Tom Dunstan <pgsql(at)tomd(dot)cc> writes:
> The enum patch [1] does exactly this with an ANYENUM pseudo-type. It
> should provide a pretty good overview of what will be required.
ANYENUM? What's the use-case for that? These special cases in the
type system are enough of a pain-in-the-neck for the code that I'm
disinclined to add one without a very solid argument for it.
regards, tom lane