On Sat, Apr 11, 2009 at 01:43:35PM -0700, David Fetter wrote:
> On Sat, Apr 11, 2009 at 04:30:02PM -0400, Tom Lane wrote:
> > Bruce Momjian <bruce@momjian.us> writes:
> > > I assume the 'type' column will identify triggers, i/o functions
> > > (cstring), window functions, and maybe aggregates too; this solves
> > > several problems at once.
> >
> > +1 for all except i/o functions. The cstring check for that was always
> > flat-out wrong, and getting it right is far more expensive than it's
> > worth --- AFAICS you'd have to grovel through all entries in pg_type.
>
> I'll leave it out :)
>
> > But aggregates are only relevant if we decide to start showing
> > aggregates in \df --- is there consensus for that?
>
> I'd throw 'em in.
It occurs to me that we ought to allow for a possibility that a
function can be more than one special case. For example, sum() is
both an aggregate and a windowing function, while rank() is only a
windowing function.
Working on a patch that allows a concise description of both.
Cheers,
David.
--
David Fetter <david@fetter.org> http://fetter.org/
Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter
Skype: davidfetter XMPP: david.fetter@gmail.com
Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate