[01:34] Huh, sqlobject doesn't have GROUP BY? [01:35] cjwatson: SQLObject has just about nothing. [01:35] cjwatson: Stormify that stuff. [01:35] But you probably want DISTINCT distroseries, pocket anyway. [01:35] Stormifying xPPH seems like it might take a while [01:36] Hm, what are you doing? [01:37] Looking at the next stage of optimising publisher phase A, so changing getPending*Publications to return everything; seemed like it'd be easiest to group it by series to get roughly the same ordering (hence logging) as before [01:38] That's ORDER BY, not GROUP BY. [01:38] Oh, er, cough. I speak SQL good, I learn him from a book. [01:38] Heh [01:39] And SQLObject has ORDER BY, good. [01:39] Also, getPendingPublications doesn't use any of the other xPPH methods; you can Stormify it alone without a problem. [01:40] Oh, I guess that's true, yes [01:41] Hm [01:42] Tempting to fix all the suite references while I'm touching almost the same code for archive references. [01:50] Also not sure whether to leave the reference for primary archives as 'ubuntu', or to make it 'ubuntu/primary' like the rest. [01:51] I always found ubuntu/primary weird. I think just ubuntu is an improvement. [01:51] Marginally. === Guest41889 is now known as wallyworld [06:59] * wgrant tries to track down all of the ancestry implementations. [07:00] There's NascentUpload.get(Source|Binary)Ancestry and lp.soyuz.adapters.overrides, which I think are the two that matter, but then I'm pretty sure there's at least another two elsewhere that are used for varying purposes. [07:01] PackageUploadSource.getSourceAncestryForDiffs [07:01] And check_copy_permissions [07:04] I think all except getSourceAncestryForDiffs are relevant for ubuntu-rtm, as check_copy_permissions assumes a package will require manual approval if no overrides are found. [08:35] cjwatson: Oh, https://code.launchpad.net/~wgrant/launchpad/archive-references-everywhere/+merge/226061 basically includes your pcj-repr changes. [08:35] I can back those bits out, I guess. [08:35] wgrant: Oh, I can just withdraw my branch [08:36] That's probably easier. I had tests running overnight after I basically grepped through and replaced almost every reference to archive.owner.name and archive.name. [08:36] So it wasn't pushed yet, sorry. [08:37] Not a problem. [08:40] cjwatson: For the ubuntu-rtm needing to inherit ubuntu's overrides problem, I'm thinking we might just want a flag on DistroSeriesParent which lets overrides fall back to a parent series' primary archive. [08:40] Also, did we decide that they wanted to skip NEW in the fallback case? [08:46] A flag would seem sensible. I thought we wanted to just do NEW as normal in the fallback case though. [08:46] I believe we agreed it was OK for #ubuntu-release to manage that. [08:46] I couldn't remember, either works. [08:46] getBinaryAncestry and getSourceAncestry die tonight. [08:48] http://paste.ubuntu.com/7769369/ is my notes for the various cases atm. [08:49] We only want to force component to universe/multiverse in some limited cases. [09:10] cjwatson: Don't we always want to do that if there's no ancestry in either archive? [09:15] New binaries uploaded to Ubuntu with source in main default to main. [09:15] For source, yes. [09:15] Er yeah, forgot that binary override complexity, good point. [09:27] is it still the case that backports can't build-depend on other backports? or has that been fixed? [09:28] That's been fixed. [09:28] cool. [09:29] https://bugs.launchpad.net/launchpad/+bug/888665 [09:29] <_mup_> Bug #888665: Backports can't build-depend on other backports [12:15] cprov: Thanks. [12:16] wgrant: no problem, I am still owning you few reviews. === daker_ is now known as daker === lifeless1 is now known as lifeless === ajmitch_ is now known as ajmitch