At Sun, 22 May 2022 09:55:37 +0200, Przemysław Sztoch <przemyslaw@sztoch.pl> wrote in
> David G. Johnston wrote on 5/19/2022 3:46 AM:
> > As an aside, I'd rather overcome this particular objection by having
> > the CREATE object command all accept an optional "COMMENT IS" clause.
> >
> I believe that it is not worth dividing it into a separate program.
>
> The same --comment argument is needed for the createdb command.
David didn't say that it should be another "program", but said it
should be another "patch/development", because how we implement the
--comment feature is apparently controversial.
It doesn't seem to be explicity mentioned that "createuser is mere a
shell-substitute for the SQL CREATE ROLE", but I feel the same with
Shinya that it is. We could directly invoke "COMMENT ON" from
createuser command, but I think it is not the way to go in that light.
We can either add COMMENT clause only to "CREATE ROLE" , or "COMMENT
IS" clause to all (or most of) "CREATE object" commands, or something
others. (Perhaps "COMMETN IS" requires "ALTER object" handle comments,
and I'm not sure how we think about the difference of it from "comment
on" command.) We might return to "comment on" in the end..
Anyway, after fixing that issue we will modify the createrole command
so that it uses the new SQL feature. I find no hard obstacles in
reaching there in the 16 cycle.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center