IRC log of #zodb for Friday, 2016-06-10

*** J1m has quit IRC00:38
*** J1m has joined #zodb00:49
*** J1m has quit IRC01:02
*** povbot has joined #zodb06:50
*** jamadden has quit IRC08:09
*** jensens has joined #zodb10:20
*** jamadden has joined #zodb13:25
*** J1m has joined #zodb15:20
J1mWe should probably revert my RelStorage changes.16:08
jamaddenbecause you're going in a different direction in ZODB itself?16:14
jamaddenbased on those comments in ZODB#5616:15
J1mMore based on our dicussion about IMVCC.16:15
jamaddenwhich you just referenced in ZODB#56, so it's all very meta16:15
J1m:)16:16
J1mYes16:16
J1mBut I also just discussed other things, like making transaction boundaries more explicit.16:16
jamaddentrue. which I think is a good thing16:16
J1m(I'm drafting a message to python-transaction about that.16:16
J1mor maybe not...16:30
*** J1m_ has joined #zodb17:29
*** J1m_ has left #zodb17:29
*** aclark has joined #zodb17:45
jamaddenIs it OK to update the ZODB dependency in RelStorage for 1.7.0? Currently it's at ZODB3 >= 3.7.0 (!), but tests are broken when 4.3.0 is used, and PyPy support needs at least 4.2.0. These can be worked around in various ways, but it's nicer to update the dep18:02
J1mNot sure what you're proposing, but sure. :)18:03
jamaddenproposing to update RelStorage 1.7.0 to depend on ZODB >= 4.3.0 (instead of ZODB3>3.7.0)18:04
J1m+118:04
jamaddencool. on it.18:05
J1mBut what is 1.7.0 for?18:08
J1mMy opinion is that we should back out the changes I made.18:08
J1mUnfortunately, I don't think I can do the IMVCCStorage everywhere work soon.18:09
J1mI think it's going to be rather involved.18:09
jamaddenI'm fine with that. So that would just leave 1.7.0/master as the place to land the updates that were too much for 1.6.0, like PyPy support, better exceptions, various other extent PRs18:09
J1mwhich makes me think that RelStorage should grow an upper bound to prevent it being used with ZODB5.18:10
jamaddenI might have some time to contribute to helping with IMVCCStorage if you can lay out a plan, but I suspect my time may better be spent on something like bringing Python 3 support to RelStorage18:10
jamaddengood point18:10
J1mThe tricky part of the adapter will be getting the invalidartions right.18:11
J1mEspecially the plumbing.18:11
J1mMaybe it won't be too bad, but I really want to finish a bunch of other stuff first.18:12
jamaddenany particular way you want to handle reverting the change to relstorage?18:12
J1m+1 on Python 3 support for RelStorage, to say nothing of CI.18:12
J1mI would use git revert commit undoing transactions. Do you know of an easier way?18:13
J1mI can do this if you want.18:14
jamaddenI do not know an easier way. I've never actually "reverted" something in git that had commits on top of it (other than manually), even if they don't conflict, so that would be new to me. I'd be happy to take a crack at it and learn though18:15
J1mThere's a relatively easy way to do it in a local repo. but now that it's in github that won't work.18:16
J1mI'll take care of it using git revert.18:16
J1mI was hoping for some automatiin in github, but I dn't see anything in the ui. Og well.18:17
jamaddenThat's what I meant, I've done it locally but not after a push.18:17
jamaddenIf you come up with a nice command to to it, I'd appreciate seeing it18:17
* J1m seens a revert button on the pr...18:18
J1mha!18:18
jamaddenha, no way!18:18
J1mway!18:19
J1mhttps://github.com/zodb/relstorage/pull/3418:19
J1mOf course, one of my changes was needed ti get tests to pass. :)18:20
J1ms/ti/to18:20
jamaddenYes, the 'doctest' thing for one. If you want to go ahead and merge it, I'm working on a PR anyway to fix the remaining tests (that were still failing) and I can roll re-fixing doctest, etc, in to that18:22
J1mk18:22
J1mdone18:23
*** J1m has quit IRC19:15
*** aclark has quit IRC21:20
*** J1m has joined #zodb21:25
*** jamadden is now known as jamadden_away21:25
*** jamadden_away is now known as jamadden21:40

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!