined #buildstream | 18:13 | |
| *** awacheux[m] has joined #buildstream | 18:13 | |
| *** abderrahim[m] has joined #buildstream | 18:13 | |
| *** albfan[m] has joined #buildstream | 18:13 | |
| *** Demos[m] has joined #buildstream | 18:13 | |
| *** doras has joined #buildstream | 18:13 | |
| *** dbuch has joined #buildstream | 18:13 | |
| *** Trevinho[m] has joined #buildstream | 18:13 | |
| *** dineshdb[m] has joined #buildstream | 18:13 | |
| *** gimpnet-irc[m] has joined #buildstream | 18:13 | |
| *** jjardon[m] has joined #buildstream | 18:13 | |
| *** kailueke[m] has joined #buildstream | 18:13 | |
| *** pro[m] has joined #buildstream | 18:13 | |
| *** m_22[m] has joined #buildstream | 18:13 | |
| *** rafaelff[m] has joined #buildstream | 18:13 | |
| *** theawless[m] has joined #buildstream | 18:13 | |
| *** ssssam[m] has joined #buildstream | 18:13 | |
| *** waltervargas[m] has joined #buildstream | 18:13 | |
| *** segfault3[m] has joined #buildstream | 18:13 | |
| *** mattiasb has joined #buildstream | 18:13 | |
| *** krichter[m] has joined #buildstream | 18:13 | |
| *** skullone[m] has joined #buildstream | 18:13 | |
| *** mrmcq2u[m] has joined #buildstream | 18:13 | |
| *** tchaik[m] has joined #buildstream | 18:13 | |
| *** reuben640[m] has joined #buildstream | 18:13 | |
| *** verdre[m] has joined #buildstream | 18:13 | |
| *** nielsdg has joined #buildstream | 18:13 | |
| *** phoenix has joined #buildstream | 18:58 | |
| *** bochecha has joined #buildstream | 19:55 | |
| *** johnward is now known as jward | 20:03 | |
| *** cs-shadow has quit IRC | 21:16 | |
| gitlab-br-bot | cs-shadow opened (was WIP) MR !1459 (chandan/mypy->master: Add type hints to public API code) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1459 | 22:13 |
| *** phoenix has quit IRC | 22:27 | |
| *** bochecha has quit IRC | 23:17 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!