>>I disagree. You only really know you need to code against it because you get an error. It's not like trying to access record 5 in an empty table. You'd know intuitively that would fail. I don't feel that _vfp.datatoclip with an empty table is necessarily the sort of thing you'd code against intuitively.
>>
>>I agree with Peter and Cetin that it's a bug in the documentation.
>
>But then I'd rather see the code fixed, not the documentation. The function could simply return empty string for zero records. Wouldn't that be nice?

But then you couldn't use TRY/CATCH to process easily. You'd still have to check for "no records" situation for that code to work (unless no records was OK).