Thread: BUG #4528: (rounding?) error with some timezone names

The following bug has been logged online:

Bug reference:      4528
Logged by:          chris wood
Email address:      chrisj.wood@sympatico.ca
PostgreSQL version: 8.3.4
Operating system:   Debian
Description:        (rounding?) error with some timezone names
Details:

again sorry I do not have access to 8.3.5, but it should be a quick check
for anyone who does.

postgresbugs=# select version() ;
                                          version
----------------------------------------------------------------------------
----------------
 PostgreSQL 8.3.4 on i486-pc-linux-gnu, compiled by GCC gcc-4.3.real (Debian
4.3.2-1) 4.3.2
(1 row)

postgresbugs=# select * from pg_catalog.pg_timezone_names where name =
'right/Pacific/Wake' ;
        name        | abbrev | utc_offset | is_dst
--------------------+--------+------------+--------
 right/Pacific/Wake | WAKT   | 12:00:00   | f
(1 row)

postgresbugs=# select '2008-01-09 07:00:00-00'::timestamptz at time zone
'right/Pacific/Wake' ;
      timezone
---------------------
 2008-01-09 18:59:37   <<<<<<<<<???????
(1 row)

postgresbugs=# select '2008-01-09 07:00:00-00'::timestamptz at time zone
'WAKT' ;
      timezone
---------------------
 2008-01-09 19:00:00
(1 row)


In version 8.3.1 there were 586 timezone names that manifest a similar
error.

"chris wood" <chrisj.wood@sympatico.ca> writes:
> postgresbugs=# select '2008-01-09 07:00:00-00'::timestamptz at time zone
> 'right/Pacific/Wake' ;
>       timezone
> ---------------------
>  2008-01-09 18:59:37   <<<<<<<<<???????
> (1 row)

The "right" timezones take leap seconds into account, so this isn't
surprising.

(We don't actually support leap seconds, so I wonder whether we
shouldn't filter those zones out of the view.  They're not there at all
if you set up Postgres to use an internal tz database, but I suppose we
have to accept that they're likely to be present in a system-supplied tz
database.)

            regards, tom lane