Ilir Gashi <[EMAIL PROTECTED]> writes:
> I saw this behaviour in PostgreSQL 7.2. (Once again, I know this is an old
> release but I do not have a newer version installed, and I am only using
> the server for research purposes). If you execute the following statement
> SELECT (CAST('01.01.2004 10:01:00' AS TIMESTAMP) - CAST('01.01.2004
> 10:00:00' AS TIMESTAMP)) + CAST('01.01.2004 10:00:00' AS TIMESTAMP);
More recent releases fail with
regression=# SELECT (CAST('01.01.2004 10:01:00' AS TIMESTAMP) - CAST('01.01.2004
10:00:00' AS TIMESTAMP)) + CAST('01.01.2004 10:00:00' AS TIMESTAMP);
ERROR: operator does not exist: interval + timestamp without time zone
HINT: No operator matches the given name and argument type(s). You may need to add
explicit type casts.
(There is a timestamp + interval operator, so you could make it work by
flipping around the outer addition.)
It looks like what is happening in 7.2 is the parser is choosing the
closest available operator, which happens to be time + date, and
then coercing the interval and timestamp to time and date respectively.
The former isn't so bad but the latter loses your 10AM data...
Since then (I think in 7.3) we tightened the rules so that
information-losing coercions, such as timestamp to date, won't be
invoked implicitly.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 6: Have you searched our list archives?
http://archives.postgresql.org
