> Embedded SQL Guide Chapter 3 "Working with Databases"

Embedded SQL is talking about writing applications using gpre and
C/C++.

> so I assumed Interbase did support cross-database joins
> too. I guess I was wrong.

Interbase does not provide direct support, it only provides indirect
support via two-phase transaction commit facilities. It is upto the
client application to implement the capabilities using the transaction
facilities. gpre supports the Embedded SQL syntax described in the
manual you mentioned, allowing you to specify SQL which can be parsed
through gpre and then linked to the application.

(I think thats right, I've never actually done it myself.)

> I'm possibly going to use a table from a third-party database
> and some tables from my own database. Ability to run select
> statements joining the tables and some kind of referential
> integrity would be nice. What's the best strategy?

Obviously you will never get server managed referential integrity
while using different database types. It must always be upto the
client to handle it. Since you are using different database types
then any "joining" you do at the client (such as using master/detail
facilities like I described) would need to use TDataset based
components (TIBOQuery etc). You will have to implement cascaded
updates and deletes and other cross-database referential constraints
in your client code.

> I could use some kind of replication to be able to work
> with my database only but I'm not sure if this is the best
> way. Any ideas greatly appreciated.

Goto: http://www.interbase2000.org/ib_ReallyUseful.htm

And find the link for "Some Solutions to Old Problems" by Dalton
Calford. His article discusses solutions for implementing replication
using Interbase.

HTH

Geoff Worboys
Telesis Computing