>> Ambiguity?
>
> Yeah, I'm worried about the possibility of parser failing to resolve
> which operator is meant.
Any idea of how to test for that? What kind of situations would make
things difficult for the parser?
Also, how is this any different for any optional data type which
overrides = ?
> No, I don't think so. For people for whom the right thing is happening,
> you'll risk making it (a) wrong and (b) lots slower.
Well, I'm dubious that current behavior is the right thing for anybody.The best I could do to answer that would be an
informalcommunity survey.
> For people for
> whom the wrong thing is happening, maybe you'll fix it so it's
> semantically right, but if indexes don't work they still won't be happy.
So I'd need to add operator classes and indexing support functions as
well, then, presumably. Annoying, but not impossible.
--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com