> > hmm, connection pooling is a production technique for scaling up your
> > access. I'm afraid this does not sit well for database design
> > development
> > work.. You need to develop your database in a development environment.
> > You
> > can sync your metadata changes in one go when you're ready after you
> > disconnect the pool and establish exclusive access to the production db
> > (after you back it up).
> > Alan
>
> This doesn't work well in the concept of an auto-deployed and
> multi-database web application. Typically, you start the web
> application, which starts up the connection pool. Then when the
> administrator that's installing it connects first to it, he doesn't see
> the application but goes through a wizard to setup preferences and
> select which database to use. Firebird is an alternative we're trying
> to add, but MySQL, PostgreSQL, Oracle and HypersonicSQL are already
> supported and working well. When the database is selected, the initial
> structure is installed. Of course this structure has been design and
> developed elsewhere beforehand. It's however very logical to create a
> proven structure in one go inside a running application. It's weird
> that Firebird prohibits this solely for foreign keys.
>
> --
> Geert Bevin Uwyn bvba

All you would need to do this startup process is use a separate exclusive
connection to create your database. Even if others work thru the pool I
wouldn't do it. Maybe with MySQL it wouldn't matter at all they don't have
foreign keys anyway??? but to create the db I would be running thru a non
production outlet.
Alan