Oh, sorry I missed that.
Still if C1 and C2 are interchangable, a rule could force C1 <= C2 and
swap them if necessary.
Richard Huxton wrote:
>
> On Friday 27 Sep 2002 5:17 pm, Jean-Luc Lachance wrote:
> > What's wrong with
> > CREATE UNIQUE INDEX foo_both_uniq ON foo(a,b);
> > ???
>
> Because he specifically wanted values of ('a','b') and ('b','a') to be treated
> as equivalent (see quote).
>
> > > > INSERT INTO test (c1,c2) VALUES('a','c');
> > > > INSERT INTO test (c1,c2) VALUES('c','a');
>
> Note Stephen Szabo's observation that I'd missed the obvious need for some
> separator so ('a','ab') is different from ('aa','b') - Doh!
>
> - Richard Huxton