> I don't suppose there's any way that, as part of this work, the size of
> the cache could also be limited DURING a transaction? ...
That didn't look like part of Dieter's proposal to me <wink>.
It carries its own risks, and I'd rather not tie them together. For
example, _v_ attributes could start vanishing "in the middle of" a
transaction then, and although there was never a guarantee that wouldn't
happen, it's easy to guarantee that some code out there will break if it
started to happen.
For more information about ZODB, see the ZODB Wiki:
ZODB-Dev mailing list - ZODB-Dev@zope.org