Non-standard TIMESTAMP WITH TIME ZONE literal handling

On Thu, 7 Aug 2003, Troels Arvin wrote:

Hello,

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.

Unless Melton+Simon are wrong, PostgreSQL is not completely following
SQL:1999 regarding TIMESTAMP-like literal parsing.

I think they're correct and we're wrong:

SQL92 5.3 Syntax rules:
17)The data type of a <timestamp literal> that does not specify
<time zone interval> is TIMESTAMP(P), where P is the number of
digits in <seconds fraction>, if specified, and 0 otherwise.
The data type of a <timestamp literal> that specifies <time zone
interval> is TIMESTAMP(P) WITH TIME ZONE, where P is the number
of digits in <seconds fraction>, if specified, and 0
otherwise.

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