[19:57] <barry> thomi: howdy!
[19:57] <thomi> hey barry
[19:59] <barry> thomi: so the good news is that all the zope stuff has cleared -proposed now.  the last package to port is gunicorn.  hope that's still something you want
[19:59] <thomi> yes please!
[19:59] <thomi> I think my version is somewhat hacky, but it's an apprach we've used in other packages before
[19:59] <thomi> barry: if you change it to no longer have the 'gunicorn3' command then I'll need to update some charms is all
[20:00] <barry> thomi: i'm going to refer to it, but probably redo it.  it's not team maintained so we either have to carry deltas in ubuntu (undesirable for the long-term) or make sure the debian maintainer is happy with the changes (my planned approach).  yep, i'm still planning on keeping the gunicorn3 command
[20:01] <thomi> awesome
[20:01] <barry> thomi: just to verify - that's the last blocker for you guys, right?
[20:02] <thomi> barry: well, it's not a blocker since we deploy with a PPA, 'less-than-desirable' would be a better description
[20:02] <thomi> we mainly want to make sure the packages aren't doing something really bad, since we're now deploying a production service with these hacks in place
[20:03] <barry> thomi: cool.  you should basically be able to grab all the utopic packages for everything else now
[20:03] <barry> and sorry it took so long.  zope.* is a *complicated* stack (especially throwing in the ubuntu-only packages)
[20:04] <thomi> no worries
[20:05] <thomi> it seems like we've (well, you did all the work) substantially improved the ability of web devs to use the distro packages
[20:05] <thomi> rather than have to rely on pypi
[20:05] <thomi> which can only be a good thing
[20:06] <barry> that's the idea :)
[20:06] <barry> thanks for sticking w/ py3
[20:07] <thomi> barry: now cgoldberg is doing another similar project with the Ci team and is hitting similar problems :-/
[20:07] <barry> thomi: different packages?
[20:07] <thomi> yeah
[20:08]  * barry looks for a rock with sufficient crawl space
[20:08] <thomi> swiftclient and ampqlib are the two we know about right now
[20:08] <thomi> I was going to look at them yesterday, but had to take the day off sick
[20:08]  * thomi looks into swiftclient now
[20:09]  * barry nods
[20:10] <thomi> ugh.
[20:10] <barry> that bad?
[20:10] <thomi> swiftclient tests don't run, I think because the utopic tox command has changed the way it parses it's config files
[20:10] <thomi> tox.ConfigError: ConfigError: substitution key 'posargs' not found
[20:12] <thomi> looks like swiftclient will also need the following ported to py3: eventlet, keystoneclient
[20:13] <thomi> hmmm, but eventlet appears to be py2 only
[20:16] <barry> thomi: https://bitbucket.org/eventlet/eventlet/commits/65845073a4eb759ae1cf3c62ec7e953a54548de5
[20:16] <thomi> actually, it seems that Dependency is innacurate
[20:16] <thomi> swiftclient depends on eventlet but never imports it
[20:17] <thomi> and the changelog has an entry about removing it
[20:17] <thomi> I guess the DPM never caught up with that
[20:17] <thomi> so... "good news everyone"
[20:17] <barry> \o/