At 02:27 AM 6/10/00 +1000, Helen Borrie wrote:

>
>
>I understand that bit...
>>or avoid trying to create objects with conflicting names.
>
>... I don't understand that bit.
>

One could reasonably assume that one could check RDB$RELATIONS
to see if a table name was in use. One would be wrong. Stored
procedures come out of the same name space (so they can be used
unambiguously in a table list) but are represented in the official
list of tables. Ah, you say, also check RDB$PROCEDURES! Well,
duh.

>
>According to a poster in a recent thread, he always uses suspend in FSD
>constructs in exec stored procedures because he believes it's a way to
>guarantee he gets a hit for every row processed. I can't see the point in
>it, but apparently the compiler doesn't mind. Not a significant point,
>just FYI.
>

When something is obscure, neither the user not the implementor knows
exactly what should be done.

>>
>>Personally, I think the right solution is to settle on a single
>>well-defined language with type-specific interface objects.
>
>So do I. Could you do that by June 30?
>

Morning or afternoon?

Jim Starkey