IRC log of #zodb for Friday, 2014-05-30

*** fdrake has quit IRC01:25
*** cropleyb has joined #zodb01:27
*** cropleyb has quit IRC03:47
*** cropleyb has joined #zodb04:26
*** cropleyb has quit IRC06:50
*** cropleyb has joined #zodb08:02
*** agroszer has joined #zodb08:57
*** cropleyb has quit IRC09:32
*** cropleyb has joined #zodb11:24
*** cropleyb has quit IRC12:52
*** agroszer has quit IRC13:28
*** cropleyb has joined #zodb13:34
*** cropleyb has quit IRC15:09
*** namregz has joined #zodb15:37
namregzhello, is it normal to have a .lock, .tmp, and .index left behind after accessing a zodb ... such as with the tutorial at
namregzI added a db.close() and still the lock is left behind15:44
J1mnamregz, yes.16:27
J1mKeeping the lock around is important for it's correct functioning.16:28
namregzafter it's closed? after the program exits?16:33
*** fdrake has joined #zodb16:34
J1mnamregz, yes17:29
J1mThe goal of that lock is to guard against multiple processes accessing the same file.17:29
namregzI see, I thought when the process was done it should remove the lock ...17:30
namregztypical dotlock17:31
namregzthought wrong I guess, thank you17:31
J1mThe lock is using fcntl-style locking. If it isn't there, there's a race condition between 2 processes creating it at the same time.17:38
J1m(nfc what happens on windows :)17:38
J1mI suppose it could be supplemented with a only-create-if-new flag, at least on posix systems.17:39
namregzwindows ... I think that's what the CIA uses now instead of waterboarding17:52
*** namregz has quit IRC19:44
*** srichter has quit IRC20:47

Generated by 2.15.1 by Marius Gedminas - find it at!