>>>>>Tommy, were you getting the error on use (add/edit)? Or on the creation of the
>>>>>initial INDEX? I thought you were getting it occasionally on the INDEX ON
>>>>>command itself.
>>>>
>>>>The error would always occur on the line of code that would assign the value
>>>>of 5 to 'status' of the new record in ss_io.
>>>
>>>
>>>So the SET value of CENTURY at that point differed from the value when the INDEX was created?
>>
>>
>>No, the same. SET CENTURY ON and SET STRICTDATE TO 0, in the old and new code
>
>
>I guess I'm not understanding how this is happening then. I could see if the CENTURY was ON and it created an index key along the lines of "MM/DD/YYYY" and then you tried to create a key which was "MM/DD/YY" as possibly generating an error. And I could see if it created a key with CENTURY OFF using "MM/DD/YY" and then you tried to create a key later with it ON which was "MM/DD/YYYY" and definitely generating an error.
>
>But how does it generate an error under any condition when the key is the same length? Seems like a bug in VFP9SP1 and not anything wrong with your logic or usage.

I may have made an error in gathering the data here for analysis.

Don't sweat it. Switching to DTOS fixed the problem and it works ok now, so we'll chalk it up to some bug in VFP9SP1.

Thanks