Re: [Emacs-diffs] master 085c7f6 2/2: Test format-time-string with zone arg

[Eli Zaretskii]

> Cc: address@hidden, address@hidden
> From: Paul Eggert <address@hidden>
> Date: Mon, 1 May 2017 10:49:17 -0700
> 
> On 05/01/2017 12:43 AM, Eli Zaretskii wrote:
> > I've found 2 issues with this test in the MinGW build on MS-Windows. 
> 
> Yes, clearly MS-Windows is buggy in this area, but only time nerds like 
> me care about this sort of thing so I installed the attached, which 
> should stop testing the buggy timestamps in early 1970 (while avoiding 
> portability problems with timestamps after mid-1972).

Thanks.  This version still has problems in the MinGW build:

 . The Windows runtime supports only 3-letter TZ abbreviations, so
   things like NZST and AEST confuse it.  (It doesn't care about the
   letters, it just needs to be 3 of them.)

 . The Windows runtime doesn't support Posix DST transition rules in
   the TZ setting, it uses the "old USA rules" (see
   https://msdn.microsoft.com/en-us/library/90s5c885.aspx for the
   details), so "AEST-10AEDT,M10.5.0,M2.5.0/3" doesn't produce the
   correct result, even if AEST and AEDT are replaced with 3-letter
   zones.  So if we want to test that DST works, the rules must be
   compatible with the old US rules, i.e. between the 1st Sunday of
   April and the last Sunday of October.