Обсуждение: COMMENT on function's arguments

Does it make sense to have a comment on function's arguments? Of course 
it is possible to include these comments in a function's comment, but 
may be better to have them in more formalized way like comments on 
columns of a table. IDEs may use this information when providing hints 
for a function like in other languages.

On Tue, Jun 12, 2012 at 10:59 PM, Vlad Arkhipov <arhipov@dc.baikal.ru> wrote:
> Does it make sense to have a comment on function's arguments? Of course it
> is possible to include these comments in a function's comment, but may be
> better to have them in more formalized way like comments on columns of a
> table. IDEs may use this information when providing hints for a function
> like in other languages.

This would be somewhat tricky, because our COMMENT support assumes
that the object upon which we're commenting has an ObjectAddress, and
individual arguments to a function don't, although perhaps the
sub-object-id stuff that we currently use to handle comments on table
columns could be extended to handle this case.  I guess I wouldn't
object to a well-done patch that made this work, but creating such a
patch seems likely to be tricky, owing to the fact that there's
nothing in the system that thinks of the individual arguments to a
function as separate objects at present.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Robert Haas <robertmhaas@gmail.com> writes:
> On Tue, Jun 12, 2012 at 10:59 PM, Vlad Arkhipov <arhipov@dc.baikal.ru> wrote:
>> Does it make sense to have a comment on function's arguments?

> This would be somewhat tricky, because our COMMENT support assumes
> that the object upon which we're commenting has an ObjectAddress, and
> individual arguments to a function don't, although perhaps the
> sub-object-id stuff that we currently use to handle comments on table
> columns could be extended to handle this case.  I guess I wouldn't
> object to a well-done patch that made this work, but creating such a
> patch seems likely to be tricky, owing to the fact that there's
> nothing in the system that thinks of the individual arguments to a
> function as separate objects at present.

Also, once you'd created the infrastructure needed to *store* such
comments, what would you actually *do* with them?  I find it hard to
imagine squeezing them into \df+ displays, for instance, without
impossible clutter.

Like Robert, I stand ready to be proven wrong by a well-designed patch;
but this seems like something that would take a lot more work than
it's really worth.
        regards, tom lane