David G Johnston <david.g.johnston@gmail.com> writes:
> Just looking for recollection regarding why these were omitted initially and
> if anyone has concerned adding them in follow-up.
My recollection is that RANGE requires some data-type-specific behavior
that we don't have any provision for in PG's datatype extension framework
(something about increment/decrement I think, but too lazy to consult the
archives for details). The original window-function patch had some klugy
hard-wired behavior for a small set of datatypes, which we quite properly
rejected as not being up to project standards: datatype extensibility is
one of PG's defining features, and we're not going to give it up lightly.
Nobody's yet done the work to get something that would pass muster.
> With the recent
> hypothetical work being done maybe these can be re-evaluated in a fresh
> light?
AFAIK those functions are unrelated to this problem.
regards, tom lane
