>Also, when it does crash, it 'forgets' whatever changes were made to the main table since the program was started. IOW- if someone goes in and changes some data and the program crashes, it reverts. It also leaves the date/time of the dbf file at whatever it previously was. The error also doesn't happen (at all?) when I'm tracing the code. I can't seem to make it happen. However, in both cases the disk activity jumps up as does the processor usage...
The behavior is normal for buffered access; if VFP fails before a TABLEUPDATE() occurs, the data should 'revert', because the only changes to the data were made in the buffer, but they were never written back to the main table.
A rapid increase in disk activity might be the swap file growing; this could be caused by lots of buffers being held, awaiting TABLEUPDATE() or TABLREVERT() to be applied. What buffer mode are you using with the tables/cursors in your DE?
>
>Cole