2009/4/11 Andrew Gierth <andrew@tao11.riddles.org.uk>:
>>>>>> "Tom" == Tom Lane <tgl@sss.pgh.pa.us> writes:
>
> >>> Perhaps more to the point: the previous round of discussion about
> >>> this already rejected the idea of treating window functions as a
> >>> category fundamentally separate from plain functions --- that is,
> >>> we are not following the "aggregate" model of having separate
> >>> commands for aggregate functions.
>
> >> I hadn't seen any such a consensus.
>
> Tom> We do not have CREATE WINDOW FUNCTION, DROP WINDOW FUNCTION,
> Tom> ALTER WINDOW FUNCTION, etc. If psql uses \dw it will be
> Tom> presenting a different world view than exists at the SQL level.
>
> I'm not sure why that would matter. The fact that it is CREATE
> FUNCTION ... WINDOW rather than CREATE WINDOW FUNCTION doesn't mean
> that window functions aren't a distinctly different animal to normal
> functions. The usage and syntax is different enough that putting them
> all together under \df seems forced.
Yeah, but all the window functions are stored in pg_proc.
Regards,
--
Hitoshi Harada