[00:19] blr: ui-project-setbranch has conflicts; can you merge devel and resolve them? [00:20] And would it be worth cherrypicking the semi-unrelated menu changes into a separate branch, to get the "Configure translations" -> "Configure Translations" bits out of the big diff? [00:20] ugh, yep sure [00:21] wgrant: can do sure, if it is obscuring the primary intention of the branch. [00:22] blr: I think it should be pretty easy to put in a separate branch, and it would shrink the 2200-line diff by quite a bit... [00:22] Just a bzr merge -i from the old branch into the new plus a couple of fixups where they overlap should work. [00:22] Then we can review and land the new branch before the old, and the diff will be a whooooole lot more sensible. [00:24] wgrant: sounds reasonable [00:30] blr: New JS is much better, thanks. [00:31] yep I think that should work well enough. [00:32] will let you know when I have the 2 branches [00:33] I think there should only be like one file that needs the two sets of changes disentangled. [02:32] wgrant: wow, managed to blow up more tests than anticipated. Also update the BMP factories. Ready for you now if you have a moment: https://code.launchpad.net/~blr/launchpad/recycle-commit-message/+merge/261793 [02:34] blr: Explosive tests are fun. [03:45] wgrant: is there a reason there's no argument for 'description' on resubmit? (in the interface) [03:46] blr: Someone probably forgot to add it there when the field was added. [03:46] Probably an idea to fix that too. [03:46] wgrant: yep, I did - just checking I wasn't missing something :) [03:53] as much as I like the keyboard on the lenovo, it is such an improvement using a proper mechanical keyboard... [03:53] As laptop keyboards go, the T4[45]0s have by far the best I've used. [03:53] But yes, a proper desktop mechanical keyboard is hard to beat. [03:54] had a look at the new macbook the other day - impressive hardware, but there's even less travel on the keys now than previous models... it would be intolerable I think heh [03:55] The one-port-to-rule-them-all model? [03:55] Yeah, certainly shiny and expensive. [03:55] hah yeah [03:55] But the keyboard is way worse than an MBA [03:55] And that's saying something... [04:32] wgrant: hmm I didn't receive a pqm notification for that last lp-land, and I don't see anything in the queue.. [04:39] blr: buildbot had spuriously failed. [04:42] wgrant: ah that old chestnut. [07:32] lp:~person-name-100028/ubuntu/+source/unique-from-factory-py-line3422-100026/+git/gitrepository-100029 [07:32] The factory creates the best URLs. [07:33] wgrant: Not enough -deactivatedaccount-deactivatedaccount in the person portion [07:37] StevenK: I'll have you know that no account has more than 17 of them. [07:38] https://launchpad.net/~deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount-deactivatedaccount for reference [07:38] Which isn't actually deactivated. [07:39] Also I guess that technically has 18, since I missed the first one, as it doesn't start with - [07:39] Hahaha [12:09] cjwatson: Oh, good catch. [12:09] Thanks. [12:10] Some day I'll find a problem with one of your MPs that isn't entirely trivial :-) [12:11] I'm normally sufficiently pedantic about things like that :/ [12:11] Almost finished syncing everything over, should be back in the saddle properly this afternoon [12:11] Yay [12:12] Trying to sort out my git-detect-merges branch, which is proving a bit awkward because passing a mock hosting client through the event infrastructure would be nightmarish, so I'm having to bite the bullet and sort out a fixture [12:12] But despite being on PyPI now using turnip proper as the fixture is rather difficult - I don't want LP deployments to have to have a usable version of pygit2, and we don't have test_requires in LP at the moment [12:12] Considered ZopeUtilityFixture? [12:13] So I'm doing a quick Twisted emulation of the relevant bits of the API in the short term [12:14] Hm, yeah, that might be easier if I turned GitHostingClient into a utility, it's true [12:14] Easier than writing something that emulates the other end of the API and then having a fixture to override the endpoint URL, which is what I'd been doing [12:15] Yep. [12:15] There are also various libraries around that mock out requests responses. [12:15] I've gone through three so far for webhooks stuff [12:15] And GHC really ought to be a utility. [12:15] And they all mostly work. [12:15] But they're a bit weird. [12:16] I have the guts of the Twisted version, but it means converting things over to AsynchronousDeferredRunTests and then drinking heavily. [12:16] So maybe that's a blind alley. [12:16] You got the order wrong, but sure. [12:17] I think order is immaterial, they can be done concurrently [12:17] Indeed. === anthonyf is now known as Guest45939 === infinity1 is now known as infinity [15:18] cjwatson: whats the drinking heavily for? Is there framework stuff we can do to make it less drenched? [15:19] cjwatson: also, the glasgow haskell compiler should be a utility? [15:21] GitHostingClient :-) [15:23] y'all still advertise AsynchronousDeferredRunTest as "use at your own peril"; I'm happy to modify tests that are already using it but for something new where all I need is for LP to use a mock version of a utility, I think wgrant is right that ZopeUtilityFixture is simpler [16:21] ... and indeed it is. /me pushes [20:15] cjwatson: oh, we should remove that terror warning [20:16] I'll consult with jml