[02:05] <StevenK> wgrant: http://pastebin.ubuntu.com/5991234/
[02:09] <wgrant> StevenK: Does it work?
[02:09] <StevenK> wgrant: Total: 562 tests, 0 failures, 0 errors in 6 minutes 12.954 seconds.
[02:10] <StevenK> That's -t test_product, which hits a lot, I'm kicking off -m blueprints now
[02:11] <wgrant> Right
[04:10] <StevenK> wgrant: Hm, do you think bug 1212703 is due to PTUJ ?
[04:10] <_mup_> Bug #1212703: translations imports say wrong uploader <Launchpad itself:New> <https://launchpad.net/bugs/1212703>
[04:10] <wgrant> StevenK: That isn't about packages, is it?
[04:11] <StevenK> wgrant: I'm not sure, TBH
[04:12] <wgrant> Those URLs are within a project.
[04:12] <StevenK> Right, so unrelated.
[04:12] <wgrant> I'd investigate the code to see how the uploader is determined.
[04:12] <wgrant> It could, for example, be from the last-changed address in the pofile.
[04:13] <wgrant> In some circumstances.
[04:13] <StevenK> wgrant: I wasn't going to investigate it, I was just wondering if PTUJ might have broken it, that's all.
[04:15] <wgrant> Ah
[04:16] <StevenK> (Read as: I'm still fixing test failures)
[04:52] <wgrant> StevenK: I'm about to land an upgrade to bzr 2.6.0, with new bzr-git, bzr-svn, dulwich, cscvs. I'll want to do a fair bit of QA on this, so if you have something ready that you want deployed soon, complain now :)
[04:53] <StevenK> wgrant: Ooooh
[04:53] <StevenK> wgrant: Any real fallout from the bzr upgrade?
[04:53] <wgrant> cscvs was using a removed deprecated method
[04:53] <wgrant> And LP needed a few lines of changes.
[04:53] <wgrant> The main reason for this is the new bzr-git and dulwich, but I thought I might as well do the whole lot.
[04:54] <StevenK> Right
[04:55] <StevenK> wgrant: Nothing I'm really attached too
[04:55] <StevenK> wgrant: But, microservices? :-)
[05:06] <wgrant> Slow test suite is slow.
[05:07] <StevenK> Bleh
[05:07] <StevenK> I've deleted it from the property cache, why you not work :-(
[05:10] <StevenK> Hm, except I haven't, it seems
[05:16]  * StevenK peers at what is calling getDelta
[05:16] <StevenK>  0.2s getDelta: get_property_cache(self).__dict__={'linked_branches': []}
[05:16] <StevenK>  0.2s getDelta:
[05:16] <StevenK>       get_property_cache(self).__dict__={'linked_branches': [], 'work_items': []}
[05:36] <cjwatson> Has anyone managed to get the dogfood builders upgraded so we can QA the ABORTED status changes?
[05:37] <wgrant> We were discussing that this morning.
[05:37] <wgrant> But not yet.
[05:41] <cjwatson> I'm at DebConf and my home ADSL has gone down, so I'm rather lacking in e-mail and IRC context at the moment.
[05:41] <wgrant> Oh, fun.
[05:41] <wgrant> (though this discussion was by voice, so you didn't miss IRC context)
[05:44] <cjwatson> Right.  Just in case anyone was expecting me to know anything that lives in the external bits of my brain. :-)
[05:46] <cjwatson> I'm on vacation next week, so if it takes until then, then somebody else will have to deal with unblocking the deployment pipeline.
[05:46] <StevenK> cjwatson: wgrant is about to stop it up away
[05:46] <StevenK> New bzr version
[05:46] <cjwatson> Actually, hmm, I suppose that the proper QA for that branch is to make sure that it doesn't break with the *old* builders
[05:47] <wgrant> We'd ideally upgrade some fraction of them
[05:47] <wgrant> To test that it works with both
[08:09] <jtv> Is anyone else having login trouble?
[08:09] <jtv> I enter my username and password, and instead of the SSO page I get "Your page was stale."
[08:10] <jtv> Deleted all launchpad cookies, reloaded everything... no difference.
[08:14] <jtv> I'm finding some references on the intertubes, but they're fairly old.
[08:19] <wgrant> jtv: You're probably not sending Referer to SSO
[08:19] <jtv> No..  should I put it in a SASE?
[08:20] <jtv> This is Chromium BTW...
[08:21] <wgrant> Have you tried another browser, or a fresh profile?
[08:22] <jtv> Firefox works fine.  It seems ScriptSafe is blocking referers — I have *.launchpad.net and *.ubuntu.com in my whitelist, but maybe the whitelist doesn't apply to referer blocking, or maybe there's another domain I need to enter?
[08:23] <wgrant> login.launchpad.net and login.ubuntu.com are the two domains that matter.
[08:24] <jtv> Then ScriptSafe is probably failing to honour the whitelist for click-through referer blocking.
[08:25] <jtv> Grr..  Disabling referer blocking didn't do the trick.
[08:27] <jtv> Ah, there's _also_ an option for sending a fake referer, which disables separately.
[08:29] <jtv> wgrant: that worked!  \o/
[08:50] <wgrant> jtv: Excellent
[08:52] <wgrant> jelmer: Hah. Combining those two bug fixes breaks Google Code imports again. Google Code's HTTP server errors unless you support thin-pack, which we were accidentally before because of the negotiation bug in dulwich, which explains the ref deltas that revealed the bzr-git bug. Do you recall why thin-pack is disabled on fetch and push by bzr-git? It seems to work fine.
[08:54] <lifeless> Maybe Jelmer can fix :)
[08:55] <wgrant> It's trivial to fix, I'm just trying to work out why it's been disabled since 2009.
[08:55] <lifeless> wgrant: I meant on the Google server side ;)
[08:55] <wgrant> Oh, heh.
[08:56] <wgrant> It sadly just seems to ISE in the face of any missing extensions :(
[12:26] <jelmer> wgrant: hi
[12:27] <jelmer> wgrant: I think the reason thin pack is disabled is that it can require expensive operations on the client side
[12:27] <jelmer> so it's a CPU vs bandwidth thing