>>>>>>>Hello,
>>>>>>>
>>>>>>>I worked on a table which has a primary index called "Key". This table is included in the DataEnvironment of one of my form, with Private DataSession. Problem occur when I tried to add (by APPENDing) a new record with a value which had been deleted previously. The message I got is "Uniqueness of Index KEY is violated." By PACKing the table after DELETEing a record would solve the problem. Is there any other way (other than RECALLing) which I could use back the same KEY value without having to pack the table? 'Cos I don't want to open the table as EXCLUSIVE.
>>>>>>>
>>>>>>>Thanks,
>>>>>>>juan
>>>>>>
>>>>>>You touched the reason why primary key should be surrogate, e.g. sequential number, which will never be repeated.
>>>>>You can define your primary index to "fieldname for not deleted()". There's been a bit of discussion on this subject, but I think that it is a valid alternative.
>>>>
>>>>This is not healthy alternative, primarily because index tags with FOR filter cannot be used for Ruhshmore purposes.
>>>True enough. I also have the primary field indexed without a fliter clause.
>>
>>Yes, but it will not be 'primary' key anymore and you cannot use it for RI triggers or for View key fields etc.
>You are correct, but I still have the primary index (with the for not deleted()) that I can use for the RI and Views. I have the added advantage of using the regular key field index to see all of the changes made a "record".
>
>A little off the subject...Be aware that I had to correct the RI code for cascading deletes ( a bug verified by MS) when using SQL delete/insert updates

What kind of bug? AFAIK, RI has problems dealing with compound keys, but surrogate keys should work Ok.