I checked out this thread
http://www.llblgen.com/tinyforum/Messages.aspx?ThreadID=167
but that was before the latest release, so I'm not sure if it applies to me.
I have a SQL Server database and and an Access database (hey Frans, thanks for doing the Access driver!). The latter is a subset of the former, basically people are "checking out" some data, so the tables will be similar but not exactly the same. Some may be denormalized (because they are read-only). Certainly PKs in SQL Server with IDENTITY property will just be integers in Access.
At first I assumed that I would have to go with adapter scenario, because the manual suggests using adapter if you have multiple dbs. But why? Can I not just create two DALs, one for each database? What does the adapter scenario really buy me here? I like the self-servicing scenario, it seems so much more OOP to me (objects know how to persist themselves). I don't want my choice of ORM architecture to be driven solely by the fact I have a second database to deal with.