Gregory Stark <stark@xxxxxxxxxxxxxxxx> writes: > ISTM this is the one that's wrong. "CET" is standard time, it, GMT+1. > If you want a timezone which switches between CET and CST automatically you > should use something like Europe/Paris. Well, actually he *is* using such a zone: regression=# select * from pg_timezone_names where name = 'CET'; name | abbrev | utc_offset | is_dst ------+--------+------------+-------- CET | CEST | 02:00:00 | t (1 row) But regression=# select * from pg_timezone_abbrevs where abbrev = 'CET'; abbrev | utc_offset | is_dst --------+------------+-------- CET | 01:00:00 | f (1 row) The problem is that one of these two statements is using the abbrev meaning and the other is using the timezone meaning. We don't have much control over the zone definition, so I'm thinking maybe the abbrev should be removed from the tznames lists. But that seems a bit sucky too. Does anyone have any idea if the zic folk would be responsive to a complaint that defining a timezone with the same name as an abbreviation is a bad idea? regards, tom lane
