| *** narispo has quit IRC | 06:23 | |
| *** narispo has joined #buildstream | 06:25 | |
| *** narispo has quit IRC | 06:34 | |
| *** narispo has joined #buildstream | 06:35 | |
| *** mohan43u has quit IRC | 06:37 | |
| *** mohan43u has joined #buildstream | 06:39 | |
| *** narispo has quit IRC | 06:54 | |
| *** narispo has joined #buildstream | 06:54 | |
| *** traveltissues has joined #buildstream | 08:47 | |
| *** coldtom4 is now known as coldtom | 09:06 | |
| *** santi has joined #buildstream | 09:54 | |
| *** tpollard has joined #buildstream | 09:59 | |
| *** jonathanmaw has joined #buildstream | 10:16 | |
| coldtom | anyone against merging https://gitlab.com/BuildStream/buildstream-docker-images/-/merge_requests/157 ? | 10:24 |
|---|---|---|
| benschubert | coldtom: I actually had one comment still, just noticed it today | 10:26 |
| *** phildawson-ct has joined #buildstream | 10:45 | |
| *** phildawson has quit IRC | 10:47 | |
| *** cs-shadow has joined #buildstream | 10:53 | |
| benschubert | WSalmon: one amendment to what I said at FOSDEM about breaking changes on plugins, there will be at least one, for adapting for the RemoteAssetAPI (Formerly FetchAPI) | 11:07 |
| WSalmon | ah, thanks for the heads up benschubert | 11:13 |
| benschubert | WSalmon: that would probably only affect source plugins and should not be a huge hard change | 11:14 |
| WSalmon | ack benschubert | 11:14 |
| benschubert | coldtom: thanks :) Do you have merge rights or want me to merge? | 11:16 |
| WSalmon | i had assumed that remote things could be don so as to not impinge on the plug-ins them selves only how they are invoked, but i can see that the subtleties of how they work especially around sources could lead to breaks | 11:16 |
| coldtom | i have merge rights for that repo, thanks benschubert! | 11:17 |
| WSalmon | ok, it seems no one moved coldtom in to the merge rights group once his first commits landed could juergbi or some one add him to the list? | 11:18 |
| WSalmon | i assume we still work like that | 11:18 |
| *** traveltissues has quit IRC | 11:18 | |
| *** traveltissues has joined #buildstream | 11:19 | |
| juergbi | WSalmon: buildstream or docker-images merge rights are missing? | 11:20 |
| WSalmon | i think he has docker but should have buildstream | 11:20 |
| coldtom | ^ that is correct | 11:20 |
| WSalmon | *should have both | 11:20 |
| juergbi | coldtom: fixed | 11:25 |
| coldtom | ta juergbi | 11:25 |
| WSalmon | thanks juergbi | 11:30 |
| *** narispo has quit IRC | 11:31 | |
| *** narispo has joined #buildstream | 11:41 | |
| *** traveltissues has quit IRC | 12:18 | |
| *** narispo has quit IRC | 12:20 | |
| *** narispo has joined #buildstream | 12:20 | |
| *** narispo has quit IRC | 12:59 | |
| *** narispo has joined #buildstream | 13:01 | |
| *** lachlan has joined #buildstream | 13:02 | |
| *** lachlan has quit IRC | 13:04 | |
| *** lachlan has joined #buildstream | 13:04 | |
| *** traveltissues has joined #buildstream | 13:12 | |
| *** lachlan has quit IRC | 13:15 | |
| *** lachlan has joined #buildstream | 13:16 | |
| *** lachlan has quit IRC | 13:19 | |
| *** narispo has quit IRC | 13:26 | |
| *** narispo has joined #buildstream | 13:27 | |
| *** phildawson-ct has quit IRC | 13:44 | |
| *** lachlan has joined #buildstream | 13:47 | |
| *** phildawson-ct has joined #buildstream | 13:48 | |
| *** lachlan has quit IRC | 13:56 | |
| *** phildawson-ct has quit IRC | 14:15 | |
| *** phildawson-ct has joined #buildstream | 14:15 | |
| *** phildawson-ct has quit IRC | 14:26 | |
| *** phildawson-ct has joined #buildstream | 14:27 | |
| juergbi | oh, we have duplicate but outdated tar plugin in bst-plugins-experimental :-/ | 14:31 |
| juergbi | that was very confusing | 14:31 |
| *** phildawson-ct has quit IRC | 14:32 | |
| *** phildawson-ct has joined #buildstream | 14:32 | |
| juergbi | benschubert: I assume ^^^ is temporary but apparently it was incorrect for more than two months | 14:34 |
| juergbi | is there something we can improve to avoid this happening again? | 14:35 |
| coldtom | the problem is that the tar plugin is massively wrapped up with the rest of the test suite i believe, so the MR to migrate it out of the main repo never landed | 14:38 |
| coldtom | for example, the docs also depend on it | 14:38 |
| juergbi | ok but duplicate code for months is not exactly a solution either | 14:40 |
| juergbi | https://gitlab.com/BuildStream/bst-plugins-experimental/-/merge_requests/81 | 14:40 |
| coldtom | agreed, i think there was talk of a decorator to allow external plugin deps for some tests to allow this | 14:41 |
| juergbi | coldtom: I'm a bit out of the loop with the external plugin work. do I need to bump a version in buildstream git (tox.ini?) to get the fixed plugins after my MR is merged? | 14:42 |
| juergbi | or does it use plugins-experimental master in buildstream master CI? | 14:43 |
| cs-shadow | IMHO we shouldn't merge either MR until both MRs are ready (i.e. the MR on plugins repo and the buildstream MR). | 14:43 |
| cs-shadow | I think we've been too eager with the plugin move lately | 14:43 |
| juergbi | I agree | 14:43 |
| coldtom | +1 | 14:43 |
| juergbi | shall we rip tar out in that case? | 14:43 |
| cs-shadow | juergbi: both buildstream and plugins run against a fixed tag of each other . For buildstream, it's captured here: https://gitlab.com/BuildStream/buildstream/-/blob/master/tox.ini#L11 | 14:44 |
| juergbi | or will that cause any problems? | 14:44 |
| juergbi | cs-shadow: ok, I guessed correctly, then. thanks | 14:44 |
| cs-shadow | I'd personally be fine with removing tar from plugins-experimental since it's not solving anything at this stage | 14:45 |
| cs-shadow | the only downside I can think of is if someone started depending on that version of it | 14:45 |
| cs-shadow | but I don't know if that has happened already or not | 14:45 |
| juergbi | pip is also duplicated | 14:45 |
| coldtom | cs-shadow: i can say that none of the projects i've done with bst master have depended on it, not sure about others though | 14:46 |
| juergbi | ok, and otherwise it should be trivial to change those projects back to the core plugin | 14:46 |
| juergbi | same holds true for the pip plugins, I suppose | 14:47 |
| cs-shadow | yeah, I vote for removing tar then, since it's more hairy | 14:47 |
| juergbi | that was merged very recently, though. maybe pip removal on buildstream is almost ready | 14:47 |
| cs-shadow | pip I think is close to being ready in https://gitlab.com/BuildStream/buildstream/-/merge_requests/1810 so maybe we juts need to do that | 14:47 |
| juergbi | ok, will just remove tar | 14:48 |
| *** narispo has quit IRC | 15:06 | |
| tpollard | pip just needs the pytest support for the cache key test, unless anyone thinks otherwise | 15:09 |
| juergbi | cs-shadow: oh, the deb plugin depends on the tar plugin | 15:15 |
| cs-shadow | ah! I hadn't realized that until now | 15:16 |
| *** rdale has quit IRC | 15:16 | |
| *** rdale has joined #buildstream | 15:16 | |
| cs-shadow | that's probably also wrong on its own though | 15:17 |
| cs-shadow | so, we manually update tar for now? | 15:17 |
| juergbi | I guess that's the simplest approach | 15:18 |
| juergbi | hopefully not necessary for too long | 15:18 |
| *** bochecha_ has joined #buildstream | 15:23 | |
| *** bochecha has quit IRC | 15:23 | |
| *** bochecha_ is now known as bochecha | 15:23 | |
| *** bochecha_ has joined #buildstream | 15:24 | |
| *** bochecha has quit IRC | 15:26 | |
| *** bochecha_ is now known as bochecha | 15:26 | |
| *** traveltissues has joined #buildstream | 15:31 | |
| *** bochecha has quit IRC | 15:32 | |
| *** jonathanmaw has quit IRC | 15:33 | |
| *** jonathanmaw has joined #buildstream | 15:34 | |
| *** phildawson-ct has quit IRC | 15:38 | |
| *** phildawson-ct has joined #buildstream | 15:41 | |
| *** phildawson-ct has quit IRC | 15:47 | |
| juergbi | cs-shadow: are you aware of the issue with the tests-plugins-master test in buildstream? | 15:58 |
| juergbi | INTERNALERROR> pkg_resources.VersionConflict: (BuildStream 1.93.0+44.gb57dbff9 (/builds/BuildStream/buildstream/src), Requirement.parse('buildstream>=1.93.0.dev0<2.0.0')) | 15:58 |
| juergbi | (or anyone else) | 15:58 |
| cs-shadow | juergbi: I wasn't aware of this. Seems like setuptools doesn't think that 1.93.0.gitsha is >= 1.93.0.dev0 | 16:00 |
| cs-shadow | if one of them is considered strictly less than the other we can probably fix this nicely, otherwise maybe just do a > 1.92.10 or something. I need to run to a meeting but I'll have a look at fixing this later this evening unless you fix it already | 16:02 |
| juergbi | ok, ta. I might take a closer look, will let you know if I do | 16:04 |
| *** traveltissues has quit IRC | 16:05 | |
| *** xjuan has joined #buildstream | 16:19 | |
| *** phildawson-ct has joined #buildstream | 16:20 | |
| juergbi | cs-shadow: hm, odd, this returns True: version.parse("1.93.0+44.gb57dbff9") >= version.parse("1.93.0.dev0") | 16:35 |
| juergbi | wondering whether it's somehow using a version check function that is not compliant with PEP440 | 16:36 |
| juergbi | no, I don't think so | 16:40 |
| juergbi | the issue could be that it considers the git URL for buildstream specified in plugins-experimental tox.ini to be a better match than the buildstream branch | 16:43 |
| *** lachlan has joined #buildstream | 17:06 | |
| gitlab-br-bot | cs-shadow opened issue #1260 (Tests are failing against bst-plugins-experimental master) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1260 | 17:13 |
| cs-shadow | juergbi: that's odd - created ^ for record. Seems like `pip` is happy to install it, but setuptools/pkg_resources fails to find it | 17:15 |
| *** lachlan has quit IRC | 17:31 | |
| benschubert | juergbi: 1) I will send an email to the ML tomorrow for a proposal on handling more involved tests when we extract plugin so we can correctly excise tar from BuildStream | 17:36 |
| benschubert | 2) We have a problem with bst-plugins-experimental specifying buildstream as a dependency. I want to remove it, but I think some people were against it? And that is the reason for your second problem I think | 17:36 |
| juergbi | ok, thanks for the info | 17:38 |
| juergbi | benschubert: if 2) is also an issue if we tag a new plugins-experimental release, then this is a blocker for !1761 | 17:38 |
| gitlab-br-bot | MR !1761: mtime support https://gitlab.com/BuildStream/buildstream/-/merge_requests/1761 | 17:38 |
| benschubert | juergbi: Is it? tests do pass | 17:39 |
| benschubert | or am I missing something? | 17:39 |
| juergbi | benschubert: the broken/outdated tar plugin from bst-plugins-experimental 0.15.0 is the cause of the userchroot test failure in the mtime MR | 17:40 |
| juergbi | the tar plugin update is merged now to bst-plugins-experimental master | 17:41 |
| juergbi | but if upgrading buildstream CI to latest bst-plugins-experimental is blocked by (2) above, it's hence also a blocker for !1761 | 17:41 |
| juergbi | I can get the test to pass by dropping the bst version check in bst-plugins-experimental install-requirements.txt | 17:43 |
| benschubert | Let's do that | 17:43 |
| *** jonathanmaw has quit IRC | 18:01 | |
| juergbi | benschubert: https://gitlab.com/BuildStream/bst-plugins-experimental/-/merge_requests/82 | 18:03 |
| juergbi | benschubert: ah, I just saw your https://gitlab.com/BuildStream/bst-plugins-experimental/-/merge_requests/76 | 18:04 |
| juergbi | wondering whether that might fix the issue as well | 18:04 |
| *** santi has quit IRC | 18:04 | |
| benschubert | juergbi: it would not I think | 18:04 |
| benschubert | but we can definitely merge it :) | 18:04 |
| benschubert | ah no, I need a rebase | 18:05 |
| benschubert | let me do that | 18:05 |
| juergbi | ok, let's merge that after rebase and then let's do a quick test in buildstream CI just in case | 18:05 |
| benschubert | done | 18:07 |
| benschubert | good to merge? | 18:07 |
| juergbi | benschubert: yes, approved | 18:12 |
| benschubert | set to merge | 18:13 |
| juergbi | I actually misread the diff before. you're right, that won't fix the other issue | 18:13 |
| benschubert | I'm not even sure why we want the plugin to depend on buildstream :/ | 18:13 |
| juergbi | benschubert: my guess was that it's needed to test bst-plugins-experimental via tox, i.e., to have it automatically install a suitable version of buildstream in the venv | 18:15 |
| juergbi | but that might be completely wrong | 18:15 |
| benschubert | I think we pin the version in tox itself, I'm not sure anymore | 18:17 |
| benschubert | if removing version pinning doesn't break, that means it's not handled by this part | 18:17 |
| cs-shadow | no, I think this is more of a philosophical question - should plugins depend on the thing they are plugins for? | 18:17 |
| cs-shadow | one side of the argument goes that since they `import` something, they should. Whereas the other side goes that it's implicit that the plugins need what they are plugins for and should only list additional stuff | 18:18 |
| juergbi | right. strictly to indicate a requirement/dependency, the latter does seem logical | 18:19 |
| juergbi | however, with pip auto-installing dependencies, it gets a bit odd | 18:19 |
| benschubert | and we don't want to have to do pip install --no-deps because it gets weird :) | 18:20 |
| juergbi | yep, so unless we can disable auto-install somehow for the dependency on 'buildstream', I'd rather not have the dependency listed at all | 18:20 |
| juergbi | (assuming this doesn't break tox in plugins-experimental) | 18:21 |
| benschubert | we have our ways of adding that to tox, it's definitely not a problem | 18:21 |
| juergbi | that said, I don't have a very strong opinion on it. I mainly want to avoid breakage in buildstream core | 18:21 |
| benschubert | +2 on that | 18:22 |
| juergbi | benschubert: I think the version number in install-requirements.txt also had the effect that test-master-fedora-31 wasn't actually testing buildstream master: https://gitlab.com/BuildStream/bst-plugins-experimental/-/jobs/425797477 | 18:31 |
| juergbi | otherwise that error message would also appear in other branches | 18:31 |
| benschubert | -_- right | 18:31 |
| *** xjuan has quit IRC | 20:40 | |
| *** xjuan has joined #buildstream | 22:11 | |
| *** ikerperez has quit IRC | 22:12 | |
| *** xjuan has quit IRC | 22:14 | |
| *** xjuan has joined #buildstream | 22:14 | |
| *** xjuan has quit IRC | 22:33 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!