Thanks Brian, I'll give that a try.
Metakit has exactly the right semantics for what I'm trying to do, but
it seems that I'm pushing it in directions it doesn't want to go. I've
given some thought to writing a database (probably in Python, probably
layered on top of Berkeley DB) with the same semantics (near as I can
manage) but different implementation/performance tradeoffs. Try to
trade speed in those areas where Metakit works well for better
scalability in areas where Metakit doesn't work well.
It can be done, I think, but it's clearly non-trivial.
Andrew
_____________________________________________
Metakit mailing list - [email protected]
http://www.equi4.com/mailman/listinfo/metakit
