[04:09] <hallyn_> stgraber: creating saucy containers fails, becuase "chroot $rootfs /var/lib/dpkg/info/openssh-server.postinst configure" fails with "invoke-rc.d: initscript ssh, action "start" failed."  (reproducible by hand)
[15:17] <stgraber> hallyn_: yeah, I saw that here too, something must have changed in the way ssh is restarted from the postinst, I'll try to figure out a patch this weekend
[16:58] <cjwatson> stgraber: I fixed openssh-server's postinst a little while ago to be policy-compliant and use invoke-rc.d
[16:58] <cjwatson> stgraber: So if you were relying on its policy-non-compliance then something might have broken
[17:00] <cjwatson> OMG what on earth is lxc-ubuntu doing with ssh.conf - couldn't you folks at least sed the existing one or something? :)
[17:04] <stgraber> cjwatson: oh, yeah, ignore the changes we do in trim(), nobody uses that anymore and I'm hoping to kill it for good by 14.04.
[17:04] <stgraber> cjwatson: in the standard code path, what we do is move ssh.conf to ssh.conf.disabled, call the postinst then move it back
[17:05] <stgraber> cjwatson: as it was the only easy way to deal with that back in 13.04, now that openssh respects invoke-rc.d, I'll just add a policy-rc.d (exit 101) and that should fix the issue (we still need the .disabled stuff until 12.04 EOL though...)
[17:54] <cjwatson> stgraber: Yeah, policy-rc.d should do it.  In <13.10, though, you could have handled it by diverting initctl
[17:55] <cjwatson> And maybe start-stop-daemon too
[17:58] <infinity> cjwatson: Does remove-package commit in a single transaction, or is it doing a round-trip removal for each binary I list on the command-line?
[17:58] <infinity> (Wondering if removing, say, kernel NBS, is being split across multiple publisher runs)
[17:59] <infinity> Not that it matters, in practice, since NBS stuff shouldn't be needed anymore, but it feels a bit weirdly wrong if it ends up spanning runs.
[17:59] <cjwatson> infinity: If you remove an entire source that's in one go, but if you're removing lots of binaries without removing their corresponding source (as in NBS) that's multiple API calls
[17:59] <cjwatson> Since it's BPPH.requestDeletion
[17:59] <infinity> Kay.
[18:00] <infinity> I wonder if that could do with a batch mode API call.  Would also speed it up tremendously.
[18:00] <infinity> (Today's multiple-kernels-plus-a-couple-libraries NBS run takes 10-15m with a high latency connection to LP)
[18:00] <cjwatson> I wouldn't object, although I don't think it's the most important thing on the list or anything
[18:01] <infinity> Nah, not dreadfully important.  I'm used to just watching it scroll slowly in the background, as it's doing right now. :P
[18:02] <cjwatson> There's already PublishingSet.requestDeletion, which would be the right thing to export, but I'm not sure how to export *Set things
[18:02] <infinity> I've just been doing some NBS-clearing uploads in my wind-down-to-travel lull overnight, so it was on my mind.
[18:03] <cjwatson> At least in cases where there isn't an obvious collection
[18:03] <cjwatson> Maybe it would have to be something artificial like a method on Archive; don't know
[18:04] <infinity> I wonder if people naming their projects ever read them out loud.
[18:04] <infinity> "kraptor" sounds like an 80s cartoon villian who lives in the sewers.
[18:13] <infinity> Laney / xnox: You guys are TIL on vtk, do you happen to know what's gone wrong with that's making yade FTBFS now? https://launchpad.net/ubuntu/+source/yade/0.97.0-4
[18:17] <infinity> Hrm, maybe that's a missing build-dep (or a missing dep)...
[18:43] <infinity> Laney / xnox: Ahh, that's http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=714935 ... I wonder why it didn't seem to pop up until now.
[23:03] <Noskcaj> roaksoax, It seems testdrive-gtk is broken in version 3.20