| *** juergbi has quit IRC | 02:55 | |
| *** juergbi has joined #buildstream | 03:15 | |
| *** delli3_ has joined #buildstream | 03:24 | |
| *** dellie__ has quit IRC | 03:26 | |
| *** narispo has joined #buildstream | 04:17 | |
| *** dellie__ has joined #buildstream | 06:24 | |
| *** delli3_ has quit IRC | 06:26 | |
| *** phildawson has joined #buildstream | 08:19 | |
| *** traveltissues has joined #buildstream | 08:33 | |
| *** tme5 has joined #buildstream | 09:41 | |
| *** dellie__ has quit IRC | 09:48 | |
| *** slaf has quit IRC | 09:56 | |
| *** jonathanmaw has joined #buildstream | 10:27 | |
| *** slaf has joined #buildstream | 10:36 | |
| *** slaf has joined #buildstream | 10:36 | |
| *** slaf has joined #buildstream | 10:36 | |
| *** slaf has joined #buildstream | 10:36 | |
| *** slaf has joined #buildstream | 10:37 | |
| *** phildawson-ct has joined #buildstream | 10:58 | |
| *** phildawson has quit IRC | 10:58 | |
| *** santi has joined #buildstream | 11:37 | |
| *** phildawson-ct has quit IRC | 11:50 | |
| *** pointswaves has joined #buildstream | 12:15 | |
| coldtom | would it be possible to get some review and hopefully a merge on !1806 please? | 13:13 |
|---|---|---|
| gitlab-br-bot | MR !1806: Stop plugin tests using private API https://gitlab.com/BuildStream/buildstream/merge_requests/1806 | 13:13 |
| *** pointswaves has quit IRC | 13:18 | |
| gitlab-br-bot | tpollard approved MR !1806 (coldtom/testing-api->master: Stop plugin tests using private API) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1806 | 13:21 |
| tme5 | is there any reason why a test using a git repo wouldn't be doing a full fetch? | 14:03 |
| tme5 | tempfiles sticking around? would have thought they'd be cleared | 14:03 |
| tme5 | or a reason why some status messages might not be printed out during a test | 14:23 |
| coldtom | tme5: i suspect the source cache may be your issue | 14:23 |
| tme5 | i've tried rm -rf .tox/py37/tmp | 14:24 |
| tme5 | is there somewhere else i should delete? | 14:26 |
| tme5 | but tbf the "running host command" of the thing which checks if it even needs to fetch is not showing up | 14:26 |
| coldtom | integration_cache is the offender i believe | 14:27 |
| * coldtom just had a similar problem with https://gitlab.com/BuildStream/bst-plugins-experimental/merge_requests/72 | 14:28 | |
| tme5 | integration_cache where? | 14:28 |
| benschubert | Normally, all caches and such are set to new temporary directories so they should always be empty at the start of a specific test | 14:29 |
| coldtom | should be in the project root i think | 14:29 |
| tme5 | n'existe pas | 14:29 |
| tme5 | i think it might be that the logs are incomplete somehow | 14:30 |
| coldtom | i only experienced this issue when using buildstream.testing.runcli.cli_integration rather than buildstream.testing.cli | 14:31 |
| tme5 | ah, looking in the full log file shows me everything | 14:31 |
| *** phildawson has joined #buildstream | 14:32 | |
| coldtom | unrelated, but https://gitlab.com/BuildStream/bst-plugins-experimental/merge_requests/72 could do with a review if anyone has time | 14:32 |
| tme5 | benschubert, another issue with my branch currently is that a user will get the no-ref-in-track warning when opening a workspace on a shallow-fetched repo | 14:39 |
| benschubert | that is expected if you get it when you fetch it the first time too | 14:40 |
| benschubert | or do you mean you can fetch correctly but get it at workspace open time? | 14:40 |
| tme5 | the check happens in validate_cache | 14:41 |
| tme5 | i'm seeing the warning when i open the workspace | 14:41 |
| benschubert | Do you see it too when you track/fetch ? | 14:42 |
| tme5 | not when fetching | 14:46 |
| benschubert | when tracking? | 14:47 |
| tme5 | nope | 14:50 |
| benschubert | If it is not when tracking nor fetching, then there is something wrong with your local cache. Would something get modified between? | 14:51 |
| *** xjuan has joined #buildstream | 15:39 | |
| tme5 | ah looks like I misread. it's actually after fetching that I get the error | 15:41 |
| tme5 | what is the purpose of the warning? | 15:46 |
| *** santi has quit IRC | 15:47 | |
| *** santi has joined #buildstream | 15:47 | |
| tme5 | i suppose to make sure the element is not in some unresolved configuration. but if a project makes this warning an error they will be unable to use the 'git' source with tag refs | 15:47 |
| benschubert | this should probably be checked differently for tag refs then | 15:50 |
| benschubert | I really wonder whether we should not instead slim down the gitsourcebase, separate cleanly git/git_tag/git_lfs/git_submodule and try to have smaller plugins that do one thing well (sharing common parts and living under a single repo is a big plus :) ) | 15:54 |
| coldtom | benschubert: i think that's a nice idea with some implementation details to be figured out | 15:55 |
| coldtom | for example, we likely have a combinatorial explosion of plugins, such as git_lfs_tag_submodule, if i want to track tags on a git submodule that uses lfs, say | 15:55 |
| tme5 | personally i don't think that solves any of our issues | 15:55 |
| benschubert | coldtom: not necessarily. a git submodule is still a git repo. So you could have a first 'git_tag' source, then a 'git_submodule' source transform and finally a 'git_lfs' source transform in the sub directory | 15:56 |
| benschubert | tme5: could you elaborate? :) | 15:57 |
| coldtom | ah, i see if we use source transforms for lfs then it makes sense | 15:57 |
| coldtom | at present we have git-lfs support as part of git_tag, but only in bst-external i think | 15:57 |
| tme5 | what is a source transform? | 15:57 |
| benschubert | Yeah, lfs cannot live anywhere else than in a git repo. So it seems to be a good fit for a source transform | 15:57 |
| benschubert | tme5: it's a source that cannot be the 'top level' source, and gets access to its parent sources | 15:57 |
| benschubert | (when you track, you already have your parents in the staging area) | 15:58 |
| benschubert | For example, we have 'patch' and 'quilt' that are source transforms | 15:58 |
| benschubert | or pip_source | 15:58 |
| benschubert | and it is allowed to operate on the parent source | 15:58 |
| benschubert | coldtom: I would be worried to move 'lfs' support in the 'git' plugin. It would end up doing many different things, and require 2h of documentation read to know how to implement correctly, and would be harder to know what's happening at a glance | 16:01 |
| tme5 | I see. But I think that's beside the point right now. This fetch optimisation would have this issue regardless of which specific git plugin I might add it too | 16:04 |
| benschubert | oh, about only fetching with --depth=1 ? | 16:05 |
| tme5 | yes | 16:05 |
| benschubert | gah, I misunderstood you, I thought that was when using the 'git_tag' functionallity moved to the 'git' plugin | 16:06 |
| tme5 | oh sorry, I'm talking about my new branch/MR | 16:06 |
| benschubert | I jsut did: | 16:09 |
| benschubert | git clone $my_repo --depth 1 | 16:09 |
| benschubert | cd my_repo | 16:09 |
| benschubert | git branch --list master --contains $(git rev-parse EHAD) | 16:09 |
| benschubert | This does give me the ref. Is there something else I'm missing? | 16:09 |
| tme5 | yes, you cloned the master branch ref | 16:10 |
| tme5 | if you fetch a tag you get no branch git refs | 16:10 |
| benschubert | _, tag = self.check_output( | 16:10 |
| benschubert | [self.host_git, "tag", "--list", self.tracking, "--contains", self.mirror.ref], | 16:11 |
| benschubert | cwd=self.mirror.mirror, | 16:11 |
| benschubert | ) | 16:11 |
| benschubert | Should be taking care of that part no? | 16:11 |
| tme5 | hmm, yeah it should | 16:13 |
| tme5 | ah no it shouldn't | 16:14 |
| tme5 | because it uses self.tracking | 16:14 |
| tpollard | I was literally talking about that snippet elsewhere | 16:15 |
| tpollard | how weird | 16:15 |
| benschubert | tme5: not sure I follow around the 'self.tracking' | 16:28 |
| tme5 | the command will run .e.g git tag --list master | 16:28 |
| tme5 | but master will not exist in the local repo | 16:29 |
| benschubert | gah I see | 16:44 |
| tme5 | i would suggest i depth=1 fetch all refs but i don't think that would solve anything, since the refs would just be floating and the --contains would probably not work | 16:47 |
| tme5 | will try though | 16:48 |
| *** traveltissues has quit IRC | 16:52 | |
| benschubert | and that will potentially pull a lot of things x') | 16:58 |
| benschubert | could we also fetch the branch in addition to the tag or something? | 16:58 |
| tme5 | the whole branch? | 17:16 |
| tme5 | that would work, but again that's a lot of things | 17:17 |
| benschubert | or everything between both commits? I don't know if that is feasible or not | 17:17 |
| benschubert | yeaha ctually the whole branch would remove the interest of the patch | 17:18 |
| *** santi has quit IRC | 17:31 | |
| *** tme5 has quit IRC | 17:36 | |
| *** xjuan has quit IRC | 17:41 | |
| *** pointswaves has joined #buildstream | 17:41 | |
| *** xjuan has joined #buildstream | 17:42 | |
| *** xjuan has quit IRC | 17:46 | |
| *** santi has joined #buildstream | 17:48 | |
| *** pointswaves has quit IRC | 17:53 | |
| *** pointswaves has joined #buildstream | 17:57 | |
| *** pointswaves_ has joined #buildstream | 18:00 | |
| *** pointswaves has quit IRC | 18:00 | |
| *** pointswaves_ has quit IRC | 18:02 | |
| *** pointswaves_ has joined #buildstream | 18:03 | |
| *** xjuan has joined #buildstream | 18:03 | |
| *** xjuan has quit IRC | 18:06 | |
| *** pointswaves_ has quit IRC | 18:12 | |
| *** pointswaves has joined #buildstream | 18:16 | |
| *** pointswaves_ has joined #buildstream | 18:21 | |
| *** xjuan has joined #buildstream | 18:22 | |
| *** pointswaves_ has quit IRC | 18:22 | |
| *** pointswaves has quit IRC | 18:22 | |
| *** xjuan has quit IRC | 18:34 | |
| *** jonathanmaw has quit IRC | 18:49 | |
| *** santi has quit IRC | 19:40 | |
| *** santi has joined #buildstream | 20:01 | |
| *** pointswaves_ has joined #buildstream | 20:35 | |
| *** cs-shadow has quit IRC | 21:51 | |
| *** santi has quit IRC | 21:55 | |
| *** pointswaves_ has quit IRC | 22:36 | |
| *** cgmcintyre[m] has quit IRC | 23:00 | |
| *** verdre[m] has quit IRC | 23:00 | |
| *** reuben640[m] has quit IRC | 23:01 | |
| *** theawless[m] has quit IRC | 23:01 | |
| *** cgmcintyre[m] has joined #buildstream | 23:12 | |
| *** verdre[m] has joined #buildstream | 23:14 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!