[03:31] wgrant: are webhooks at a point now where I could look at building out a view perhaps? [04:12] blr: Not at the moment -- hopefully later in the week. Is setbranch all done from your perspective/ [04:14] wgrant: yep, the cherry picked link name changes are here: https://code.launchpad.net/~blr/launchpad/project-configuration-rename/+merge/261918 [04:14] once that's landed, imagine it will be much easier to review the other branch [04:15] blr: Oh, can you flip that out of WIP? [04:15] wgrant: gah, sure thing [04:16] One day things like this will be sensible. [04:16] eg. bugs will say whether they're open or closed [04:16] And MPs will do the same, except with an additional state that is neither. [04:16] Rather than expecting everyone to know and notice the particular statuses... [04:16] that sounds sensible [04:19] can you think of anything that needs addressing atm? (otherwise I'll find some more bugs to poke at) === anthonyf is now known as Guest37830 [04:27] blr: One thing we need is a ref picker. [04:27] blr: For +register-merge, mostly. [04:27] Currently you have to type the target ref manually. [04:27] It's the most annoying bit about Git MPs atm. [04:28] https://app.asana.com/0/29290342588691/37033052254451 [04:34] wgrant: great, will have a look after I tidy up this MP. [04:39] asana? [04:44] lifeless: Lightweight task tracking service. [04:51] wgrant: using it for LP development? [04:54] lifeless: Where it makes sense, yes. [05:10] seems a little ironic ;) [05:12] lifeless: No moreso than using kanban like in the old days. [05:12] LP bugs is good for tracking some types of things. [06:25] blr: Are you around to fix the broken test? If not, I'll do it. [06:42] wgrant: back sorry [06:42] blr: No worries. [06:50] wgrant: https://code.launchpad.net/~blr/launchpad/project-configuration-rename/+merge/261930 [06:51] blr: Thanks. [07:37] wgrant: buildbot happy, will get the other branch merged. === anthonyf is now known as Guest35437 [20:45] morning [20:47] Morning blr [20:47] cjwatson: Do you believve the GitHostingClient regressions to all be fixed now? [20:53] wgrant: did something blow up, or are you just up early? :/ [20:53] blr: Just an early meeting. [20:54] ah good [21:06] wgrant: Yes - they weren't regressions (well, except from unmerged code) since they only affected the patch case, and I've confirmed that to be working now [21:07] Oh, yeah. [21:07] No empty responses yet. [22:02] * wgrant gets a lot of coffee and dives back into blr's setbranch [22:04] wgrant: ah thanks [22:04] blr: Thanks for cutting the irrelevant bits out of the diff. [22:04] It's still 1700 lines, but that's a lot better than 2300 :) [22:04] hopefully easier to deal with now [22:04] yes :) [22:05] And the changes are mostly to a few files rather than EVERYWHERE. [22:05] in retrospect I should have made the renaming changes first... will try to think about decomposing branches before going down the rabbithole next time [22:07] wgrant: when looking at it again yesterday, I did wonder if it needs an additional test for the setDefaultRepository case [22:08] blr: It's often difficult to foresee quite how deep the rabbithole is. [23:33] wgrant: just lodging a bug suggesting that the test factories should use the first available Person.name rather than throwing an exception - is that reasonable? [23:37] introducing more randomness to the property is probably the only way to manage it without a db query [23:39] wgrant: if you have any comments https://bugs.launchpad.net/launchpad/+bug/1465438 [23:39] Bug #1465438: Factories should ideally use first available Person.name [23:46] blr: It shouldn't cause a serious performance regression; NameAlreadyTaken is raised when getByName returns something, so it's already performing an extra query. [23:47] blr: The login code uses generate_nick to create a unique name. === anthonyf is now known as Guest93759 [23:48] wgrant: a minor annoyance. I'll make a note about generate_nick thanks [23:48] Yeah, it's quite irritating, but usually only a problem in the harness rather than in tests. [23:48] Could also include a timestamp in the generated names (it doesn't just affect Person), or possibly only do that outside tests. [23:49] But tests shouldn't be depending on the specifics of the name generation, so always including something unique would probably be fine. [23:49] wgrant: yep, that seemed like the least obtrusive way to improve it