Chris Withers wrote at 2006-10-4 15:06 +0100:
>The interesting thing is that it looks like the transactions where the
>time appears to go backwards are duplicates of earlier transactions:
>position in file tid time from tid
>31025376233 0x03689abb582f1311 2006-10-03 04:43:20.668098
>31025376508 0x03689abdbbe5f000 2006-10-03 04:45:44.038639
>31025376783 0x03689abddbe6be55 2006-10-03 04:45:51.539377
>...lots of transactions...
>31025646913 0x03689abb582f1311 2006-10-03 04:43:20.668098
>31025647188 0x03689abdbbe5f000 2006-10-03 04:45:44.038639
>31025647463 0x03689abddbe6be55 2006-10-03 04:45:51.539377
>Would this seem to be an accurate reading of the attached log?
It looks as if you had given the same incremental file twice
>1. Could repozo have a bug that resulted in this?
Maybe, especially when the same file is twice integrated
>2. If repozo has no bug, should it have checking that makes sure it
> doesn't build insane .fs files, or is fstest the way to go for that?
As "fstest" found this problem, it was not too bad.
>3. If repozo is not to blame, what could be?
One possibility would be a bad call.
For more information about ZODB, see the ZODB Wiki:
ZODB-Dev mailing list - ZODB-Dev@zope.org