=== Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [21:34] is the LoC delta policy still being operated? [21:37] Laney: To a reasonable extent. [21:37] If someone wants to add 5000 lines without taking anything away I will probably tell them to go away. [21:40] See #launchpad earlier; I just want to expose a property on PackageUpload. Shouldn't be that big (code-wise, never sure about tets). [21:40] That's fine. [21:40] Will need a bit of extra code to preload the relevant objects to avoid timeouts, but still pretty easy. [21:42] * Laney nods [21:58] Laney: I think the PU tests are reasonably rational these days. [21:58] lib/lp/soyuz/tests/test_packageupload.pyu [21:58] -u [21:58] Actually that wouldn't be for webservice stuff would it [22:00] Oh yeah, the webservice tests are in there as well. [22:00] class TestPackageUploadWebservice [22:15] wgrant: I've landed everything I want in the next lp-buildd release, FWIW. Not going to have time to QA it tonight [22:16] The master side is -106 so far although without handling builder_version yet [22:17] cjwatson: Yeah, it should be a bit cleaner now you can avoid handing both the dict and list around. [22:18] I can only assume the code that mangled it into a dict was written well after the original slave code, or else that whoever wrote it didn't know that you could pass dicts through xmlrpc. [22:19] It was written well after the slave code. [22:19] That makes some sense at least. If they'd been written together it would have been manifest lunacy [22:19] I think the dicts happened with the introduction of BFJB in 2009. [22:19] Because the BFJB can extend the dict with its own keys. [22:20] Yeah. Which I've destroyed. [22:20] (that's why it gets both the sentence and the dict) [22:20] Ah, good :) [22:20] Because it needed to die. [22:20] It wasn't providing any significant value and means the BFJB is involved in yet another place it shouldn't be. [22:26] Oh, I see you went the whole way and matched the names to the existing ones on the master side. I guess that makes sense. [22:27] I figured why not. [22:27] Yeah [22:27] Otherwise I have to have a totally pointless thing that renames them. [22:27] builder_status is probably sufficiently clearer that it makes sense. [22:27] Yep [22:31] cjwatson: I haven't noticed any other regressions. Have you seen any? [22:31] From 116/117? No [22:31] Great. [22:31] The depwait one has the amusing effect of turning things into upload failures [22:31] Yes [22:31] But that was obvious once I looked at it [22:31] That's how I first noticed it. [22:32] I got Alexander to land my builddfitzer change and he said he'd include a reminder to update local checkouts in his report, so hopefully other webops will use it [22:32] Yep, saw that. [22:32] Thanks. [22:32] But it might be worth watching out for cases where builders appear to have been reset rather than rebuilt