>>>>>I was using the VFP's RI code generator to cascade my updates. So when >>my parent record is deleted my child record will be too and when my >>parent is changed my child will reflect that change automatically. My >>one-to-many relationship is based on an expression PO+REFNUM. When I >>choose to generate my RI code I get a message saying:
>>>>>
>>>>> "Cascade updates based on expression-based index keys may not behave as >>expected."
>>>>>
>>>>>Does anyone know more about this? What are the possible pitfalls here? I >>know the best way to do this would be to have the tables related by a >>seperate ID field, but we can't add any fields to our table at this >>point.
>>>>
>>>>VFP RI code doesn't work w/ compound keys. you'll need to write your own >or use something like xcase...
>>>
>>>Ah, so "MAY not behave as expected" should read "WILL not behave as expected." That makes a big difference. But, shame on you, Dave, for suggesting I write my own RI code. Shame on you! ;).
>>
>>i'd give you mine but it's out of date. we're using xcase or stonefield >now (maybe both -- i'm not currently involved in that side of the job)
>
>Gee, you actually have job deliniation there? Must be nice. I suppose you have a class librarian and class designer too? :-P I've always had to wear many...er, all...the hats during development.

almost (we're not big enough for that). i got added to this project about 3 months after it was supposed to start and told to write up the specs. now that i'm 3 1/2 months behind......