David Binger  wrote:
> On Dec 15, 2005, at 2:59 PM, Jesus Cea wrote:
>> The obvious path is using reference counting with objects.
>
> That's not sufficient, since you can have lost cycles.

If he is sure there are no reference cycles then I bet it's hard to
come up with anything better than reference counting.  OTOH, if you
are sure there are no reference cycles then I would question if you
really need to be using an object database.  Probably some other
kind of specialized storage would work better.

> The storage server serializes operations by handling one at a
> time, in a single loop in a single thread.

If there are lots of clients then the simple single-threadedness of
the storage server will probably start to hurt.  That's another
design tradeoff we made.  A more scalable storage server would be
multi-threaded or would be more careful to not let clients cause it
to block.  This is one area where ZODB is definitely better.

  Neil