I did not read Nancy's post. What I did read was Ed's description of what was not a job for Tx's. In fact, the scenario he outlined was definitely a job for Tx's, one where all updates to all tables have to occur, or none should occur. This is clearly a scenario for Tx's. He stated that buffering was the way to handle this.

>John,
>
>I have to defend Ed here. I had the same initial reaction and then I went back and reread Nancy's message to Ed. He was referring to the description of the problem with unbuffered tables and transactions that Nancy had described. He was correct, the scenario that Nancy described is not a candidate for a transaction.