[01:23] * StevenK stabs the blueprints tests [01:23] test_hasspecification is calling _valid_specification on a projectgroup with an open spec, and an obsolete spec, and asserting both are returned [01:25] Which makes me think the test is buggy, since valid specifications are not obselete or superseded ... [03:27] wgrant: How goes the BFJO murder? [04:07] StevenK: https://code.launchpad.net/~wgrant/launchpad/handleStatus-refactor-1/+merge/144049 [04:07] StevenK: As a bonus it makes buildd-manager logging suck a bit less [04:07] Hah [04:07] When a build completes it will actually log that the build completed, and what its status was [04:08] Currently you get, on some types of failure, '******* rother is MANUALDEPWAIT *******' [04:08] Without any reference to what the actual build was [04:08] And on success or a real build failure, you get no logging at all :) [04:12] 8- @classmethod [04:12] 10+ @defer.inlineCallbacks [04:12] Already fixed and retesting locally [04:13] Right [04:13] Was hoping you wouldn't notice :) [04:13] It was the first change in the diff, bit hard [04:13] It doesn't actually make any difference, but yeah [04:14] wgrant: That is my only concern with that branch [04:14] Thanks [04:15] I love the _handleStatus_generic_failure refactor [04:17] StevenK: It's pushed [04:17] And passes tests too [04:20] wgrant: r=me [04:20] Thanks sir [04:20] Right, hopefully the thinko I've fixed makes the tests pass. [04:21] :) [04:21] Aw. [04:21] It fixed one test, though [04:22] How many failures remain? [04:22] 4 [04:23] ah [04:23] ez [04:23] From -m blueprints -m registry [04:23] And I've also bent {Distro,Product}Series to my will, and destroyed Specification.completeness_clause [04:23] Excellent [04:23] As I plotted [04:24] 14 files changed, 383 insertions(+), 765 deletions(-) [04:24] Hmm [04:24] Only a 50% win [04:24] Although I guess it also fixes some omissions [04:24] It drops a massive amount of duplication [04:25] Right, I just expected more [04:25] Perhaps I've missed some bits that can die, though [04:34] Hm, I think my LEFT JOIN on product is screwing up my productseries call [04:34] StevenK: bugtasksearch does the same thing [04:34] You can probably steal ideas from there [04:34] It might well omit the check if called in a product or productseries context [04:36] It does it differently [04:37] How differently? [04:38] Currently, I do it all the time, and bugtasksearch will only do that check if product, distribution, productseries and distroseries are all unset. [04:38] Right, that makes sense. [04:38] The way bugtasksearch does it [04:39] Possibly because I wrote it :) [04:39] distribution/distroseries are obviously fine because they can't have a product [04:39] Yeah [04:39] So it gets included for person/project group [04:39] And product/productseries are fine because they aren't accessible unless they're active or you can see inactive projects. [04:39] person/projectgroup/*, yes [04:40] Now I have to pass in the context [04:40] You weren't already? [04:40] No, I pass in a base clause [04:40] Specification.productID == self.id or so [04:40] Ah [04:47] I could check base_clauses[0].expr2.table, but you might murder me [04:48] Might? [04:49] :-) [04:51] wgrant: So handleStatus_OK is up next? [04:52] storeBuildInfo is being demolished [04:52] Well, substantially revised [05:27] Ugh, I *think* this will fix the productseries failures [05:28] Indeed, it does. [05:34] wgrant: Is http://pastebin.ubuntu.com/1554189/ a 2.7 failure? [06:13] StevenK: There shouldn't be any more 2.7 failures, but that's possible. [06:14] Hm [06:16] I've seen this before [06:16] Perhaps 18 months ago [06:16] But i don't remember quite what it was [06:17] I've ignored it for the time being, trying to figure out these person failures [06:21] StevenK: Oh, I wonder if it's proxied [06:23] Yeah [06:23] 2.7 doesn't like securityproxied dicts [06:23] 2.6 is fine [06:25] Haha, so it is a 2.7 failure [06:25] I wonder how that was missed [06:25] Perhaps it is new [06:25] Ah [06:25] A change in September [06:26] - Issue #15801: Make sure mappings passed to '%' formatting are actually [06:26] <_mup_> Bug #15801: mozilla-firefox-locale-tr: new changes from Debian require merging < https://launchpad.net/bugs/15801 > [06:26] subscriptable. [06:26] That would be how it was missed, I guess. [06:26] A regression in 2.7.something [06:26] I can rSP to work around it [06:27] http://hg.python.org/cpython/rev/2801bf875a24/ [06:29] (I assume) [06:29] Plausible [06:31] StevenK: shouldn't need to, just make sure all the attributes needed for the new check pass [08:46] good morning === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away [14:46] wgrant,StevenK: I have a QAed DB patch - do you think I could have an FDT? [14:58] having our only maintenace folks in the one timezone is not ideal :( [14:59] I'm not in a desperate rush beyond "next couple of days would be nice" [20:09] * mwhudson laughs at cjwatson getting 30k+ lines of LoC credit in one go [20:10] the graph looks awesome :) [20:11] http://people.canonical.com/~cjwatson/tmp/loc-cum.png [20:14] would be bad if that was a stock market index [20:15] cjwatson: waaaa? [20:17] what did you nuke? [20:23] database/schema/launchpad.html, last updated 2009 [20:24] make target to build it locally in a few seconds if anyone cares [20:25] kind of cheating of course, but misleadingly obsolete doc is a maint problem imo [20:25] Never hesitate to poke fun at the arbitraryness of a metric? :-) [20:33] well, its accurate so far :> [20:33] cjwatson: nice catch [21:56] cjwatson: Sure, we should be able to do it in a few hours