Обсуждение: BUG #9955: The time zone value for South Africa (SAST) is 01:00:00 not 09:30:00

The following bug has been logged on the website:

Bug reference:      9955
Logged by:          Valentin Gora
Email address:      valentin.gora@vitheia.com
PostgreSQL version: 9.2.0
Operating system:   Windows, Linux ( OpenSuse )
Description:

Wrong value in pg_timezone_abbrevs ( generated by pg_timezone_abbrevs()
function, implemented internally )

valentin.gora@vitheia.com writes:
> Wrong value in pg_timezone_abbrevs ( generated by pg_timezone_abbrevs()
> function, implemented internally )

This isn't terribly helpful.  What do you think should be changed,
exactly, and to what?

            regards, tom lane

On 04/11/2014 06:54 PM, Tom Lane wrote:
> valentin.gora@vitheia.com writes:
>> Wrong value in pg_timezone_abbrevs ( generated by pg_timezone_abbrevs()
>> function, implemented internally )
> This isn't terribly helpful.  What do you think should be changed,
> exactly, and to what?

It's in the subject.  The SAST utc_offset is incorrect for South Africa
and shows instead South Australia because the abbreviation is
ambiguous.  Not sure if there's anything to do about that.
--
Vik

On 04/11/2014 11:07 PM, Vik Fearing wrote:
> On 04/11/2014 06:54 PM, Tom Lane wrote:
>> valentin.gora@vitheia.com writes:
>>> Wrong value in pg_timezone_abbrevs ( generated by pg_timezone_abbrevs()
>>> function, implemented internally )
>> This isn't terribly helpful.  What do you think should be changed,
>> exactly, and to what?
> It's in the subject.  The SAST utc_offset is incorrect for South Africa
> and shows instead South Australia because the abbreviation is
> ambiguous.  Not sure if there's anything to do about that.

On the other hand, a Google search for "sast time zone" shows nothing
but South Africa, and "sast time zone australia" still shows South
Africa except for the first result.

And a grep for SAST in the source tree shows:

src/timezone/tznames/Australia.txt:SAST    34200    # South Australian
Standard Time (not in zic)
src/timezone/tznames/Africa.txt:SAST     7200    # South Africa Standard
Time
src/timezone/tznames/Default:SAST    34200    # South Australian
Standard Time (not in zic)

I'm not sure exactly what that means but it would make sense to me that
the default should be the one not marked "not in zic".

--
Vik

On 4/11/2014 2:07 PM, Vik Fearing wrote:
> It's in the subject.  The SAST utc_offset is incorrect for South Africa
> and shows instead South Australia because the abbreviation is
> ambiguous.  Not sure if there's anything to do about that.

TZ abreviations /are/ ambiguous.    CST is both Central Standard Time
(in the USA) and China Standard Time (in the PRC).



--
john r pierce                                      37N 122W
somewhere on the middle of the left coast

Vik Fearing <vik.fearing@dalibo.com> writes:
> On 04/11/2014 06:54 PM, Tom Lane wrote:
>> This isn't terribly helpful.  What do you think should be changed,
>> exactly, and to what?

> It's in the subject.  The SAST utc_offset is incorrect for South Africa
> and shows instead South Australia because the abbreviation is
> ambiguous.  Not sure if there's anything to do about that.

Oh, I failed to read the wrapped-around part of the subject line :-(

We do have a plan for dealing with ambiguous abbreviations, but it
seems like it's not being followed here.  (Looking at the comments in
the africa timezone file, it seems like SAST may be a relatively recent
newcomer to the IANA TZ database; so perhaps we missed doing something
about it when it was added.)

To wit: there's already an Australia-specific timezone abbreviation file
that Aussies are supposed to install if they want their zone abbreviations
to be recognized in preference to other meanings given to the same
abbreviations elsewhere in the world.  Given that, it seems like SAST
ought to mean the South African zone by default, and you get the Aussie
version only if the "Australia" abbreviation set is installed.

BTW, the IANA database says that the SA meaning for SAST is +2:00,
not +1:00.  If the OP really wants us to use +1:00 he'd better present
some evidence that IANA is wrong.

            regards, tom lane