[Seaside] Database

Avi Bryant seaside@lists.squeakfoundation.org
Fri, 6 Dec 2002 13:35:24 -0800 (PST)


On Fri, 6 Dec 2002, Adrian Lienhard wrote:

> So, I really don't know... I think to tend to Postgres and try to use
> Tantalus. On the other hand it would be probably more comfortable to use a
> object database.

Although Tantalus is certainly immature, we are using it in production and
actively working on it.  Alain might have some insights on using it with
Postgres, that's not a configuration I've used (I've used it largely
with MySQL, and test it with SQLite over ODBC).  If you decide that an RDB
is the way to go, it's not a bad choice.  Be aware though that using an
O/R mapping layer will "infect" your entire object model - you still have
to be thinking about tables and relationships, even if you seem to be
dealing with objects and collections, and your design can't help
but to reflect that.

You shouldn't avoid using an OODB just because of full text searching - if
you can find Scott Crosby's text indexing code, for example, it should
work very well with something like Magma or Stephen's BDB stuff (you won't
have to load the entire DB into memory, just the index).

I can't speak to GLORP, but I'd like to hear from anyone that's used it in
production.

Avi