Craig,

>>>Can you show me where Codd or Date explicitly stated that a record is a non-deleted tupple?
>>
>>I didn't say this.. I said, that a non-deleted is the equivalent of a tuple. Since a deleted tuple is a non existing tuple, a deleted record should be a non existing record.
>
>The RM does not include NOR exclude deleted records when describing a tuple. It's your defination that deleted records be excluded.

Yes based on the assumption that either CODD and DATE regard the existance of deleted records. Remember that with the introduction of the RM xBase did not exist yet, neither did the existance of delete records.

>But also, technically speaking, what Codd & Date described doesn't exist.

CODD describes a theory (the RM); the theory certainly exists. Date describes both the RM and (R)DBMSs features. Both the RM theory and the (R)DBMSs certainly exists.

>>The RM does implicetely describes that you can reuse a PK (So please don't come with the story that you should generated PKs, this is another topic). Why should we not be able to do this in VFP ???
>
>There is a solution...if you don't like the way it works in VFP, then don't use it. There are plenty of other solutions. MSDE is free and works the way you want the DB to work.


You draw the wrong conclusion, I certainly think there are things wrong in VFPs implementation of a RDBMS, but this is also the case with others (also MSDE). It's not a case of using or not using a certain (R)DBMS, but how to handle the problems. See my latest reply to Jim.

Walter,