| *** awacheux[m] has joined #buildstream | 00:13 | |
| *** Trevinho[m] has joined #buildstream | 01:05 | |
| *** tlater[m] has joined #buildstream | 01:19 | |
| *** dineshdb[m] has joined #buildstream | 01:26 | |
| *** pro[m] has joined #buildstream | 01:55 | |
| *** nielsdg has joined #buildstream | 02:02 | |
| *** ssssam[m] has joined #buildstream | 02:02 | |
| *** m_22[m] has joined #buildstream | 02:22 | |
| gitlab-br-bot | juergbi opened MR !1576 (juerg/synthetic-file-index->master: node.pyx: _SYNTHETIC_FILE_INDEX must not be module-private) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1576 | 05:47 |
|---|---|---|
| Kinnison | A direct back port won't be possible since it was done when we'd already changed the yaml internals a lot | 06:03 |
| Kinnison | But it shouldn't be complies to replicate on the branch | 06:03 |
| Kinnison | s/complies/complicated/ | 06:04 |
| *** narispo has quit IRC | 07:53 | |
| *** narispo has joined #buildstream | 07:54 | |
| gitlab-br-bot | juergbi opened MR !1577 (juerg/fork->master: Replace safeguard for fork with multiple threads) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1577 | 07:57 |
| juergbi | Kinnison: if you're in the mood for reviewing more flaky test fixes: https://gitlab.com/BuildStream/buildstream/merge_requests/1577 | 08:00 |
| * Kinnison loads | 08:01 | |
| jjardon | bochecha: probably yes; can you open an issue please? | 08:01 |
| gitlab-br-bot | valentindavid opened MR !1578 (valentindavid/update-requirements-1.4->bst-1: Update requirements for 1.4) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1578 | 08:12 |
| Kinnison | juergbi: How expensive is constructing a CASRemote each time it's used in that context manager mechanism> | 08:13 |
| Kinnison | ? | 08:13 |
| juergbi | Kinnison: hm, the branch is not creating more CASRemote instances, is it? | 08:14 |
| juergbi | it just uses it as a context manager to ensure its resources are released | 08:14 |
| * Kinnison expands up more context and realises you weren't removing instantiation from __init__ methods | 08:15 | |
| juergbi | in general, CASRemote construction itself is cheap but .init() is relatively expensive (remote check) | 08:15 |
| * Kinnison nods | 08:15 | |
| *** rdale has joined #buildstream | 08:17 | |
| Kinnison | juergbi: One nit in the remote execution bit | 08:17 |
| Kinnison | You create `channel` inside the `with CASRemote` block, and then use it afterwards -- is that normally accepted in Python coding, or should really the use of it be scoped inside the `with` block too? I realise variable scoping in Python is a bit 'special' but it tweaks my 'variables should be lexically scoped ideally' nerve | 08:18 |
| juergbi | Kinnison: which method/line? | 08:20 |
| Kinnison | sandboxremote.py line 451 | 08:21 |
| juergbi | that's not created in `with CASRemote` block | 08:21 |
| juergbi | it's in a different method | 08:21 |
| Kinnison | Oh FFS, gitlab is doing a VERY bad job at giving me context | 08:22 |
| Kinnison | I apologise | 08:22 |
| Kinnison | Today I am clearly not good at reviewing :( | 08:22 |
| juergbi | yes, context is often a pain on gitlab, I typically review non-trivial things locally | 08:22 |
| Kinnison | Mmm but then giving feedback is a pig to find how to feed it back in | 08:22 |
| * Kinnison misses code-review-on-mailing-list | 08:22 | |
| gitlab-br-bot | BenjaminSchubert approved MR !1459 (chandan/mypy->master: Add type hints to public API code) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1459 | 08:22 |
| juergbi | yes, have to switch between local diff and browser for this :-/ | 08:23 |
| Kinnison | Okay, over-all I dislike the number-of-threads thing, but I get why you need it; otherwise 👍 | 08:24 |
| *** traveltissues has joined #buildstream | 08:25 | |
| gitlab-br-bot | BenjaminSchubert approved MR !1576 (juerg/synthetic-file-index->master: node.pyx: _SYNTHETIC_FILE_INDEX must not be module-private) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1576 | 08:25 |
| gitlab-br-bot | danielsilverstone-ct approved MR !1577 (juerg/fork->master: Replace safeguard for fork with multiple threads) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1577 | 08:25 |
| juergbi | yes, I was also not too happy that I needed this but I don't really see another way for xdist/execnet. thanks for the review | 08:26 |
| gitlab-br-bot | marge-bot123 merged MR !1459 (chandan/mypy->master: Add type hints to public API code) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1459 | 08:26 |
| juergbi | sensible error messages instead of hangs in case of future background threads should be worth it | 08:26 |
| * Kinnison nods | 08:27 | |
| bochecha | jjardon: will do | 08:42 |
| benschubert_ | juergbi: agreed, I do like the PR in general :) | 08:45 |
| jjardon | bochecha: FYI: https://gitlab.com/BuildStream/buildstream/merge_requests/1578/ | 09:06 |
| Kinnison | valentind: That MR confuses me | 09:08 |
| Kinnison | valentind: You say in the description we still need 0.15 | 09:08 |
| Kinnison | valentind: but the MR updates to 0.16 | 09:08 |
| Kinnison | valentind: Also you probably want to say >= 0.16, <= 0.17 (or whatever the pip syntax is) | 09:08 |
| valentind | "required" | 09:08 |
| valentind | That was in the past. | 09:09 |
| *** ahmed89 has joined #buildstream | 09:09 | |
| valentind | Kinnison, Do we know that it will not work with 0.17? | 09:09 |
| Kinnison | valentind: I think ruamel explicitly don't guarantee things across minor updates yet | 09:09 |
| Kinnison | valentind: But if you're happy with it as-is then I wouldn't block it :D | 09:10 |
| valentind | Kinnison, I do not think we should be strict here. We can always install another version. The problem was to forbid 0.16. We should not do that. | 09:10 |
| Kinnison | Okay | 09:10 |
| * Kinnison withdraws conern | 09:11 | |
| valentind | And it was known that later 0.15 was broken for us. | 09:11 |
| Kinnison | concern | 09:11 |
| Kinnison | Everything went bangsplat inside ruamel though | 09:12 |
| Kinnison | in your pipeline | 09:12 |
| *** jonathanmaw has joined #buildstream | 09:13 | |
| valentind | I have to look at why it fails. | 09:14 |
| *** ahmed89 has quit IRC | 09:18 | |
| *** ahmed89 has joined #buildstream | 09:19 | |
| *** narispo has quit IRC | 09:21 | |
| *** narispo has joined #buildstream | 09:21 | |
| tpollard | anyone having an issue with running the tests on master and virtualenv? | 09:21 |
| bochecha | valentind: updating to 0.16 like this doesn't require any change in the code? I thought 0.16 was incompatible with 0.15 and that was why bst was blocked :-/ | 09:22 |
| tpollard | I'm getting "ERROR: py37: InvocationError for command /usr/bin/python3 -m virtualenv --python /usr/bin/python3 py37 (exited with code 1)", I've reinstalled virtualenv with pip too | 09:25 |
| benschubert_ | bochecha: that was what the ruamel developper were saying in their docs, turns out that the way we used it didn't break... yet | 09:27 |
| bochecha | benschubert_: ah, cool | 09:28 |
| bochecha | a happy accident then :) | 09:28 |
| qinusty | Can anyone provide some insight why buildstream requests the output directory that it does for remote sandbox builds? Currently it appears to request the entire contents of the sandbox rather than just the working directory where the elements, project.conf and outputs will be | 09:31 |
| valentind | bochecha, apparently we needed one line fix. | 09:42 |
| *** raoul has joined #buildstream | 09:44 | |
| * tpollard switches away from pip3 venv | 09:45 | |
| juergbi | qinusty: this is something that we should try to optimize. however, in some cases we need the whole tree (e.g., integration commands in one action followed by actual build in a separate action) and with buildbox-fuse it's fairly cheap | 09:51 |
| juergbi | without fuse it might not be that cheap, so we should look into optimizations. but it should still work | 09:52 |
| qinusty | Okay, I'm just curious due to some of the nuances of Buildbarn which place some arbitrary restrictions on file paths :/ | 09:52 |
| qinusty | If whole tree is sometimes required I will find a workaround | 09:53 |
| juergbi | qinusty: is the issue still that buildbarn doesn't support `../*` paths at all or is there a separate issue with requesting the whole tree? | 09:54 |
| qinusty | I'm working on an upstream fix for the issue | 09:55 |
| jjardon | valentind: mmm, is it needed to install gcc to install bst 1.4? https://gitlab.gnome.org/GNOME/gnome-build-meta/-/jobs/416627 | 09:56 |
| benschubert_ | jjardon: ujson will require GCC yes | 09:56 |
| juergbi | qinusty: it might not be that difficult to modify buildstream to start with a working directory at the input root and then issue a `cd` | 09:56 |
| juergbi | we already insert `cd` for following commands that require different working directory | 09:57 |
| qinusty | The issue is more about the output directory | 09:57 |
| qinusty | As buildstream wants the entire tree | 09:57 |
| juergbi | well, if the working directory was the top-level, the output directory could then simply be '.' | 09:57 |
| juergbi | i.e., no need for '..' anymore | 09:57 |
| qinusty | Ah yes | 09:57 |
| qinusty | that could work | 09:57 |
| * qinusty hadn't considered fixing buildstream since buildbarn doesn't meet the spec | 09:57 | |
| Kinnison | Surely buildbarn should be fixed to be compliant to the protocol though | 09:57 |
| qinusty | ^^^ | 09:58 |
| juergbi | yes, I agree | 09:58 |
| jjardon | ok, thanks benschubert_ | 09:58 |
| valentind | jjardon, maybe install ujson on the docker image. | 09:58 |
| qinusty | However that could be a temporary fix juergbi. Effectively bringing buildstream into similar behaviour to bazel | 09:59 |
| * Kinnison fears that if we do that, there'll be no impetus to get buildbarn fixed | 09:59 | |
| Kinnison | and then someone else will hit the same problem | 09:59 |
| * qinusty will hack around for another few hours on buildbarn | 10:00 | |
| *** ahmed89 has quit IRC | 10:01 | |
| qinusty | If I can't see a fix which would produce a viable upstream patch, then I'll attempt a buildstream patch for the issue and detail the upstream issue @ buildbarn | 10:01 |
| *** ahmed89 has joined #buildstream | 10:01 | |
| jjardon | valentind: yeah, I will simply use the same docker image as in fdsdk | 10:05 |
| jjardon | coldtom: is there anything missing at https://gitlab.com/BuildStream/bst-plugins-experimental/merge_requests/27 ? | 10:07 |
| Kinnison | qinusty: Please @ me if you do any github stuff related, so I can keep track. @danielsilverstone-ct on github | 10:07 |
| coldtom | jjardon: not as far as I'm aware | 10:07 |
| jjardon | benschubert_: do you have any other comment about https://gitlab.com/BuildStream/bst-plugins-experimental/merge_requests/27/ ? | 10:08 |
| qinusty | There is an issue raised currently at https://github.com/buildbarn/bb-remote-execution/issues/18 Kinnison. Will @ you if I make an MR or something | 10:08 |
| Kinnison | ta | 10:08 |
| coldtom | what is the behaviour of buildstream when there are multiple of the same junction? e.g. i junction project A and project B, but project B junctions project A | 10:09 |
| coldtom | does project B build against the version of project A i have junctioned? | 10:09 |
| benschubert_ | jjardon: let me have a look | 10:09 |
| juergbi | cs-shadow: interesting approach. definitely more convenient than clicking through the artifact | 10:10 |
| * cs-shadow is glad someone else also finds it useful :) | 10:11 | |
| gitlab-br-bot | valentindavid merged MR !1578 (valentindavid/update-requirements-1.4->bst-1: Update requirements for 1.4) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1578 | 10:15 |
| benschubert_ | jjardon: review done, there is still quite some work to do :) | 10:19 |
| jjardon | benschubert_: tvm :) | 10:20 |
| gitlab-br-bot | valentindavid opened MR !1579 (valentindavid/news-1.4.1->bst-1: Update NEWS) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1579 | 10:24 |
| *** jonathanmaw has quit IRC | 10:24 | |
| *** phildawson_ has quit IRC | 10:24 | |
| *** tpollard has quit IRC | 10:24 | |
| *** tiagogomes has quit IRC | 10:24 | |
| *** traveltissues has quit IRC | 10:24 | |
| *** raoul has quit IRC | 10:24 | |
| *** tpollard has joined #buildstream | 10:26 | |
| *** jonathanmaw has joined #buildstream | 10:26 | |
| *** traveltissues has joined #buildstream | 10:26 | |
| *** raoul has joined #buildstream | 10:26 | |
| *** tiagogomes has joined #buildstream | 10:27 | |
| *** phildawson has joined #buildstream | 10:27 | |
| jjardon | bochecha: ^ | 10:28 |
| bochecha | jjardon: yes? | 10:32 |
| jjardon | 1.4.1 coming with support for ruamel >= 0.16 | 10:33 |
| bochecha | jjardon: already 1.4.1? o_O | 10:34 |
| bochecha | or you mean valentind's MR was merged so when it comes out 1.4.1 will have that? | 10:34 |
| gitlab-br-bot | marge-bot123 merged MR !1577 (juerg/fork->master: Replace safeguard for fork with multiple threads) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1577 | 10:35 |
| valentind | bochecha, We are preparing the release. | 10:35 |
| valentind | With just that fix. | 10:35 |
| jjardon | depend on a specific unstable version of that library is annoying for other distros as well | 10:39 |
| benschubert_ | jjardon: I think ruamel 0.15.54 that we had before is still compatible, we could whiletelist it too maybe? | 10:43 |
| jjardon | benschubert_: sorry, not sure I understand what you mean by that? | 10:58 |
| jjardon | any idea why https://docs.buildstream.build/ still shows 1.2.8 as the latest? | 10:58 |
| benschubert_ | jjardon: there is no reason for us to require ruamel > 0.16, we can also allow the previous version we had (0.15.54) | 10:59 |
| benschubert_ | which might help other distros | 10:59 |
| jjardon | benschubert_: I think a code change is needed , so not sure we can allow both | 10:59 |
| benschubert_ | jjardon: your PR to bump to 0.16: https://gitlab.com/BuildStream/buildstream/merge_requests/1506/diffs | 11:00 |
| benschubert_ | doesn't seem so :) | 11:00 |
| jjardon | benschubert_: ah yeah, in master was totally fine | 11:00 |
| jjardon | benschubert_: do you want me to revert that? | 11:01 |
| benschubert_ | On master? No, that perfectly fine on master :) | 11:01 |
| benschubert_ | For bst 1.4, I will let other answer, I don't have problems, we are working from master branch internally | 11:01 |
| gitlab-br-bot | jjardon opened MR !1580 (jjardon/badge->master: README.rst: Point badge to latest bst 1.x release) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1580 | 11:02 |
| jjardon | benschubert_: seems this small change was required when bumping the required version: https://gitlab.com/BuildStream/buildstream/merge_requests/1578/diffs#736a3dbaba7ae3029bf6fe19c22b6d3eed42789e_156_155 | 11:03 |
| benschubert_ | jjardon: which does not break previous version I would think? | 11:05 |
| gitlab-br-bot | marge-bot123 merged MR !1579 (valentindavid/news-1.4.1->bst-1: Update NEWS) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1579 | 11:05 |
| jjardon | benschubert_: ah, I see what you mean, I will prepare a MR | 11:15 |
| jjardon | benschubert_: used to be ruamel.yaml < 0.15.52, you mean to change that to ruamel.yaml > 0.15.52 ? | 11:17 |
| benschubert_ | ah no, that would not work | 11:18 |
| benschubert_ | but maybe <0.15.52 | >0.16<0.17 ? | 11:19 |
| jjardon | mmm, Is that syntax valid on requirements.txt? "," is "AND" but not sure about the "|" | 11:23 |
| jjardon | cs-shadow: hey, I think you have access to the pip account? can you upload 1.4.1 there please? | 11:23 |
| benschubert_ | jjardon: I'm not sure actually, would have to check the spec ;) | 11:25 |
| cs-shadow | jjardon: can do. Unfortunately the pypi keys are on my personal laptop so I'd be able to do that later today once I get home | 11:26 |
| *** traveltissues has quit IRC | 11:26 | |
| *** raoul has quit IRC | 11:26 | |
| * cs-shadow will look into automating this on BuildStream repo similar to some of the others | 11:26 | |
| *** tpollard has quit IRC | 11:26 | |
| *** jonathanmaw has quit IRC | 11:27 | |
| *** phildawson has quit IRC | 11:27 | |
| *** tiagogomes has quit IRC | 11:27 | |
| *** tiagogomes has joined #buildstream | 11:28 | |
| *** traveltissues has joined #buildstream | 11:28 | |
| *** raoul has joined #buildstream | 11:28 | |
| *** tpollard has joined #buildstream | 11:29 | |
| *** jonathanmaw has joined #buildstream | 11:29 | |
| *** phildawson has joined #buildstream | 11:30 | |
| jjardon | cs-shadow: ok, thanks! | 11:32 |
| juergbi | HTTP test server / netrc tests still causing issues. let's simplify that code | 11:44 |
| gitlab-br-bot | juergbi opened MR !1581 (juerg/http-test-server->master: tests/testutils/http_server.py: Drop queue to avoid lingering thread) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1581 | 11:48 |
| bochecha | valentind: oh wow, don't feel like you have to release just for me /o\ | 12:29 |
| *** ahmed89 has quit IRC | 12:49 | |
| * tlater[m] wonders why jquery found its way into his buildstream directory | 13:09 | |
| tlater[m] | docs? | 13:09 |
| tlater[m] | Eww, coverage.py includes it in its package | 13:10 |
| gitlab-br-bot | marge-bot123 merged MR !1580 (jjardon/badge->master: README.rst: Point badge to bst-1 branch) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1580 | 13:14 |
| jjardon | would it be possible to have a 1.9.x snapshot of current master? last one was 1.3.0 and it can be a bit confused now that 1.4.0 is out | 13:52 |
| valentind | There was supposed to be an IRC meeting now according to my calendar. | 14:03 |
| valentind | But there is nobody around. | 14:03 |
| valentind | Is it canceled? | 14:04 |
| Kinnison | Laurence said he was unable to run it as he is in hospital | 14:05 |
| Kinnison | but I don't think anyone else stepped up to run it | 14:05 |
| valentind | Also tristan is not around | 14:06 |
| valentind | So I suppose it is canceled. | 14:06 |
| *** traveltissues has quit IRC | 14:29 | |
| *** raoul has quit IRC | 15:25 | |
| ironfoot | I'm trying to use the overlap-whitelist configuration | 15:28 |
| ironfoot | but I see: https://paste.gnome.org/p58zgkufe | 15:28 |
| *** traveltissues has joined #buildstream | 15:38 | |
| juergbi | ironfoot: in 1.x or master? | 15:43 |
| juergbi | there was a fix in master to support/expect absolute paths but I'm not sure whether that was ever backported | 15:43 |
| juergbi | if you're on 1.x, you may have to remove the leading slash | 15:44 |
| juergbi | but in that case we should probably backport the fix. see !968 | 15:44 |
| gitlab-br-bot | MR !968: Allow absolute paths in whitelist https://gitlab.com/BuildStream/buildstream/merge_requests/968 | 15:44 |
| ironfoot | 1.2.7 | 15:44 |
| ironfoot | aha, I'll do that | 15:44 |
| ironfoot | (remove the leading slash) | 15:44 |
| ironfoot | that worked, thanks juergbi! | 15:48 |
| ironfoot | I feel like backporting a breaking change could be difficult | 15:48 |
| juergbi | ironfoot: as per above MR, in 1.x we would support both for backward compatibility | 15:50 |
| gitlab-br-bot | marge-bot123 merged MR !1581 (juerg/http-test-server->master: tests/testutils/http_server.py: Drop queue to avoid lingering thread) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1581 | 15:59 |
| *** raoul has joined #buildstream | 16:32 | |
| gitlab-br-bot | traveltissues opened issue #1118 (ProvenanceInformation has no attribute `toplevel`) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1118 | 16:46 |
| *** bochecha has quit IRC | 16:48 | |
| gitlab-br-bot | traveltissues opened MR !1582 (traveltissues/1118->master: Fix typo in node attribute) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1582 | 16:48 |
| gitlab-br-bot | BenjaminSchubert approved MR !1582 (traveltissues/1118->master: Fix typo in node attribute) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1582 | 16:49 |
| traveltissues | ty benschubert_ | 16:50 |
| benschubert_ | traveltissues: good catch :) | 16:51 |
| benschubert_ | I'm surprised it even compiled... | 16:51 |
| traveltissues | i ran into while debugging workspaces via cas things | 16:51 |
| traveltissues | i was also surprised | 16:52 |
| tlater[m] | grmbl, I never got to figuring out how our docs are supposed to be published | 16:55 |
| * tlater[m] needs to reverse engineer this bash script | 16:56 | |
| *** traveltissues has quit IRC | 16:56 | |
| *** jonathanmaw has quit IRC | 17:10 | |
| *** bochecha has joined #buildstream | 19:18 | |
| bochecha | heh, now that I can build 1.4.1 in Rawhide against ruamel.yaml 0.16… I can't build it in Fedora 30 which has 0.15 of course! | 19:30 |
| bochecha | I'll just drop the version requirements for the package, since bst is clearly compatible with both versions | 19:31 |
| bochecha | alright, 1.4.1 is on its way in Rawhide, F31 and F30 | 19:55 |
| jjardon | cool! thanks bochecha | 20:29 |
| *** bochecha has quit IRC | 22:01 | |
| *** rdale has quit IRC | 23:00 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!