| *** jmuchemb has quit IRC | 00:11 | |
| *** J1m has quit IRC | 00:46 | |
| *** povbot has joined #zodb | 01:31 | |
| *** J1m has joined #zodb | 01:44 | |
| *** J1m has quit IRC | 02:07 | |
| *** jamadden has quit IRC | 08:09 | |
| *** jensens has joined #zodb | 10:17 | |
| *** jamadden has joined #zodb | 13:25 | |
| *** J1m has joined #zodb | 13:53 | |
| *** jamadden has left #zodb | 16:17 | |
| *** J1m has quit IRC | 16:36 | |
| *** J1m has joined #zodb | 17:03 | |
| *** jmuchemb has joined #zodb | 18:35 | |
| *** J1m has quit IRC | 19:33 | |
| *** jmuchemb has quit IRC | 20:11 | |
| *** J1m has joined #zodb | 20:16 | |
| *** jmuchemb has joined #zodb | 20:35 | |
| jmuchemb | Hi Jim, I just saw your question about DemoStorage.store and I don't understand | 20:37 |
|---|---|---|
| jmuchemb | I think I missed a few changes in the interfaces | 20:39 |
| jmuchemb | In fact, my intention was to leave the choice between reporting at store or tpc_vote | 20:40 |
| jmuchemb | the question is mainly for simple storages like FileStorage & DemoStorage, for which keeping the list conflicts in some collection may be a bit annoying | 20:43 |
| jmuchemb | I wonder if there's a benefit in forcing storage to report conflicts only at vote ? Is there ? | 20:45 |
| J1m | I realized after commenting that you had different assumptions about the API. | 20:46 |
| J1m | I guess I don't object to store optionally returning booleans. | 20:46 |
| J1m | as long as it's spelled out in the interface. | 20:47 |
| J1m | I'm losing track of what the contract is. | 20:47 |
| J1m | So, you can disregard my comment, but, we need to update the interface. | 20:48 |
| J1m | I guess storeBlob should always return None. | 20:48 |
| J1m | Because we don't have conflict resolution for blobs. | 20:49 |
| jmuchemb | but, for performance and code simplify, I may like you proposal of a stricter interface :) | 20:50 |
| jmuchemb | (checking savepoints ...) | 20:51 |
| J1m | I don't feel strongly about returning booleans from store, one way or the other. I just want the contract to be written down. | 20:52 |
| jmuchemb | sure | 20:53 |
| J1m | I also want there to be tests for storages that return all of the resolved oids in vote. | 20:54 |
| jmuchemb | there's already testDemoStorage which tests the 2 cases | 20:54 |
| jmuchemb | by forcing conflicts to be reported by tpc_vote, I see some code simplication in Connection so I'm going to prefer that | 20:56 |
| J1m | Except that it didn't exercise blobs or undo, which bit me when porting ZEO. | 20:56 |
| jmuchemb | yes, sorry for that | 20:56 |
| J1m | That's OK. An easy way to do this would be with an adapter/wrapper of file-storage that holds the resolved oids till vote. | 20:57 |
| J1m | similar to the current adapter. | 20:57 |
| J1m | or have file storage hold the resolved oids | 20:58 |
| jmuchemb | I prefer the 2nd option | 20:59 |
| jmuchemb | that can be done partly in BaseStorage so that it's done like this for FileStorage and DemoStorae | 20:59 |
| J1m | k | 21:00 |
| jmuchemb | PR 83, so that ZODB4 does not accept store to return True | 21:17 |
| jmuchemb | coming back in ~20 minutes | 21:19 |
| *** jmuchemb has quit IRC | 21:19 | |
| *** jmuchemb has joined #zodb | 21:44 | |
| *** J1m has quit IRC | 22:06 | |
| *** jmuchemb has quit IRC | 22:08 | |
| *** J1m has joined #zodb | 22:16 | |
| *** J1m has quit IRC | 22:31 | |
| *** jmuchemb has joined #zodb | 23:42 | |
Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!