> -----Original Message-----
> From: Ann Harrison [mailto:harrison@...]
> Sent: Jueves 2 de Noviembre de 2000 11:23

> 2) Extend the foreign key constraint definition language with
> the option [[NO ]INDEXES]. The default would be INDEXES.

And maybe provide a way to reverse the default setting, so it can become NO
INDEXES?

> 3) Extend the foreign key constraint definition language with
> the option [[IN]ACTIVE]. The default would be ACTIVE.
> INACTIVE will create the constraint definition (so higher
> level languages that use it for navigation) but not enforce
> it, except in terms of dependencies - an error if you try
> to delete columns or tables referenced in the constraint.
> This is a bit harder, but, I think, not very hard.

I don't understand what you mean: allow the indexes to be created inactive
for the PK (so natural is used) or tell the engine to allow any user to
break the RI by not checking the FK declarations at all when inserting,
deleting and updating?

While IB was in beta testing, I asked for built-in detection of things like
these:
- PK and UNIQUE declarations that use exactly the same fields in the same
sequence (A,B,C) and order (ASC/DESC).
- Manually defined indexes that use the same fields, sequence and order than
automatically created indexes (and the inverse, if the manual index already
exists).
- Overlapped unique and non-unique indexes that apart from the uniqueness,
have the same declaration.

C.