>Raymond Kennington wrote:
>
>
>
>>(I am amazed that the language allows for something that appears to be very useful but is not functioning as described or is considered poor design.)
>>
>>
>Alexander V.Nevsky wrote:
>
> I don't know when and for what this feature was implemented, but I
>incline to treat it at least as incomplete.
>

Computed fields that reference other tables have been part of the
database since about day one. When it was pointed out to the designer
that they couldn't always be restored from a backup, he suggested that
the feature was useful and fixing backup would be a hassle. The
documentation group, who had pointed out the problem, suggested that he
wasn't going to document the feature unless it was more completely
handled. And there it sat. I believe there's mechanism in dfw now that
could handle the problem correctly, if it were taught to do so. The
restore problem is actually not in gbak, but in dfw. As for problems
with views etc., I think they should work. During compiliation, the
computed field is turned into something like a sub-select.

Regards,


Ann