Problem with zero year
We added the following commit in 8.4:
/src/backend/utils/adt/datetime.c
tgl
Reject year zero during datetime input, except when it's a 2-digit
year (then it means 2000 AD). Formerly we silently interpreted this
as 1 BC, which at best is unwarranted familiarity with the implementation.
It's barely possible that some app somewhere expects the old behavior,
though, so we won't back-patch this into existing release branches.
The problem is that the 2-digit year check is for <=2 digits, not
exactly two digits:
/*
* When processing a year field, mark it for adjustment if it's only one
* or two digits.
*/
if (*tmask == DTK_M(YEAR))
*is2digits = (flen <= 2);
This leads to some unexpected outputs:
test=> select '1-1-0'::date;
date
------------
2000-01-01
test=> select '1-1-0 BC'::date;
ERROR: date/time field value out of range: "1-1-0 BC"
LINE 1: select '1-1-0 BC'::date;
^
test=> select '1-1-0 AD'::date;
date
------------
2000-01-01
test=> select '1-1-000'::date;
ERROR: date/time field value out of range: "1-1-000"
LINE 1: select '1-1-000'::date;
I think the BC part is fine because that can't possibily be 2000 AD, but
having '0' interpreted as 2000 seems counter to the commit message text;
should the assignment be changed to:
if (*tmask == DTK_M(YEAR))
*is2digits = (flen == 2);
--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com
+ If your life is a hard drive, Christ can be your backup. +
Bruce Momjian <bruce@momjian.us> writes:
The problem is that the 2-digit year check is for <=2 digits, not
exactly two digits:
...
This leads to some unexpected outputs:
test=> select '1-1-0'::date;
date
------------
2000-01-01
We've interpreted that like that since 7.4, without complaints; and
I think it was an intentional change then (since 7.3 doesn't accept it).
I do not recommend changing it.
regards, tom lane
Tom Lane wrote:
Bruce Momjian <bruce@momjian.us> writes:
The problem is that the 2-digit year check is for <=2 digits, not
exactly two digits:
...
This leads to some unexpected outputs:test=> select '1-1-0'::date;
date
------------
2000-01-01We've interpreted that like that since 7.4, without complaints; and
I think it was an intentional change then (since 7.3 doesn't accept it).
I do not recommend changing it.
OK, the release note text will be:
Reject year '0 BC' and years '000' and '0000' (Tom)
Previously these were interpreted as 1 BC.
--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com
+ If your life is a hard drive, Christ can be your backup. +