>>>>>Your explanation doesn't make sense to me, I suggest you spend some more time and fix whatever is wrong in your code.
>>>>
>>>>Nick Mason reminded us that insert won't work when you're at eof() in the current alias. That's explanation enough to me.
>>>
>>>That's incorrect information from Nick, SQL commands behaves differently from xBase commands.
>>
>>I've already done my head scratching last night, thanks... had a case where a "blank fields for {condition} in {alias}" invariably blanked only the first two records out of five. And a different alias would be selected. Can't find any active timers. Eventually got it to work by rewriting it into a scan for - blank - endscan loop.
>
>Any indexes on the table? If the replace "moves" a record to the end of the table in the current order, the BLANK will stop since it reaches the end, (IIRC)

Nope, checked that. The sorting expression is kind of complicated, and since this is a SPT cursor, it was much easier to order it properly on SQL side. So it had no indexes at all.