Re: Defining a foreign key with a duplicate column is broken

David Rowley <dgrowleyml(at)gmail(dot)com> writes:
> On Sat, Aug 9, 2014 at 3:34 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> There's no need for a new error message I think, because we should just
>> ignore such indexes. After all, there might be a valid matching index
>> later on.

> hmm, but if the user attempts to define the foreign key that contains a
> duplicate column in the REFERENCES part, then we'll never "find" any
> indexes, so there's no point in looking at all.

OK, now that I'm a bit more awake, I agree with that.

> I've attached a version of the patch that's a little smarter when it comes
> to doing the duplicate checks in the attnums array...

Applied with some cosmetic adjustments. I didn't bother with the
regression test either --- this doesn't seem like something that needs
permanent memorialization as a regression test.

regards, tom lane