Alan,

>...Are there any advantages of using a CA over a local view of native data? Will the fact that the datasources are free-tables make any difference?

I'm about to dive into a similar project and will know more about free-table "issues" very soon. Meanwhile, one of the big advantages of CA is that you can set up your own top-level subclass of CursorAdapter class, put enhanced functionality there (such as indexing cursors based on an array property that you add (or collection) to hold indexing info, or other specific enhancements related to making it work better with free tables.)

Then, you can subclass from there for each CA you need and inherit all the good stuff from above.

To me, the subclassability of CAs (and XMLAdapters) is one of the greatest advantages.

You also may find that if you eventually need to convert to SQL Server or another ODBC backend, it may be a bit easier to just tweak the top-level CA instead of having to create remote views or SQL Passthrough commands for every table.