> Thanks Roman. You'd think they'd understand the performance increase
> that batching provides. In most databases the difference between
> batched and non-batched updates is pretty drammatic as the number of
> round trips to the server is greatly decreased.

Exactly this is the problem. It is clear that our goal is to eliminate
excess of the roundtrips - there is an agreement on this, but so far I have
seen no agreement how to do this.

The issue has to do with the wire protocol and with the public API. Jim has
a proposal for a new API and he had suggested to introduce a new wire
protocol, but so far it did not went behind the suggestion. It would be
quite easy to extend current API with a new calls that would take array of
parameters or string statements and would return the update count, but that
does not nicely fit the current Firebird. So, I expect rather hot discussion
in Prague, since the issue is not easy. And if new API and the protocol is
introduced, it will take some time until it gets to production.

So, the only possibility for next year or two I see is the type 3 driver,
like the InterClient 1.6 and 2.01 were. Though I'm not sure if that is right
thing to do in the long run. Maybe it is...

Roman