>Since you already have it isolated to a small piece of code, perhaps there is a recurringly faulting line. You should investigate the possibilities of getting a trace of the last run code. Either move the offending process to a standalone exe and run this under vfp9 with the coverage profiler, get your customer to run your code in a vfp8 IDE with coverage set on for the faulting process or build your own small protocoller sprinkling some code...
>
>If it *never* shows with profiling enabled just say you fixed a timing issue <bg>. Happend to me once in a FPW exe, when I ran into a kernel error. As long as I echoed to a textfile, everything ran ok - and since it was not a time critical part I always overwrote a log of a few hundred lines.
>
>HTH


That's pretty much what I'm down to at this point.

Dan