Nadya,

If your table2 does not contain any deleted records or the join index is filtered for !DELETED(), you may want to try :

SCAN FOR INDEXSEEK(PropID,.f.,"Table2")

INDEXSEEK does not retrieve records from the table but searches trough the index itself. I can't think of any faster method.

Walter,


>Hi everybody,
>
>I have Table1 (child) with ~5mln.records and Table2(parent) with ~4mln.records
>
>I want to find and cound each record in Table1 key PropID, what doesn't have
>corresponding record in Table2.
>
>Two ways of doing that:

>scan Table1
>seek PropID && in Table2
>if not found()
>   *** This is an orphan.
>endif
>endscan

>
>or
>
>select PropID from Table1 where PropID not in (select PropID from Table2).
>
>Both tables, of course, have indexes on PropID.
>
>My question is: what is the fastest way of doing this?
>Currently we're using method 1 and I have not patience enough to perform test2.
>
>What do you think?
>
>Thanks in advance.