Helen Borrie wrote:

>
> 1. Even on an Internet connection?

Well, of course nobody can guarantee an immediate response over the
Internet :-) But it's still a matter of a single round-trip. Obviously,
some timeout value would be required to avoid long/infinite lock-ups
caused by slow/problematic networks.

> 2. What of an application that's designed to do event callbacks constantly (such as DML caching in IB Objects)? How could that work on Classic?

I don't see any problems here. As soon as the secondary connection is
established, it will be used for events delivery until the client
application detaches from the database. So it's only the first call for
events that could be affected from the performance POV (if someone else
is doing the same at the same time). But the same applies to SuperServer
as well.

> 3. Is the event table inexorably anchored to one port or can it be made so that the auxiliary port # can be tied to the PID?

The event table has nothing to do with ports, they're layered
separately. I seem to get your line of thinking, but I'm afraid this
doesn't fit the architecture.


Dmitry