Thanks, Nadya. Now, as I mentioned in other parts of this thread, the actual SQL - SELECT command is more complicated.

About your FAQ, I think you should mention the alternative of a subquery. The code is much shorter and - once you understand subqueries - simpler. I don't know about speed; but sometimes you want to get results quickly, meaning, without writing a lot of code!

In your example, you can show the results in a browse window with a single command:

select * from Table2 where KeyId not in (select KeyId from Table1)

What I am dreaming about is to integrate subqueries with the SCAN (or something similar). I think this can be done with a UDF, but I have to think about it a little more.

Regards, Hilmar.

>Hilmar,
>
>Check my FAQ "Best way to find orhpans". You can use either:
>
>scan for indexseek(KeyField,.f.,'Table2','KeyIndex')
>endscan
>
>of using KeyMatch, which would be faster.
>
>Another way would be to set relation and
>scan for found()
>endscan
>
>
>>I want to process, in a SCAN loop, all records that satisfy a given condition. The condition is most easily expressed as a SELECT - SQL - but what is the best way to combine the two?
>>
>>

>>Pseudocode:
>>
>>SCAN where KeyField in (SELECT KeyField from AnotherTable)
>>   ...
>>ENDSCAN
>>

>>
>>The processing itself is too complicated to place into a single INSERT - SQL command.
>>
>>Basic criteria: 1) code should not be too messy, 2) it should be reasonably fast.
>>
>>TIA, Hilmar.