In a message dated 95-11-10 14:26:10 EST, [log in to unmask] (Chris Bartram) writes:
 
>I'm no odbc expert, as we're just starting to experiment with it ourselves
>(and also ran into the unique-key requirement problem in MS Access), but I
>wonder if a way around it wouldn't be to have HP automatically generate a
>pseudo-item for every detail dataset - say a numeric field which gets
>initialized to the Image record# of the record... Ignore updates to that
>field
>when records are being updated but use it as an index if possible. Seems
like
>that field could then be used as (or made part of) the unique key for MS
>Access?
>                Just a thought...
>
>
 
Funny you should mention that.  This is the exact solution I presented to HP
at the last IRPOF.  I spent quite a bit of time discussing it with them since
then.  At the last 2 C/S UG meetings, HP said they were not going to do that.
 They said that using a view which as in it a call to TID with a new function
to be released, should handle the problem.  We shall see.
 
However, you still do not have to use a unique key when you attach an IMAGE
table to a MS Access, just lie and tell it that the column you are telling it
to use is unique.  MS Access won't trust you anyways, because when it does a
delete or an update, it verifies all the columns in your tuple.
 
Kind regards,
 
Denys. . .