[10:01] <mwhudson> omg i actually have to wait for a ppa build to start
[10:01] <mwhudson> haven't had that happen for months
[10:04] <wgrant> mwhudson: Hm, that's a bit weird.
[10:04] <wgrant> Oh someone is building about a billion mysqls and mariadbs.
[10:04] <wgrant> And lcy01 fell over.
[10:05] <mwhudson> i was wondering if that many buildds in cleaning was normal
[10:07] <wgrant> Well.
[10:07] <wgrant> It's normal, but not normal, if you know what I mean.
[10:07] <wgrant> Not meant to happen, but, well, lcy01 is thpecial.
[10:10] <mwhudson> i thought the cloud fixes everything
[10:10] <mwhudson> did i read that memo wrong?
[10:13] <wgrant> Clouds sometimes mean rain.
[10:14] <wgrant> mwhudson: Stabbed a few things, and the queue should be clear in a minute or two.
[10:15] <mwhudson> wgrant: awesome thanks
[10:16] <mwhudson> wgrant: now can you make the publisher a million times faster
[10:16] <wgrant> mwhudson: I'm not a wizard.
[10:17] <mwhudson> shame
[10:46] <cjwatson> Hm, not convinced that twisted.web.client.HTTPConnectionPool actually works quite the way I was thinking it did.
[10:46] <cjwatson> maxPersistentPerHost limits the number of persistent connections it retains, but not the number it's willing to open in parallel.
[10:50] <cjwatson> I think I'll need a subclass that overrides _{get,remove,put}Connection with some extra Deferreds.
[11:01] <wgrant> That is unfortunate.
[11:02] <wgrant> But Deferreds fix everything.
[11:06] <cjwatson> Including "my program is too easy to debug".
[11:07] <wgrant> That's most of the point.
[11:11] <cjwatson> I think DeferredQueue is probably my friend.
[11:27] <cjwatson> Ah, no, DeferredSemaphore.