[13:22] <cjwatson> Wow, you get amazingly long test output if you try to @provider something that you forgot to make a subclass of Interface
[13:22] <wgrant> Heh, what is the bulk of it?
[13:23] <wgrant> ZCML spam?
[13:23] <cjwatson> In this case, TypeErrors reporting duplicate enumerated types from everything else that imports the module in question
[13:23] <wgrant> ... nice.
[13:24] <cjwatson> I'm not sure I can actually do this though.  I was trying to turn SourcePackageRecipeData into a <securedutility> so that I could use its getParsedRecipe staticmethod from browser code, but its __init__ takes arguments ...
[13:25] <wgrant> component= rather than class=.
[13:25] <cjwatson> Aha
[13:26] <cjwatson> Of course, thanks
[13:27] <cjwatson> Maybe in fact just a <utility> since RecipeBranch doesn't have a corresponding Zope interface
[13:44] <cjwatson> wgrant: db-git-recipes> you got me on indexes, but why does it need deletion handling in the DB patch?  there's no such handling for base_branch, and I was going to add it to GitRepository._getDeletionRequirements (which, yes, has the same "somebody else creates a recipe that you can't delete" as for Branch, but it's not a new problem, and deleting repositories is less frequent than deleting branches anyway)
[13:46] <wgrant> cjwatson: Hm, how does base_branch get around the branch deletion tests?
[13:47] <wgrant> There is at least a test that lists all FKs.
[13:48] <cjwatson> wgrant: That test whitelists SPRD.base_branch
[13:48] <cjwatson> Exactly because of the explicit deletion handling for recipes in Branch, I'd assume
[13:49] <cjwatson> (Also sourcepackagerecipedatainstruction.branch)
[13:50] <cjwatson> And there's no equivalent of that FK listing test for Git at the moment
[13:51] <wgrant> Sounds acceptable, then.
[13:54] <cjwatson> OK, I'll fix the indexes.
[13:54] <wgrant> Thanks.
[13:54] <cjwatson> Down to 25 failing recipe tests at the moment ...
[13:54] <wgrant> Are you parsing Git recipes with bzr-builder, or switching implementations?
[13:54] <cjwatson> The former, with a small hack.
[13:55] <wgrant> Excellent.
[13:56] <cjwatson> (Hack: Git recipes may start with either "# bzr-builder" or "# git-build-recipe", the former for compatibility and the latter because otherwise it looks rather WTF to people who don't know the history.  But bzr-builder won't accept the latter, so we have to munge it.)
[18:43] <cjwatson> wgrant: Could you sort out your pending QA?  I'd like to be able to deploy.
[21:15] <wgrant> cjwatson: Oh, sorry, forgot about that.
[21:47] <wgrant> cjwatson: QA is clear.