Re: [HACKERS] Subselects and NOTs

Bruce Momjian wrote:
>
> >
> > > Added to TODO.
> > > > ... In Postgres, a column can be
> > > > designated "not null", the default being to allow nulls.  In the
> > > > default Sybase configuration, it is the other way around.  In the
> > > > interest of writing cross database compatible code, I try to always
> > > > call out columns as either "null" (nulls allowed), or "not null"
> > > > (nulls not allowed).  Unfortunately, Postgres does not support this
> > > > In short, it would be nice if Postgres would take "null" as a type
> > > > specifier as well as "not null".
> >
> > We currently get a shift/reduce parsing conflict on this since NULL can be
> > specified in other constraint clauses and since the constraint clauses are
> > only whitespace delimited. It might be that this part of the parser can be
> > redone, or perhaps the only way around is to restrict the ordering of the
> > constraints. But NULL constraint is not SQL92 and free ordering is...
>
> OK, removed from TODO.

I'll look into this since I requested it.  I'll see if I can add this
without breaking everything else.

Ocie Mitchell