Troels Arvin <troels@arvin.dk> writes:
> In Jim Melton and Alan Simon's "SQL:1999 - Understanding Relational
> Language Components" (ISBN 1-55860-456-1), they write that the following
> is to be interpreted as a TIMESTAMP WITH TIME ZONE value:
> TIMESTAMP '2003-07-29 13:19:30.5+02:00'
> PostgreSQL interprets the above as a TIMESTAMP WITHOUT TIME ZONE value of
> '2003-07-29 13:19:30.5', i.e. it simply discards the '+02:00' part and
> fails to interpret it as being of TIMESTAMP WITH TIME ZONE type.
That's true, and I think we are unlikely to change it. Postgres
interprets this construct as a special case of a general
datatype_name 'literal string'
construction. To allow the contents of the literal to determine the
datatype specification would break the general construct completely.
regards, tom lane