> A concept like "user exits" might provide flexible data validation
or
> pre-/post-processing for DBGET/DBPUT/DBUPDATE intrinsics. But this
could
> also open a can of worms in terms of complexity and robustness...

"User exits" is what Date & Codd recommend.   If memory serves I think
Dr. Codd explicitly includes the concept in the later incarncarnation
of the relational model.    IMO this is the strategic solution -- talk
of adding new data types (domains) is not.

I don't see that adding this feature is particularly risky - any more
than *any* change to the code is a risk.

Some would be concerned that such enhancements would "slow down"
IMAGE.    I think that is a silly argument when the work must be done
some where, some how.    In that Image doesn't provide certain
features many HP 3000 applications don't either leaving us with poor
quality, even corrupted data.


"Lars Appel" <[log in to unmask]> wrote in message
news:99cqde01jn8@enews2.newsguy.com...
> Jim wrote...
>
> > ... my list of suggested data type fields was USA centric.
Guilty.
>
> Ok. Sentenced to read five off-topic postings *before* deleting
them.
>
> :-) ;-) :-)
>
> > ... I do think
> >that HP should look at providing intrinsics to make editing data
like
> dates as
> >consistent and easy as possible.  And in the case of dates HP as
responded.
>
> A concept like "user exits" might provide flexible data validation
or
> pre-/post-processing for DBGET/DBPUT/DBUPDATE intrinsics. But this
could
> also open a can of worms in terms of complexity and robustness...
>
> On the other hand, nobody would have to wait for HP to implement
such a
> feature. As far as I recall, there are already a couple of
people/products
> using an "intercept" approach with Image Intrinsic "look alikes" in
some
> user level NMXL or NMRL that call XL.PUB.SYS internally.
>
> Lars.
>