[00:02] * cjwatson embarks on a bit of refactoring there [00:06] wgrant: I think it was actually broken by r17579 and we just didn't notice. inferred_vcs shows up in the output. [00:07] cjwatson: Quite possibly. Must be test ordering. [00:07] Since that test has passed since then. [00:07] Yeah, as I say that test doesn't seem as careful as it should be. [00:08] I'm up for a while yet, so I'll clean it up [00:08] I'd forgotten about the JSON cache when I approved that branch. I'm tempted to revert the inferred_vcs export. [00:09] blr: What do you think? [00:09] inferred_vcs should be temporarily, anyway, as we can backfill vcs at some point. [00:10] And it is causing every Product view to incur Branch and GitRepository queries to render the JSON representation. [00:17] I wouldn't be sad about that [00:27] wgrant: how valuable is it to expose in the api? [00:28] blr: I don't think it's very interesting at the moment. [00:28] agreed [00:28] we really only need it internally for now [00:30] wgrant: are you happy to revert the export or shall I? [00:34] blr: Could you? Most/all of the query count test fixes can probably go. [00:35] wgrant: sure [00:52] Ha, after starting on this refactoring I notice that I have a branch from March with exactly the same thing in progress [01:25] cjwatson: I hate it when that happens. [01:26] Like with the product-aps thing [01:26] I was looking on DF and wondering why the query didn't use the Product.access_policies that clearly already existed. [01:27] So I tweak the query locally and the tests fail because the column doesn't exist. [01:27] I'd worked through most of the solution on DF in February and completely forgotten all about it. [04:33] blr: All the other test changes are still necessary with that reverteD? [04:37] wgrant: hmm perhaps I missed some tests, green with regex 'query_count' and 'test_product', plus the doctest [04:39] blr: I'd just look through your two testfix merges to see if you've missed anything. [04:51] wgrant: yes, the other querycount bumps are necessary (specification, bugtask, gitlisting) [04:52] blr: Weird, OK.\ [04:52] r=me [04:52] Thanksw [05:18] Oh [05:18] I even filed a bug about the thing I forgot about, with details: https://bugs.launchpad.net/launchpad/+bug/1425430 [05:19] Bug #1425430: Bulk project access checks are slow

[08:25] blr: I think your latest change will result in duplicated result lines if there is a comment in the dirty header and a comment somewhere else in the same file [08:28] blr: You probably want http://paste.ubuntu.com/11797735/ ? [08:32] wgrant: Any word on the Soyuz redesign docs you had sketches of? [08:35] cjwatson: No. [08:35] Hmm. [08:35] buildbot is very unhappy with me. [08:35] I have a vague recollection of Storm leaking references with List attributes. [08:35] Many years ago. [08:35] It really is ... [08:36] PackageUpload.searchable_versions is a List, though. [08:36] And it doesn't seem to blow up quite so effectively. [08:50] cjwatson: err quite, thanks. [08:55] I can reproduce it locally, at least, on one test. [09:02] wgrant: shall I land this in the morning, given you and buildbot are not on good terms atm? [09:25] blr: Yeah, landing it now won't exactly work. [09:29] Ahhh, computers, how I love you. [09:29] Can reliably reproduce in product-aps-set [09:30] Cannot reproduce in current devel. [09:30] But buildbot can. [09:32] I'm going to see if I can implement my apt-get build-dep / dose-builddebcheck suggestion today, unless anyone has serious objections. [09:32] Starting with trying to backport the thing to precise ... [09:33] The whole pulling ocaml into everywhere thing is sort of unpleasant. [09:33] But worth a try, I suppose. [09:33] Only into our build-dep chain. [09:33] ocaml binaries are standalone once you get them built. [09:34] Yeah [09:34] But is ocaml in the interesting bootstrap set today? [09:34] ocaml? O.o [09:34] Oh, it's in main, so it must be. [09:34] blr: Yes, people are weird, sadly. [09:35] This kind of problem often fits functional languages quite well. [09:35] Indeed. [09:36] We could reduce bootstrap problems by allowing lp-buildd to degrade to failing versioned build-dep failures if it doesn't have dose. [09:37] ocaml is in main because of llvm-toolchain-* b-ding on dh-ocaml. Not sure why that is. [09:40] wgrant: yes, and then write libraries like this http://sumtypes.readthedocs.org/en/latest/ [09:42] wgrant: people are weird w.r.t. ocaml ? [09:44] lifeless: There are only three projects I've ever cared about written in it, and at least one of them (sks) has no reason to be. [09:44] lifeless: I've never met a quant, but I can imagine they would be very weird. [12:18] I think the leak may be through Storm's MutableValueVariable._event_system, but I'm just avoiding that part of the code entirely for now. [12:21] wgrant: Want me to organise a deployment this afternoon if possible, assuming that you defeat it before EOD? [12:31] cjwatson: Yeah, just waiting for tests to complete. [12:32] Then I get get -use landed and deployed tomorrow afternoon. [12:32] Thanks. [12:58] So perhaps the mere act of reading it is sufficient for doom. Bah. [13:04] Wow bug 1853 is in progress \o/ [13:04] Bug #1853: Project group "display name" is redundant with "title"

[13:07] The 20th-oldest still-open Launchpad bug report [20:16] I know you guys are super busy, but do you have 5 minutes to take a look to https://bugs.launchpad.net/loggerhead/+bug/1436483 ? The diff side by side in completely unusable, and I'm sure it's a 5 minutes fix [20:16] Bug #1436483: The diff page side by side has broken layout [20:43] rpadovani: certainly not looking very side by side there. [20:47] blr, :D I think it's due the radius borders someone added back in march [20:51] hmm, that's a different issue, I'm not sure that style should be applied there. [20:52] rpadovani: I think the issue is the padding [20:52] on .code, .linenumber [20:54] rpadovani: thanks for the report, will get that fixed now. I just need to see if that style is being used elsewhere (probably how this bug was introduced) [20:54] blr, thanks for the fast response, I should ping you guys more often :D [20:55] yeah, please do :) [21:07] rpadovani: wow, how long has this been broken, any idea? [21:08] bzr blame is suggesting that padding was added in 2011... [21:08] blr, so it isn't, because I used to use it until I opened the bug [21:09] rpadovani: ok, something else going on in that case (although that did fix it...) [21:12] blr, the width of .lineNumber [21:12] https://bazaar.launchpad.net/~loggerhead-team/loggerhead/trunk-rich/revision/484 [21:13] that fixes https://bugs.launchpad.net/loggerhead/+bug/310255 [21:13] Bug #310255: 6 digit line numbers get cutoff [21:15] rpadovani: looking for a diff that satisfies both [21:37] hmm I don't actually know who reviews loggerhead.. I guess we do? [21:44] rpadovani: thanks, cjwaston will probably want to look at that, but should be landed soon. [21:45] blr, thanks to you :-) [21:45] maybe we need css regression testing! [21:46] or a theme ex-novo, maybe responsing (just joking) [21:47] a responsive theme for LP would be great, but a fair bit of work I'd imagine. [21:49] tbh after the git support the thing I really would like are webhooks (and I see you are working on them) [21:49] yep, webhooks are coming :) [21:49] \o/