Here's another GUESS. . .

*IF* OPS is an alias already opened, then perhaps SQL-INSERT is using the (already) opened work area instead of opening it own.

Possibly stating the file name in the INSERT (or UPDATE) command would make a difference??

I use the opportunity to point out yet again that the VFP docs ought to tell us HOW these things are handled. Making up our own "rules" based on observation is just too risky!

Cheers,

Jim N

>>It was a simple
>>UPDATE Ops ;
>> SET Extno = SPACE(4), ;
>> LoggedOn = .F. ;
>> WHERE Extno = app.extno
>>I don't really care if it's the currently selected table or not. This moves the record pointer for "Ops" to EOF(). My view is that it should not move at all. Otherwise I'd simply use the old SCAN technique
>>* save recno() here
>>SCAN FOR Extno = app.extno
>> REPLACE Extno WITH SPACE(4), LoggedOn WITH .F.
>>ENDSCAN
>>*go to old recno here
>
>Without any other knowledge of how VFP handles this, I would expect the pointer to move because this is like saying replace extno with space(4),... for extno = app.extno. The only way this would not happen would be if the Ops table were opened again in another area for updating, leaving the original table alone.