Re: Type Categories for User-Defined Types

"David E. Wheeler" <david(at)kineticode(dot)com> writes:
> Well, there are already citext aliases for all of those operators, for
> this very reason. There are citext aliases for a bunch of the
> functions, too (ltrim(), substring(), etc.), so I wouldn't worry about
> adding more. I've added more of them since I last sent a patch, mainly
> for the regexp functions, replace(), strpos(), etc. I'd guess that I'm
> about half-way there already, and there probably are a few I wouldn't
> bother with (like timezone()).

That's exactly what I don't really want to do; if you are adding aliases
*only* to get rid of ambiguity-errors, and not to alter functionality,
then I think you're doing the wrong thing. Adding more aliases can
easily make the situation worse.

> Anyway, would this issue then go away once the type stuff was added
> and citext was specified as TYPE = 'S'?

Yeah, that's the point of the proposal. I think the issue has come up
once or twice before, too, else I'd not be so interested in a general
solution. (digs in archives ... there was some discussion of this
in connection with unsigned integer types, and I seem to recall older
threads but can't find any right now.)

regards, tom lane