IRC logs for #buildstream for Wednesday, 2021-01-20

*** tristan has joined #buildstream05:49
*** ChanServ sets mode: +o tristan05:49
*** tristan has quit IRC07:30
*** tristan has joined #buildstream07:30
*** ChanServ sets mode: +o tristan07:30
*** tristan has quit IRC07:48
*** abderrahim[m] has joined #buildstream09:02
*** tristan has joined #buildstream09:09
*** ChanServ sets mode: +o tristan09:09
*** santi has joined #buildstream09:44
*** toscalix has joined #buildstream10:36
abderrahim[m]Hi all10:53
abderrahim[m]so what's the situation with the github migration? I see issues are still off by one10:53
*** toscalix has quit IRC11:11
tristanabderrahim[m], we are still waiting on infra people for that, Justin is giving them a nudge last I heard11:26
tristanabderrahim[m], for the time being I'm keeping a mirror of the repo on github and we're still landing PRs on github, hoping that the reset will happen soon11:27
tristanSo instead of surgically replaying master on top of the gitlab version, we can push what was on github at the time, and just do the data migration - losing the short history of pull requests we've landed in the ugly window11:27
abderrahim[m]so I can just send my pull requests now in the "ugly window"?11:32
abderrahim[m]or is it better to wait11:32
abderrahim[m]another thing I wanted to ask, is anyone working (or has worked) on porting git_tag to the base class in bst master?11:33
tristanI can't easily find the thread right now, but I seem to recall this hitting the mailing list and being somewhat controversial11:35
tristanI think there are some wonky things in git_tag, although the nicer track results by itself might be nice to have (did it start doing some weird thing like supporting dual "track" configurations or smth like that ?)11:36
coldtomkinda, you can have an additional branch and it takes the latest, for if a project has a master branch + release branch, i think this would be a good opportunity to replace that with a pattern based track parameter anyway11:44
coldtomit's also worth noting that git_tag has a bunch of fixes which are not reflected in the original git plugin (and presumably also not in the base plugin in master)11:44
coldtomto answer your question abderrahim[m], i believe the state is still as in https://gitlab.com/BuildStream/bst-plugins-experimental/-/issues/3, i.e. nobody has tried because the git abstract plugin is still private11:55
tristanRight, I'm not really seeing the value of this dual branch thing; the stable branches of upstream should be reflected in the "track" parameters of BuildStream elements on a corresponding stable branch for that BuildStream project - stating that there can be multiple branches that are valid for a single mainline doesnt strike me as a pattern we should propagate11:55
coldtomthe reason for it is that fdsdk use git_tag to search for latest releases, and in projects which branch for releases, tags for new releases may be on multiple branches. the idea was that `master` would be removed from the tracking branches when fdsdk stabilised to a release branch12:07
coldtomhowever i agree that this probably isn't suitable for buildstream "blessed" plugins, as it's changing what "track" is ostensibly for12:07
*** santi has quit IRC12:39
*** santi has joined #buildstream12:40
*** tristan has quit IRC14:40
*** tristan has joined #buildstream14:56
*** ChanServ sets mode: +o tristan14:56
*** santi has quit IRC18:41
*** toscalix has joined #buildstream20:27
*** toscalix has quit IRC20:53
*** toscalix has joined #buildstream20:53
*** toscalix has quit IRC20:54
*** toscalix has joined #buildstream20:59
*** toscalix has quit IRC21:14
*** toscalix has joined #buildstream21:29
*** toscalix has quit IRC21:30
nanonymetristan, FWIW Fedora seems to have orphaned grpc :(21:41
*** toscalix has joined #buildstream23:10
*** toscalix has quit IRC23:12

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!