[01:36] <cjwatson> jelmer: Are there still remaining subvertpy blockers to getting bzr-svn or similar working with breezy?  If so can you explain how to reproduce them?  Maybe we can have somebody work on them ...
[01:38] <cjwatson> jelmer: Even if we were to decouple codeimport, it would still mean doing the py3 transition at the same time as the port to breezy if we don't have svn support working before you drop py3 support, which sounds like it could add some challenges; so I'm keen to get that out of the way if it isn't completely impossible / doesn't involve a huge pile of extra work from you
[01:38] <jelmer> cjwatson: there are basically two big challenges to getting bzr-svn working
[01:39] <jelmer> cjwatson: subvertpy's working copy API needs to be fixed to run with a newer version of subversion
[01:39] <jelmer> I got stuck getting that to work last time - it just segfaulted
[01:39] <jelmer> I don't think there is anything in svn itself that uses the old wc API anymore, so it may just be fundamentally broken
[01:39] <cjwatson> is that something subvertpy's own test suite shows up or is it only something higher-level?
[01:40] <jelmer> subvertpy's tests show that once you uncomment some of the self.skipTest() invocations
[01:41] <jelmer> the second big effort would be updating it to work with breezy's API
[01:42] <cjwatson> I don't know if this is an unanswerable question, but do you know if that's going to require changes in breezy itself?  that is, could it in principle be done to work with breezy 3.0 even if it doesn't happen before you drop py3 support?
[01:42] <cjwatson> Just trying to work out constraints
[01:43] <jelmer> cjwatson: we have no immediate plans to drop Python 3 support, I'm happy to wait with that until it's in a state where Launchpad can migrate
[01:43] <cjwatson> I have all the rest of LP ported (well, with the exception of utilities/community-contributions.py, which needs to be rewritten to work with git)
[01:43] <jelmer> cjwatson: I can't think of major changes to Breezy that would be necessary, but it's possible there will be some
[01:44] <cjwatson> ok, that sounds promising and gives me something concrete to dig into, thanks
[01:44] <cjwatson> been a long time since I looked at subversion's innards
[01:47] <StevenK> jelmer: Hopefully Python 2 support ...
[01:49] <jelmer> StevenK: euhm, yes :)
[01:49] <cjwatson> I made that exact same typo in my initial question, and caught it before hitting enter
[01:49] <cjwatson> Err
[01:49] <cjwatson> In fact I didn't catch it before hitting enter :)
[10:40] <cjwatson> https://pypi.org/project/vermin/ is an interesting (if mildly terrifying) tool
[10:41] <cjwatson> "Concurrently detect the minimum Python versions needed to run code"
[10:46] <SpecialK|Canon> cripes
[10:48] <cjwatson> It ... doesn't crash on Launchpad?
[10:50] <cjwatson> I think it might be a nice way to find obvious stuff in advance of getting to the point of running the test suite
[10:57] <SpecialK|Canon> Excellent :)
[12:46] <cjwatson> http://lpbuildbot.canonical.com/builders/lp-devel-xenial/builds/970/steps/shell_9/logs/summary   wow, what even was that code
[12:48] <SpecialK|Canon> I cannot read PackagePublishingPocket without hearing the opening line of http://holyjoe.org/poetry/anonE.htm
[12:49] <cjwatson> It is a bit
[12:50] <cjwatson> https://code.launchpad.net/~cjwatson/launchpad/+git/launchpad/+merge/378040 should fix that
[12:50] <cjwatson> (and does in my tests)
[12:51] <SpecialK|Canon> Oh wonderful!
[12:52] <SpecialK|Canon> (approved)
[12:52] <SpecialK|Canon> (heartily and enthusiastically)
[12:54] <cjwatson> Heh