[02:57] lifeless: Upgrading from fixtures 0.3.10 to 0.3.11 generates some garbage. Any ideas why? [02:58] Tests generated new (2) garbage: [02:58] [, ] [02:58] wgrant: You've really been on an upgrade kick ... [02:59] StevenK: Well, we're 3-4 years behind on most things [02:59] Last week we upgraded Zope and YUI, this week I'm upgrading everything else. [02:59] We weren't that far behind on YUI [02:59] Maybe we need to downgrade to 2 or something === cyclicflux is now known as Guest39382 === Guest39382 is now known as CyclicFlux === jam1 is now known as jam [06:43] containing_div_node.addClass('yui3-skin-sam'); [06:43] Which doesn't help [06:43] Calendar, you're beginning to annoy me [07:03] wgrant: garbage? [07:13] lifeless: gc.garbage that remains after a gc.collect() [07:13] Presumably due to reference cycles [07:13] I don't know why that isn't collected, TextIOWrapper -> BytesIO is a one way ref. [07:14] Yeah, that's what it looks like :/ [07:14] Maybe the TextIOWrapper is referenced elsewhere [07:15] But then it shouldn't be garbage [07:16] Gooooooood morning [07:17] Morning czajkowski [07:29] wgrant: so the Q is can you reproduce it. [07:29] Gnight. [07:32] lifeless: It's easy to reproduce as part of a particular LP test, but I was going to check if you knew anything before isolating. [07:32] Will do so [07:34] cjwatson: when you're about can you give me a shout please, I seem to have broken something on my machine and in need of your assitance, for this I shall send you cake and biscuits :) [08:28] czajkowski: Hi [08:29] wgrant: So, I think I ended up not using fmt:link here because an earlier version rendered a link for primary archives as well (and I've reverted to that version per your feedback) and there's no link implementation for Archives other than PPAs [08:30] wgrant: And as a result the TAL ended up being annoyingly cumbersome with fmt:link, rather than simpler as it's meant to be [08:30] cjwatson: Oh, it just leaves primary archives unlinked? [08:30] And copy archives too? [08:31] cjwatson: shall move to kernel and fill you in there [08:31] or is foundation better? [08:31] czajkowski: What's it about? [08:31] There's no foundations channel, intentionally [08:31] ahh [08:31] And I'm not a kernel developer, so I doubt that's appropriate :) [08:31] cjwatson: on start up this morning I'm http://ubuntuone.com/1REKSBU0jU7XITkpfBK0RQ [08:32] wgrant: Trying to use fmt:link on a primary archive raises NotImplementedError [08:32] cjwatson: Well that's a bit crap. [08:32] wgrant: Maybe I should add a generic link implementation for archives that does ? [08:32] Yet easily fixed, and it probably makes sense [08:32] Yeah [08:33] Copy archives should probably use the PPA icon, but primary/partner can use the distro one [08:33] And we can use that for all other archives, and probably delete some insane TAL elsewhere [08:33] They could even link directly to the distro [08:33] czajkowski: Nothing wrong with that - you might get the menu when you don't normally if a previous boot failed [08:34] But just press Enter and carry on [08:34] cjwatson: aye I did then I went to ubuntu advanced and moved down to recovery kernel. but no matter how many reboots I still up at that screen :/ [08:34] recovery kernel won't count as a successful boot [08:35] the point of this is to show the menu when people seem to be having trouble [08:36] now, if you see it after a non-recovery boot, that's a bit odd and we can investigate in #ubuntu-devel [08:37] ahhhh [08:37] cheers cjwatson :) [08:45] wgrant: Do directives in lib/lp/app/browser/configure.zcml magically pick the closest match somehow, or do I need to take inheritance into account? [08:46] Also, I happened to notice that https://bazaar.launchpad.net/~launchpad/+junk/dbpatches/view/head:/allocated.txt seems to be missing 2209-43-0 [08:47] cjwatson: It should pick the closest adapter [08:48] So an IPPA adapter will override one for IArchive for PPAs [08:51] OK. Though, hmm, I wonder how "closest" is defined given that IPPA is only done by alsoProvides [08:52] Oh, true [08:52] That's a good question... [08:52] Maybe it'd be clearer to turn PPAFormatterAPI into ArchiveFormatterAPI and do the checks explicitly [08:53] Quite likely. [09:03] wgrant: Copy archive icon - using the PPA icon would be inconsistent with ArchiveImageDisplayAPI.icon ("ArchivePurpose.COPY: '/@@/distribution'"). Do you want that changed? [09:08] cjwatson: I'd forgotten there was precedent, so go with that. [10:03] wgrant: Is there any precedent anywhere for what to do when trying to render a copy from an inaccessible private PPA? The current version is 'Sync from , ...'; if I do the naïve thing and replace that with 'Sync from , ...', I end up with "Sync from , ..." in the output which is less than ideal [10:22] I guess something like SourcePublishingRecordView.linkify_source_archive [10:43] cjwatson: For private teams we say , IIRC [10:43] Should probably do a similar thing here. [10:43] At the moment I think we must show the archive displayname [10:44] But I suppose arguably that's wrong [10:45] Perhaps also "Building private job" as precedent [10:45] Yeah, maybe [10:45] So it's tempting to say "Sync from private archive" [10:46] Indeed [11:06] wgrant: OK, would you mind re-reviewing https://code.launchpad.net/~cjwatson/launchpad/queue-copy-archive-links/+merge/159250 ? I know you approved already, but the changes are fairly substantial. [11:07] I suspect I'll have some doctest damage, which I'll let EC2 catch [11:19] cjwatson: That looks fine. Thanks. [11:25] Now, I wonder why my EC2 setup is busted ... http://paste.ubuntu.com/5715669/ [11:37] cjwatson: Seems to be a change in python-boto. http://paste.ubuntu.com/5715687/ seems to fix it [11:37] Hm, the python-boto package hasn't changed since last May [11:38] I wonder if it's an EC2 change tickling an existing bug [11:38] Could be [11:38] I use it so infrequently that I didn't bother to do more than hack it to work. [11:51] wgrant: Much better, thanks === cjohnston_ is now known as cjohnston === wedgwood_away is now known as wedgwood === wedgwood is now known as Guest86505 === wedgwood_ is now known as wedgwood === BradCrittenden is now known as bac [16:11] Grr, these failures didn't show up in ec2, probably because the test runner there apparently hung :-/ [16:12] Oh well, should be easy enough to fix [16:48] https://code.launchpad.net/~cjwatson/launchpad/queue-copy-archive-links-testfix/+merge/159442 - test fixes, could use a review === deryck is now known as deryck[lunch] === olli_ is now known as olli === deryck[lunch] is now known as deryck === wedgwood is now known as wedgwood_away