On Jan 24, 2008, at 3:12 AM, Thomas Lotze wrote:

Jim Fulton wrote:

IMO, something that packed incrementally, with disk being freed along the way, would be a big improvement. This isn't possible with FileStorage.

Just an idea, without having followed FileStorage's history: Has spreading
the file storage across multiple files been considered?

Sure, but then retaining consistency across files would become a greater challenge. It is certainly possible, that's what the Berkeley DB does.

That would be an entirely different storage implementation.

FileStorage didn't put everything into one big file which has to be
rewritten from scratch during packing. Since transactions are simply
appended to that big file it should be possible to store each transaction
in its own small file. Incremental packing might then be possible by
conflating the oldest n files into one, for some sufficiently small n.

Lots of things are possible in alternative storage implementations. That wouldn't be FileStorage.


Jim Fulton
Zope Corporation

For more information about ZODB, see the ZODB Wiki:

ZODB-Dev mailing list  -  ZODB-Dev@zope.org

Reply via email to