>>Also, to be devils advocate, nobody's mentioned the use of views and p-views (my favorite approach) instead of buffering the actual tables.
>
>Actually, I have used views and p-views for editing, and I still set buffering on the view to optimistic table buffering (change from default optimistic row buffering). For the same reason I have been saying, (and Jim has, too). I don't want to have to debug an accidental (or as Jim said, "magic") update of data, because some deeply hidden method moved my record pointer.
>
>Cheers,

Agree completely. I do the exact same thing. Just trying to add another facet to the thread.

Cheers