Re: Complexity of contrib types

> > Can't they SELECT from pg_proc?
>
> Making it a function would probably make the type-addition script
> easier.

I guess, but much less flexible.

>
> > >
> > > The real solution would of course be extending the (Postgre)SQL language
> > > to find the OIDs automatically, like Oracle currently does for its
> > > COMMENT
> > > statement, an equivalent of which could be used in PostgreSQL to insert
> > > values in pg_description on the fly.
> >
> > We return oid's as part of an INSERT.  Is that what you meant?
>
> It is very hard (probably impossible) to use them from a psql script.
>
> If I remember the syntax right (have'nt used Oracle for >=2 years),
> I could do:
>
> COMMENT "this is a nice table" ON TABLE nice_table;
> COMMENT "this is an unnecessary field from a nice table"
>   ON FIELD nice_table.unnecessary_field;
>
> Of course, to fully support it we would need a much improved foreign
> key support, so that we could set the ON DELETE CASCADE for the
> commented on tables, and do it so that the foreign key can references
> _any_ table ;).
>
> If we could manage that, we could really call PostgreSQL an OO database.

We could create a function that returned the previously inserted oid,
and use that in the next query.

    insert into test values (4);
    update test3 set val = lastoid();

Just remember the lastoid inserted in the backend code.  Seems easy.  Do
you want it added to the TODO list.

--
Bruce Momjian                          |  830 Blythe Avenue
maillist@candle.pha.pa.us              |  Drexel Hill, Pennsylvania 19026
  +  If your life is a hard drive,     |  (610) 353-9879(w)
  +  Christ can be your backup.        |  (610) 853-3000(h)