>T1=SECONDS()
>? "BAD: this uses TFF1 and is optimized"
>SELECT count(FF) FROM ALIAS() INTO ARRAY AX
>? SECONDS()-T1

Does it really make use of TFF1? optimizing Count() can be done if rushmore sees, that the count() is equal to reccount(). Due To SET DELETED OFF that is valid, as there is no filter in effect. So I think this time rushmore in fact uses the reccount() stored in the table header to optimize the query, not an index.

Bye, Olaf.