| *** J1m has quit IRC | 00:04 | |
| *** jensens has joined #zodb | 00:12 | |
| *** jamadden has quit IRC | 00:45 | |
| *** jensens has quit IRC | 13:05 | |
| *** jensens has joined #zodb | 13:50 | |
| *** jamadden has joined #zodb | 14:57 | |
| *** J1m has joined #zodb | 15:02 | |
| *** aclark has joined #zodb | 18:36 | |
| *** aclark has quit IRC | 18:38 | |
| *** aclark has joined #zodb | 18:46 | |
| *** J1m has quit IRC | 19:02 | |
| *** J1m has joined #zodb | 19:03 | |
| jamadden | J1m: The RelStorage tests consistently fail in checkResolve with a ConflictError under ZODB 5.0a1. They pass with 4.4.0. Is that something that interests you? | 20:15 |
|---|---|---|
| J1m | um | 20:15 |
| J1m | It depends on whether they fail because of a ZODB bug or a RelStorage bug. :) | 20:18 |
| J1m | Why do you think they're failing? | 20:18 |
| jamadden | Same RS code works with 4.4 so I'm hoping it's a ZODB bug :) | 20:19 |
| jamadden | I haven't actually looked into it yet, I wanted to find out if you thought they *should* be passing or if this might still be an area under active work | 20:19 |
| J1m | Not active by me. It's my understanding that ZODB5 should work with RelStorage, since it's essentiallt doing things the relstorage way. | 20:21 |
| J1m | But there were some small changes in how things are called. | 20:21 |
| J1m | I suspect this is related to poll intervals, but idk. | 20:21 |
| jamadden | poll intervals have been removed :) | 20:22 |
| jamadden | what's happening is that conflict resolution is raising a POSKeyError on `oldData = self.loadSerial(oid, oldSerial)`. I'll try to see why | 20:23 |
| J1m | Connection used to call sync(False) on close. It doesn't anymore. | 20:27 |
| J1m | idk if that's related. | 20:27 |
| J1m | I could certainly see how it would interact with poll_interval. | 20:27 |
| jamadden | this test isn't using a connection, it's working at the raw storage level | 20:27 |
| J1m | Hm, then I have no idea | 20:28 |
| J1m | and 5.0.0a1 doesn't have Juliens changes. | 20:29 |
| jamadden | Ok, it's related to the "breaking" change in tests.ConflictResolution in 4.4.0. The two branches aren't in sync on that test | 20:36 |
| J1m | what 2 branches? | 20:41 |
| *** aclark has quit IRC | 20:43 | |
| jamadden | sorry, I really meant "release", not branch. Assuming branch 4->4.4.0 and master->5.0, they're in sync now. But 5.0a1 predates 4.4.0 and the test changes. | 20:43 |
| J1m | Yup. | 20:44 |
| J1m | Want me to make a 5.0.0a2 release? | 20:44 |
| jamadden | Not especially on my account. I haven't been testing the 5.0 releases. I only noticed the problem because one of my virtualenvs *somehow* decided to upgrade to the pre-release (I didn't think they were supposed to do that anymore). | 20:46 |
| jamadden | Sorry for the noise | 20:46 |
| J1m | k, np | 20:46 |
| *** J1m has quit IRC | 20:47 | |
| *** jensens has quit IRC | 21:11 | |
| *** J1m has joined #zodb | 21:25 | |
| J1m | I went ahead and released 5.0.0a2 | 21:31 |
| jamadden | Thanks. I can confirm that the RS tests pass on 5.0a2 | 21:37 |
| J1m | and released 4.4.1 and 5.0.0a3 with the new interface | 22:27 |
| *** J1m has quit IRC | 22:49 | |
| *** J1m has joined #zodb | 22:49 | |
| J1m | Implementing the new commit protocol in ZEO leads to lots of test failures. :( | 23:24 |
Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!