>>Specifically, it fails to mention that it affects all subsequent revisions to any tables until either a END TRANSACTION or ROLLBACK is issued. In fact reading it would lead one to conclude that it *does* only affect one table at a time.
>
>ARGH! Thats exactly how I interpreted it. I built my business class so that if it contained child objects, upon save, each child object would start it's own BEGIN TRANSACTION. I went to quite a bit of trouble to ensure things worked this way.
>
>If I understand you correctly, I can just handle it all with the top level object, and (by using txnlevel() maybe), have any child objects handled by the main transaction.
>
>Would defintately make things tidier....

Well, Russ, at least you took the "safe" way. And you may well have a more flexible and robust mechanism by having done so!

Nevertheless, it's a shame that this kind of ambiguity remains so prevalent throughout the VFP documentation. I keep adding small things to this topic in hope of proving a need for a thorough re-vamp BUT I have to admit that I feel like a tiny whisper at a band paratice.

Cheers