Greg Stark <gsstark(at)mit(dot)edu> writes:
> That would be nice and simple and clean, except that postgres's type system
> doesn't work that way. It works backwards from every other language. Instead
> of every variable and function return value having a rigid type and forcing
> the interpretation of the surrounding expression, postgres seems to work top
> down.
[blinks...] I don't think so.
regards, tom lane