> It is do-able, but I'm not sure it is desirable. If there is an exact
> function or operator match, then that would be preferred, right? The
> "binary-compatible" is a cheat to help performance and to allow us to
> not have to define _every_ possible combination of functions and
> operators. So it's helped on some things, but probably shouldn't become
> the only mechanism. For one thing, if you change the underlying
> implementation of a type, then it would not longer be binary compatible
> with a second one, and you would need the explicit functions and
> operators anyway.
>
> I didn't change any of the type matching behavior of the aggregate
> functions that I can recall. Maybe I should have; if I have time I'll
> look at it. That would probably be better than bothering to implement
> something explicitly for oids:
>
> postgres=> select min(oid) from foo;
> ERROR: No such function 'min' with the specified attributes
OK. A question. Aren't oid's unsigned int, while int4 is signed. How
does binary compatable indexes handle this. Do oid's grater than 2gig
work with the int4 indexes? Do negative integers work with oid indexes?
--
Bruce Momjian | 830 Blythe Avenue
maillist@candle.pha.pa.us | Drexel Hill, Pennsylvania 19026
+ If your life is a hard drive, | (610) 353-9879(w)
+ Christ can be your backup. | (610) 853-3000(h)