Good that you did, Garrett BUT...

maybe the correct "solution" is to leave the old behaviour alone and invent a new function for the new behaviour.

After all, it will be virtually IMPOSSIBLE to find this potential 'bug' in existing code! And it really is TOO LATE to wait until a user has a problem and then isolate/fix it.

If they leave it as-is then I hazard that only TROUBLE can follow. Possibly even dissuading some from implementing VFP9!

cheers

>>In VFP 3,5,7,8
>>
>>VAL("3456-")=3456
>>
>>IN VFP 9
>>
>>VAL("3456-")=-3456
>
>Filed the behavior change. Thanks!