[ZODB-Dev] Storage API corner cases
Jeremy Hylton
jeremy at zope.com
Fri May 30 18:53:19 EDT 2003
On Fri, 2003-05-30 at 13:33, Barry Warsaw wrote:
> Uh, BTW. checkOversizeNote() are the last two failures for BDBStorage
> in ZODB3, but those tests are bogus. It's saying "try to store 128K of
> zeros as the transaction note and if that /doesn't/ fail with a
> POSError, the test fails".
>
> Whaahh? First off, BerkeleyDB has no problem storaging that huge of a
> transaction note, and second, FileStorage should be fixed so that it's
> legal. This test surely shouldn't be doing what it's doing so I'm going
> to disable it.
FileStorage uses two bytes to store the length of the transaction
metadata. At this point, it's impossible to fix for ZODB3. It's
possible to fix for ZODB4, but I wouldn't put a high priority on it.
I think it's fair for individual storages to provide different quality
of server for large metadata. I don't think it makes sense to have a
generic test for this though.
Jeremy
More information about the ZODB-Dev
mailing list