[05:18] <wgrant> cjwatson: Huh, I thought pocketlint caught all the double equals
[05:54] <cjwatson> wgrant: As far as I can tell it has no double-equals handling
[05:55] <wgrant> Hm./
[05:55] <wgrant> cjwatson: Also what's with the new assertDictEqual that requires rSP? Do we need to exempt some new attrs from checks, or is it just type-checking?
[05:56] <cjwatson> wgrant: What branch is this in, sorry?
[05:56] <wgrant> cjwatson: https://code.launchpad.net/~cjwatson/launchpad/upgrade-testtools/+merge/335395 sorry
[06:01] <cjwatson> wgrant: Looking harder I'm actually somewhat confused as to how it worked before.  unittest.TestCase.assertDictEqual does explicit isinstance checks.
[06:02]  * cjwatson pdbs
[06:03] <wgrant> We didn't monkeypatch in zope_isinstance, did we?
[06:04] <wgrant> Doesn't look like it.
[06:05] <cjwatson> I think it has something to do with our own TestCase.assertIsInstance, but checking why that's no longer used.
[06:07] <cjwatson> Oh, right
[06:08] <cjwatson> testtools now prefers unittest2, and that does self.assert_(isinstance(...)) rather than self.assertIsInstance()
[06:08] <cjwatson> Which is kind of a regression from plain unittest IMO, but at least the impact is small
[06:09] <wgrant> Ah, fair enough.
[06:10] <cjwatson> Ah, but it's fixed in newer unittest2
[06:10] <cjwatson> So maybe the right answer is to upgrade that instead
[06:10] <wgrant> Aha
[06:17] <cjwatson> wgrant: Upgrading unittest2 and reverting that change passes the relevant tests; I'll do a full test run, though
[06:19] <wgrant> cjwatson: Excellent, thanks.
[06:20] <cjwatson> (https://hg.python.org/unittest2/rev/071e84843744)
[06:21] <cjwatson> https://hg.python.org/unittest2/rev/2ce0bdfa3bff may mean we actually have to fix some legacy cruft mind you, but we'll see
[06:21] <cjwatson> That'll be an unfun diff if so
[06:22] <wgrant> If they've actually been removed since, we should probably just add compat methods to our TestCase
[06:23] <wgrant> We have a LOT of callsites.
[06:23] <wgrant> Oh only about 800, not too bad.
[06:24] <wgrant> Oh they actually removed assertEquals and co, oh dear
[06:25] <wgrant> That's another 1500 or so
[06:26] <cjwatson> Been mildly annoying me for about six years, so I might just fix in bulk ...
[06:26] <cjwatson> I mean completely unreviewable, but
[06:27] <wgrant> Indeed, a fine idea.
[06:50] <cjwatson> No obvious DeprecationWarnings showing up yet, so it may just be general cleanup rather than required for this upgrade
[07:58] <cjwatson> wgrant: Do you have a non-prod Swift setup?  I recall that the last time I tried to set one up I ended up rabbit-holing on it for the better part of a day, and eventually giving up and using Canonistack.
[07:59] <wgrant> cjwatson: I do. I can give it a run through
[07:59] <wgrant> Mine's mitaka but the API isn't completely different so should be good enough
[07:59] <cjwatson> That'd be very welcome, thanks
[08:01] <cjwatson> Have we done similar staging adjustments before?  I guess we must have done when deploying this in the first place
[08:01] <wgrant> Not since then.
[08:02] <wgrant> I've been putting it off for years but I guess we should probably fix it up.
[08:21] <cjwatson> Oh, in fact unittest2 1.1.0 must be fine - I upgraded it in my twisted-16.5.0 branch.