This is a follow-up to a question I asked earlier.
On 10/19/07, Gregory Stark <stark@enterprisedb.com> wrote:
> What you need is:
>
> CREATE UNIQUE INDEX foo_uniq_x_y on foo (canonicalize(x,y));
>
> > LOCATION: base_yyerror, scan.l:795
OK, now, what if instead of this
-> ALTER TABLE foo ADD CONSTRAINT foo_uniq_x_y UNIQUE(canonicalize(x,y));
what I was trying to do was this
-> ALTER TABLE foo ADD CONSTRAINT foo_pkey PRIMARY KEY(canonicalize(x,y));
Of course, this also elicits a syntax error, but is there a way to
achieve the same effect by creating some index?
I realize that, as far as the indexing and the enforcement of the
uniqueness constraint are concerned, the earlier solution using
"CREATE UNIQUE INDEX" works perfectly fine. But some software that I
use (Perl modules, etc.) get very confused whenever a table does not
have a primary key.
Of course, I could just add one more PRIMARY KEY constraint:
ALTER TABLE foo ADD CONSTRAINT foo_pkey PRIMARY KEY(x, y);
which would be satisfied automatically whenever the UNIQUE constraint
implicit in the index foo_uniq_x_y (see above) is satisfied. But this
amounts to maintaining an entirely superfluous index.
In short, my question is: is there a way to designate a pre-existing
UNIQUE INDEX (based on data contained in NOT NULL fields) as the basis
for a table's PRIMARY KEY?
TIA!
kj