IRC log of #zodb for Sunday, 2016-07-10

*** jensens has quit IRC01:27
*** jamadden has quit IRC08:09
*** agroszer has joined #zodb08:52
*** agroszer has quit IRC09:12
*** jensens has joined #zodb12:03
*** jamadden has joined #zodb13:25
*** J1m has joined #zodb14:45
*** aclark has joined #zodb15:06
*** J1m has quit IRC15:57
*** aclark has quit IRC16:56
*** aclark has joined #zodb17:53
*** aclark has quit IRC19:04
*** jensens has quit IRC19:05
*** jmuchemb has joined #zodb19:10
*** J1m has joined #zodb20:29
jmuchembHello Jim, the next step about the new commit protocol is to decide whether we drop support for the old one in ZODB521:32
jmuchembdo you mean prefer we discuss this on the ML ?21:32
jmuchemb(oops s/mean//)21:32
J1mI thought this was already done.21:33
J1mafaict, ZODB5 already requires the new protocol.21:33
jmuchembthere's indeed no test anymore for the old protocol, but the code still supports it21:34
jmuchembI just want confirmation that I can now cleanup and remove support for the old one21:34
J1mthen lets stop "supporting" things that aren't tested. :)21:35
jmuchemba last question about IMultiCommitStorage:21:36
jmuchembCurrently, it's unused. Personally, I always found it overkill, but I think your idea is to declare it on all storage implementation (FileStorage, etc.)21:37
J1mYes.  I like being explicit about contracts and it provides an opportunity to fail early.21:38
J1mI declare it for ZEO. I let the declaration slide on your PR. I probably shouldn't have.21:39
J1mI think we should require the interface when we stop supporting the old commit protocol.21:40
jmuchembwhat about ZODB.ConflictResolution.ResolvedSerial ? can I just delete the "ResolvedSerial = b'rs'" line ?21:41
J1mI guess :)21:41
jmuchembok I think I have all the information. I don't know when I'll submit something. I still have urgent tasks for a customer. Maybe 1 week.21:44
*** agroszer has joined #zodb21:45
J1mOK. Thanks a lot for all your work.21:45
jmuchembhave you seen ?21:45
jmuchembI think the random failures we sometimes have pypy are not pypy bugs. At least the one I had in one of my PR was really a bug in my code.21:47
jmuchembpypy seems more sensitive to race conditions21:48
J1mThat's weird.21:49
J1mThere's a pretty consistent failure in those builds.21:49
J1mOh wait, I was looking at the wrong thing.21:52
J1mI find travis a bit hard to navigate still.21:52
J1mGenerally, I ignore intermittent failures unless a) I see the same failure often and b) I feel like taking the time to run it dowm.21:53
J1mIn particular, I'll try running the test 99 times by itself, and if it fails, I dig futher.21:54
*** agroszer has quit IRC22:00
*** J1m has quit IRC22:38
*** jmuchemb has quit IRC22:49
*** J1m has joined #zodb23:12

Generated by 2.15.1 by Marius Gedminas - find it at!