Neil Conway <neilc@samurai.com> writes:
> On Tue, 2008-06-03 at 13:03 +0200, Pavel Stehule wrote:
>> this patch add support of table functions syntax like ANSI SQL 2003.
> I'm not necessarily opposed to this, but I wonder if we really need
> *more* syntax variants for declaring set-returning functions.
I've been saying right along that we don't. The proposed patch adds
no measurable new functionality; its only reason to live is standards
compliance, and I'm not convinced that's worth the confusion. Our
implementation of functions is (and always will be) far enough away
from the standard that notational issues like this are hardly the top
of the problem list for someone wishing to import a spec-compliant
function.
(It's also worth asking where the import is coming from. Who implements
the spec syntax anyway? DB2 maybe, but when was the last time we heard
from anyone trying to migrate from DB2 to PG?)
regards, tom lane
