[tz] [PROPOSED] Document POSIX versions more carefully

Paul Gilmartin PaulGBoulder at AIM.com
Sun Jan 28 22:29:26 UTC 2024


On 1/28/24 13:35:09, Paul Eggert via tz wrote:
> On 2024-01-28 07:23, Paul Gilmartin wrote:
>> Is a draft of that next POSIX publicly available?
> 
> You can get a copy by signing up as per the News section of:
> 
> https://www.opengroup.org/austin/
>  .
Thanks.

>> One significant(?)
>> UNIX-branded implementation supports only “POSIX.1-2017-like TZ strings”:
>>     <https://www.ibm.com/docs/en/zos/3.1.0?topic=variable-command-format>
> 
> Thanks, I wasn't aware that z/OS supports only POSIX.1-2017 syntax for TZ strings. This behavior was common in older platforms, and I assume can happen even today in stripped-down platforms other than z/OS.
>  .
It may be worse than that:
     <https://www.ibm.com/docs/en/zos/3.1.0?topic=services-zos-unix-system-users-guide>
     It also helps you use the functions specified in the POSIX.2
     standard (IEEE Std 1003.2-1992 and ISO/IEC 9945-1992
     International Standard;

1992?  But it supports the USA default March/November transition dates.

> As a transition aid for these systems, it's helpful to have a mapping from TZDB settings like TZ='Europe/London' to POSIX.1-2017 approximations like TZ='GMT0BST,M3.5.0/1,M10.5.0'. Perhaps TZDB itself should support such a mapping, though now's not the best time to add that as we're so close to a release. In the meantime I installed the attached documentation patch, which references the posix_tz_db package that generates mapping tables.
> .
Stumbled on:  The last line of each TZ file seems to contain the
POSIX.1-2017 format, which could be copied from the z/OS Java
distribution.

-- 
Thanks again,
gil



More information about the tz mailing list