I don't think it would be fully optimized, and the little test I just ran (using sys(3054)) seems to confirm that. However, as Hilmar has said, it may not be a problem.

Is this data stored in VFP tables? Why are there nulls, and can they be permanently eliminated?

>
>Right, the REPLACE statement should be changed as well. Why do you think, it would not be Rushmore optimizable? Do you remember, NVL in index expression does automatically turn expression to be non-optimizable?
>>