Re: BUG #13691: Postgres reverse timezone system

Thomas Munro <thomas(dot)munro(at)enterprisedb(dot)com> writes:
> It's pretty surprising that '+13' is not negated, but '+13:00' is. Is
> that expected?

Um, well, there's yet a third randomly-different standard involved here,
which is that the SQL spec says that timezones can be specified as purely
numeric GMT offsets --- using the ISO sign convention. So "+13" is
captured by that rule; while "+13:00" doesn't look like a plain number
so it gets taken in by the POSIX conventions.

We can probably find a few more standards governing PG's behavior in
this area, if you care to keep poking ;-). But looking for absolute
mathematical consistency in anything having to do with timekeeping is
a lost cause.

Since there pretty much isn't any way that the SQL spec's timezone
rules don't suck, I'd urge avoiding that particular notation. Really
the Olson-style geographically-based zone names are the least ambiguous
and the least likely to be wrong when considering past and future law
changes. If I were the OP I'd be using "Asia/Ho_Chi_Minh".

regards, tom lane