Re: RfD: more powerful "any" types

On Sep 8, 2009, at 9:57 AM, Tom Lane wrote:

> Intentionally so, because there's not a whole lot you can *do* with an
> ANY parameter, other than checking it for null.  Perhaps the real
> question is about what semantics you're expecting for these
> unconstrained parameters.

For my purposes, I guess implicit casting for comparing values, as in
    arg_a IS DISTINCT FROM arg_b

and
    arg_a = arg_b

or
    arg_a <> arg_b

It'd work if there was a way to cast one to the type of the other,
such as comparing a TEXT to a VARCHAR. But maybe that'd be too magical…

Best,

David