Re: inconsistent composite type null handling in plpgsql out variable

2009/9/1 Sam Mason <sam@samason.me.uk>:
> On Mon, Aug 31, 2009 at 07:26:59PM +0200, Pavel Stehule wrote:
>> 2009/8/31 Sam Mason <sam@samason.me.uk>:
>> > The more awkward case (to me anyway) is that the standard says (1,NULL)
>> > IS NULL should evaluate to TRUE.
>>
>> what?
>>
>> only (NULL, NULL) IS NULL is true
>
> Bah, sorry you're right! =C2=A0I was rattling my favorite tin and getting
> mixed up with the behavior with IS NOT NULL, the negation of which
> would say this row is null. =C2=A0I.e:
>
> =C2=A0SELECT NOT (1,NULL) IS NOT NULL;
>
> evaluates to TRUE. =C2=A0I think the consensus is that we should continue=
 to
> follow the spec on this, but I was getting confused as to which operator
> contains the EXISTS and FORALL operator. =C2=A0I.e. a value "v" IS NULL i=
ff
> all elements of "v" are not 'the null value', whereas "v" IS NOT NULL
> iff an element of "v" is 'the null value'.
>
>> p.s. what isn't consistent (maybe - there are more possible
>> interpretations) is
>>
>> (NULL, NULL) IS DISTINCT FROM NULL is true
>
> Yup, I'd agree with Merlin that a ROW consisting entirely of 'null
> values' should itself be 'the null value' (to use the terminology from
> the copy of the SQL spec I'm reading). =C2=A0I think this should also work
> recursively:
>
> =C2=A0SELECT ROW(ROW(NULL)) IS DISTINCT FROM NULL;
>
> should return FALSE, in my understanding.

it's question. You ask, is it (NULL, NULL) same as NULL. Without some
reduction - ROW(NULL, NULL) is really different than NULL.

Pavel

>
> --
> =C2=A0Sam =C2=A0http://samason.me.uk/
>
> --
> Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-bugs
>