""Woody"" <woody.tmw@...> wrote in message
news:001501c1dd7b$c5c7b500$166222d1@popstoy...

>
> I know that you frequent the Borland groups, so you must have seen at

least

> one of these in the past few weeks, no?

No. There are several posts and I only click on the few ones whose title
catch my attention.

> People such as Bill Todd, etc. have
> said these things.

Bill and Craig are functional experts, but they don't need to get any clue
about the internals (because you don't need to know the source code to use
the engine in most cases), so I only believe them if they are repeating
something straight from the IB team.

> They haven't specifically said the access rights might
> change but system tables can be changed to suit the engine.

Please explain how
select gen_id(...) from rdb$database
is affected by the structure of that table (or any table). It doesn't touch
any field. Whether you add or delete fields, it's not relevant. Oh, I can
only speculate about one idea: the IB team is going to put more records in
that table, so the call to gen_id() would be done for each record, resulting
in odd behavior when incrementing the generator (apart from losing some
numbers, nothing worse will happen).

> select extract (month from current_date) from rdb$nothing
>
> The rdb$nothing (or whatever you want to call it) isn't actually a

physical

> table, just a keyword used to balance the statement.

The problem is only in dynamic statements. Inside a procedure,
z = gen_id(g, N);
d = extract(day from current_date);
are allowed.

Your idea sounds reasonable, although rdb$nothing will have to fake a single
record or nothing would be returned.
:-)

I'm sorry you didn't like my suggestion for retrieving table names. I was
only trying to make your life worse.

C.
--
Claudio Valderrama C. - http://www.cvalde.com - http://www.firebirdSql.org
Independent developer
Owner of the Interbase® WebRing