Craig/Jim

>>I'd classify the fact that SET TABLEVALIDATE TO's record count check being done AFTER the write (INSERT or APPEND) as critical because the purportation (like that word?) that it fixes the lost records problem is FALSE because of that!!!
>
>Whether this is a bug in the software or with the specs, I don't know. It seems like it would be easy to check this before the insert, but we don't know. There may be technical reasons that it works the way it does. In MS speak, this is "by design". However, I agree that it would be better to check this before the write. Is this a big enough issue to warrant an SP? I don't know.
>
I probably missed the relevant thread on this one, but I don't agree with Jim's interpretation of this one. If the check was made before the write, one knows that a new write will not succeed, but not whether the current write succeeded. If it fails after the attempt to write, and the error bubbles up, the transaction can be aborted and I know exactly where I'm at.


Geoff