[00:01] Yeah, I don't think the list on the repo is that useful. [00:01] (the repo -> bug link already exists if you go through the commit history, anyway) [00:01] And for refs, it can be linkified from recent commit messages. [00:01] Usually. [00:01] Once we have a convention I imagine we could even hack our cgit to support it. [00:02] Right, exactly. [00:02] Were we just going to apply the LP: regex for changelogs? [00:02] So I guess we grow a BugGitRepository which is unique (commit_sha1, bug) [00:03] I hadn't really thought about that yet. [00:03] My main goal was to get it possible to have Git MPs with bugs, which is the main blocker for us. [00:03] Whether manually linked or otherwise. [00:03] I wonder what tarmac would do with this. [00:03] The MP bug harvesting is then actually separate from BugGitRepository. [00:03] What do you mean? [00:04] It currently expects to be able to do ... oh, it's actually fetching the bug list from bzrlib directly [00:04] Because the MP should grab any bugs referenced from commits that it involves, rather than looking at BugGitRepository at all. [00:04] Oh, is it? [00:04] rev = self.bzr_branch.repository.get_revision(rev_info[0]) [00:04] for bug in rev.iter_bugs(): [00:04] Huh, I didn't expect that. [00:04] An exotic approach. [00:05] I guess that makes sense given the LP list isn't always reliable. [00:05] Maybe they didn't trust LP, indeed. [00:05] But I do occasionally forget to --fixes and link manually on LP, which would break in that world. [00:06] Anyway, if a sensible thing were exposed on BMP's API, that would be better. [00:06] Perhaps that is fine, though. [00:06] Which is easier than coming up with a sensible thing for a branch. [00:06] Since there's an implied stop point. [00:06] I'm weighing up whether to continue to allow manual linking at all. [00:06] For MPs, at least. [00:07] But that would mean we'd need automatic linking before this would be useful at all. [00:07] Which means bikeshedding the commit message format. [00:07] Yeah, I assumed we'd bikeshedded this eight years ago and would just use that :) [00:08] Or at least have it as a valid option, so that package repositories can be sensible. [00:08] Anyway, I must sleep, happy to debate further in the morning. [00:08] If by sensible you mean wrong, then indeed :) [00:09] Well, I guess if we don't autoclose then it's OK. [00:09] If it preserves current LP behaviour then it'd be fine. [00:09] But GitHub instacloses on merge. [00:09] Night. [00:10] Yeah, I meant reference not autoclose [00:31] cjwatson: Oh, IPersonRoles really doesn't enjoy being given None. Will fix. [11:18] wgrant: Bah. Thanks. [11:19] I didn't run the full bugs test suite after that. :-( [11:30] Quis testfix ipsos testfix, etc. [16:30] rpadovani: Will you be able to do QA on your change? That is, trying things out on qastaging and making sure they behave sensibly [19:12] cjwatson, sure thing, I start right now :-) [19:26] cjwatson, so, I tested it here: https://bugs.qastaging.launchpad.net/launchpad/+bug/1391394 with my account and a private browser session, and works as expected. What's next? Should I create screenshots? [19:26] Bug #1391394: Unable to unhide my comment

[19:33] remove qa-needstesting and putting qa-ok, i'd say (/me's just trying a guess) [19:49] rpadovani: that should be fine, so you can do as mapreri says. https://dev.launchpad.net/PolicyAndProcess/QAProcess?highlight=%28qa-ok%29 [19:49] rpadovani: no need for screenshots [19:51] ... I need to read more the wiki [19:51] ty [19:52] "qa-rcfixed: when a bug is fixed in RC mode." cjwatson ? [19:52] the only meaning i'm aware of RC is "Release Critical", guess this is different? ;) [19:53] (or release candidate) [19:58] that may be obsolete, it predates me [20:49] nice