En 01 Sep 2002 13:44:37 -0400
Rod Taylor <rbt@zort.ca> escribió:
> > > On Sun, 2002-09-01 at 04:00, Neil Conway wrote:
> > > > This patch fixes a minor inaccuracy in the documentation: NOT NULL
> > > > is not synonymous with CHECK (xxx IS NOT NULL) -- for example,
> > > > consider ALTER TABLE ADD PRIMARY KEY, which checks for 'NOT NULL',
> > > > not a check constraint.
>
> The case you brought up is more of a bug or mis-implementation of alter
> table / add primary key. I guess you're right. They're not, but they
> should be.
I think it's very difficult in the general case to determine from a
CHECK constraint if it is setting some column to not null; but if it was
easy, then you better translate(*) it into the SET NOT NULL form, and leave
the ADD PRIMARY KEY code alone.
Anyway IMHO this should be documented to prevent possible confusion.
(*) but I think auto-converting some things into some other things that
may not be what the user wants is a recipe for user frustration.
--
Alvaro Herrera (<alvherre[a]atentus.com>)
"El sabio habla porque tiene algo que decir;
el tonto, porque tiene que decir algo" (Platon).
