Re: inconsistent composite type null handling in plpgsql out variable

On Fri, Aug 28, 2009 at 02:06:02PM -0400, Merlin Moncure wrote:
> 3) If we decide the sql standard is correct, so that (null, null) is
> null == true, then we should observe rule 1 and make things work in
> consistent way.  This means, for example, that null::foo and (null,
> null)::foo should not be distinct.

The more awkward case (to me anyway) is that the standard says (1,NULL)
IS NULL should evaluate to TRUE.

I'd never noticed the ROW / RECORD dichotomy before; could one of these
be made SQL compatible and the other use more sane semantics?

--
  Sam  http://samason.me.uk/