On Fri, 2009-07-24 at 14:10 -0400, Tom Lane wrote:
> Hmmm ... that suggests that we may not be handling IS DISTINCT FROM
> correctly for rowtypes. I haven't read that part of the spec, but
> I would've guessed that the spec wants it to be consistent with
> IS NULL.
>
Our behavior appears to match the standard, which covers this case in
8.15.General Rules.1.c.iv.
Regards,Jeff Davis