Re: Allow an alias for the target table in UPDATE/DELETE

Neil Conway <neilc(at)samurai(dot)com> writes:
> Did you actually test this?

No, I was just looking at the y.output file to see what would happen.

> neilc=# update t1 set set a = 500 where set.a > 1000;
> UPDATE 0
> (Using essentially the patch you posted.)

[ scratches head... ] That shouldn't have worked. I'll have to look
again.

> Well, if necessary we can just use an alternate production for the
> DELETE case, as there is no SET ambiguity to worry about.

Yeah, I thought of that too and rejected it as being too much trouble
for too small a case. I'm really considerably more worried about the
question of whether attaching a precedence to SET might cause trouble
later. But it's only a hypothetical problem at this point.

> So I'm inclined to favor #2.

OK, motion carries. I'll check what's happening in the case above
and commit if there's not something wrong.

regards, tom lane