"John P. Hartmann" wrote: > The design choice is whether to write the record to the driver before > or after it is written and consumed on the primary output.
Right, and after is certainly the right choice for EOFBACK. The problem with my COPY|SYNCH approach, though, is just that SYNCH isn't summarily stoppable, so it reads and writes the extra record on the primary even though its alternate output is already disconnected. I wonder if I'm overlooking a way to work around that. ¬R
