23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eric Hanchrow]

Please write in English if possible, because the Emacs maintainers
usually do not have translators to read other languages for them.

Your bug report will be posted to the emacs-pretest-bug@gnu.org mailing list.

Please describe exactly what actions triggered the bug
and the precise symptoms of the bug:

I simply typed M-x display-time-world, and noted that the *wclock*
window contained this text:

    Seattle   Wednesday 10 October 17:49 
    New York  Wednesday 10 October 17:49 
    London    Wednesday 10 October 17:49 
    Paris     Wednesday 10 October 17:49 
    Bangalore Wednesday 10 October 17:49 
    Tokyo     Wednesday 10 October 17:49 

The time was 2007-10-17T16:49:43Z or thereabouts; my time zone is PST8PDT.

In GNU Emacs 23.0.50.1 (i386-mingw-nt5.1.2600)
 of 2007-10-16 on FLAP
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (3.4)'

Important settings:
  value of $LC_ALL: nil
  value of $LC_COLLATE: nil
  value of $LC_CTYPE: nil
  value of $LC_MESSAGES: nil
  value of $LC_MONETARY: nil
  value of $LC_NUMERIC: nil
  value of $LC_TIME: nil
  value of $LANG: nil
  locale-coding-system: cp1252
  default-enable-multibyte-characters: t

Major mode: World clock

Minor modes in effect:
  erc-autojoin-mode: t
  erc-match-mode: t
  erc-log-mode: t
  erc-services-mode: t
  desktop-save-mode: t
  recentf-mode: t
  shell-dirtrack-mode: t
  display-time-mode: t
  iswitchb-mode: t
  tooltip-mode: t
  mouse-wheel-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  unify-8859-on-encoding-mode: t
  utf-translate-cjk-mode: t
  auto-compression-mode: t
  line-number-mode: t
  transient-mark-mode: t

Recent input:
/ l o c <tab> s r <tab> e m <tab> <M-backspace> <M-backspace> 
d a <tab> <return> C-x C-v d o <tab> e m <tab> <return> 
C-n C-n C-n <return> C-v C-x b s h e <return> t M-p 
M-b <M-backspace> C-e SPC - - v e r s i o n <return> 
<lwindow> M-x s e r v e r - s t <tab> <return> S h 
a d e s SPC o f SPC W i l l i a m SPC W e g m a n . 
M-b M-b M a <backspace> <backspace> < a SPC h r e f 
= " C-y " > C-e C-b < / a > C-x C-s C-x # C-x C-z M-x 
e m a c s - v e <tab> <return> C-h N C-x ] C-x ] C-x 
n p C-x n w C-x ] C-n C-x n p M-: M-( f e a t u r e 
p SPC ' m u l M-/ <backspace> <return> C-v M-> C-x 
n w M-x d i s p l <tab> t i <tab> - w <tab> <return> 
C-p C-p C-p ? <backspace> l a k d s j M-r y e C-x k 
<return> C-x 0 C-x F d i s <tab> p <tab> t <tab> i 
<tab> - w <tab> <return> C-n C-n C-n C-n C-n C-e C-x 
C-e C-x n w C-n C-n C-n C-n C-a C-M-d X C-x u C-x F 
<return> C-u C-n C-u C-n C-u C-n C-p C-M-d C-x F <return> 
C-u C-n C-u C-n C-u C-n C-u C-n C-u C-n C-p M-m C-s 
C-w C-w C-w C-w C-s C-s C-s C-a C-p M-0 C-l C-v C-M-n 
C-l C-x C-z M-x M-p <return> M-x r e p o r t - e m 
<tab> <return>

Recent messages:
View mode: type C-h for help, h for commands, q to quit.
Scanning for dabbrevs...100%
t
Mark set
revert-buffer: Buffer does not seem to be associated with any file
Making completion list... [2 times]
t
Undo!
Mark set
Mark saved where search started

-- 
Imagine this movie with Barbara Stanwyck and Fred MacMurray
and it would work for you.  Better still, just rent "Double
Indemnity" and the hell with it.
        -- Roger Ebert, on "Derailed (2005)"

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eli Zaretskii]

> From: Eric Hanchrow <offby1@blarg.net>
> Date: Wed, 17 Oct 2007 09:55:48 -0700
> Cc: 
> 
> I simply typed M-x display-time-world, and noted that the *wclock*
> window contained this text:
> 
>     Seattle   Wednesday 10 October 17:49 
>     New York  Wednesday 10 October 17:49 
>     London    Wednesday 10 October 17:49 
>     Paris     Wednesday 10 October 17:49 
>     Bangalore Wednesday 10 October 17:49 
>     Tokyo     Wednesday 10 October 17:49 

This is because the implementation of this command uses time-zone
specifications that are not portable enough, and in particular cannot
work on Windows.  The problem is in the value of the option
`display-time-world-list'.

Thanks for reporting this.  If no one beats me to it, I will work on
fixing it in a few days.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Jason Rumney]

Eric Hanchrow wrote:
> I simply typed M-x display-time-world, and noted that the *wclock*
> window contained this text:
>
>     Seattle   Wednesday 10 October 17:49 
>     New York  Wednesday 10 October 17:49 
>     London    Wednesday 10 October 17:49 
>     Paris     Wednesday 10 October 17:49 
>     Bangalore Wednesday 10 October 17:49 
>     Tokyo     Wednesday 10 October 17:49 
>
> The time was 2007-10-17T16:49:43Z or thereabouts; my time zone is PST8PDT.
>   

Windows appears not to support the "Continent/Region" timezone names
used in the default value of display-time-world-list. Apparently it is
falling back on London time for the unrecognized timezones.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Phil Sung]

On 10/17/07, Eli Zaretskii <eliz@gnu.org> wrote:
> >     Seattle   Wednesday 10 October 17:49
> >     New York  Wednesday 10 October 17:49
> >     London    Wednesday 10 October 17:49
> >     Paris     Wednesday 10 October 17:49
> >     Bangalore Wednesday 10 October 17:49
> >     Tokyo     Wednesday 10 October 17:49
>
> This is because the implementation of this command uses time-zone
> specifications that are not portable enough, and in particular cannot
> work on Windows.  The problem is in the value of the option
> `display-time-world-list'.

Furthermore, display-time-world does not display the current day,
which I assume it is supposed to do. (It says "10 October" rather than
the current date, "17 October", on every line.)

Perhaps display-time-world-time-format should be "%A %d %B %R %Z"
rather than "%A %m %B %R %Z"?

--Phil

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Jason Rumney]

Eli Zaretskii wrote:
> Thanks for reporting this.  If no one beats me to it, I will work on
> fixing it in a few days.
>   

I added an entry to etc/PROBLEMS already. I don't think it can be fixed
easily - there appear to be multiple problems beyond just the timezone
name, including the fact that standard libraries on Windows only read
the TZ environment variable on startup, and that the system clock on
Windows is kept as localtime so changes to the timezone have a different
effect than on systems that use UTC for the system clock.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eli Zaretskii]

> Date: Thu, 18 Oct 2007 00:46:21 +0100
> From: Jason Rumney <jasonr@gnu.org>
> Cc: Eric Hanchrow <offby1@blarg.net>, emacs-pretest-bug@gnu.org,
> 	Ivan Kanis <apple@kanis.eu>
> 
> system clock on Windows is kept as localtime

Not on W2K, Xp, Vista, etc., AFAIK.  I think what you say is correct
for Windows 9X, but not for the more modern versions.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Stephen J. Turnbull]

A pre-installed Dell server I bought in January had the clock set
to localtime.  YMMV.

On 10/17/07, Eli Zaretskii <eliz@gnu.org> wrote:
> > Date: Thu, 18 Oct 2007 00:46:21 +0100
> > From: Jason Rumney <jasonr@gnu.org>
> > Cc: Eric Hanchrow <offby1@blarg.net>, emacs-pretest-bug@gnu.org,
> >       Ivan Kanis <apple@kanis.eu>
> >
> > system clock on Windows is kept as localtime
>
> Not on W2K, Xp, Vista, etc., AFAIK.  I think what you say is correct
> for Windows 9X, but not for the more modern versions.
>
>
> _______________________________________________
> Emacs-devel mailing list
> Emacs-devel@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-devel
>

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Richard Stallman]

    Perhaps display-time-world-time-format should be "%A %d %B %R %Z"
    rather than "%A %m %B %R %Z"?

Thanks.  Done.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eli Zaretskii]

> Date: Thu, 18 Oct 2007 00:46:21 +0100
> From: Jason Rumney <jasonr@gnu.org>
> Cc: Eric Hanchrow <offby1@blarg.net>, emacs-pretest-bug@gnu.org,
> 	Ivan Kanis <apple@kanis.eu>
> 
> there appear to be multiple problems beyond just the timezone
> name, including the fact that standard libraries on Windows only read
> the TZ environment variable on startup

I don't see on my system that TZ is only read on startup.  The
attached simple test program demonstrates that TZ can be changed
during the program's run with expected results, at least on my XP SP2
box.

So it looks like just by fixing display-time-world-list we could make
this work on Windows as well.

Am I missing something?


------------------------------ cut here -------------------------
#include <time.h>
#include <stdlib.h>
#include <stdio.h>

int main (int argc, char *argv[])
{
   char tstr[1024];
   time_t t;

   t = time (NULL);
   strftime (tstr, sizeof (tstr), "%c %z", localtime (&t));
   printf ("%s\n", tstr);
   if (_putenv ("TZ=EST5EDT") != -1)
   {
      _tzset ();
      strftime (tstr, sizeof (tstr), "%c %z", localtime (&t));
      printf ("%s\n", tstr);
   }
   if (_putenv ("TZ=PST8PDT") != -1)
   {
      _tzset ();
      strftime (tstr, sizeof (tstr), "%c %z", localtime (&t));
      printf ("%s\n", tstr);
   }
   return 0;
}

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Jason Rumney]

Eli Zaretskii wrote:
> I don't see on my system that TZ is only read on startup.  The
> attached simple test program demonstrates that TZ can be changed
> during the program's run with expected results, at least on my XP SP2
> box.
>   
You're right. I was testing using 3 letter timezones, as per the
Microsoft docs. But in fact it needs 3 letters + offset (+ optionally 3
letters for DST). I suspect from past experience with MS libraries
though, that the DST changeover dates will not be handled properly.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eli Zaretskii]

> Date: Sat, 20 Oct 2007 21:54:46 +0100
> From: Jason Rumney <jasonr@gnu.org>
> Cc: offby1@blarg.net, emacs-pretest-bug@gnu.org, apple@kanis.eu
> 
> I suspect from past experience with MS libraries
> though, that the DST changeover dates will not be handled properly.

Yes, this article:

  http://support.microsoft.com/kb/932955

and also the documentation of _tzset on MSDN indicate that the MS
runtime libraries follow the US rules for the DST changeover.  But
that is better than nothing, I think.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eli Zaretskii]

> Date: Sat, 20 Oct 2007 23:21:41 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: offby1@blarg.net, emacs-pretest-bug@gnu.org, apple@kanis.eu
> 
> > Date: Sat, 20 Oct 2007 21:54:46 +0100
> > From: Jason Rumney <jasonr@gnu.org>
> > Cc: offby1@blarg.net, emacs-pretest-bug@gnu.org, apple@kanis.eu
> > 
> > I suspect from past experience with MS libraries
> > though, that the DST changeover dates will not be handled properly.
> 
> Yes, this article:
> 
>   http://support.microsoft.com/kb/932955
> 
> and also the documentation of _tzset on MSDN indicate that the MS
> runtime libraries follow the US rules for the DST changeover.  But
> that is better than nothing, I think.

I've fixed time.el so that it supports both zoneinfo-style time zones
and the legacy time zones.  Please see if the change does TRT.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eric Hanchrow]

>>>>> "Eli" == Eli Zaretskii <eliz@gnu.org> writes:

    Eli> I've fixed time.el so that it supports both zoneinfo-style
    Eli> time zones and the legacy time zones.  Please see if the
    Eli> change does TRT.

Works for me.  Thanks!

-- 
He who refuses to do arithmetic is doomed to talk nonsense.
        -- John McCarthy

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Jason Rumney]

Eli Zaretskii wrote:
> I've fixed time.el so that it supports both zoneinfo-style time zones
> and the legacy time zones.  Please see if the change does TRT.
>   

The timezones work in that format, but there are a number of problems
with them.

BST0BDT should be GMT0BST
JST-9JDT should be JST-9 (no daylight savings in Japan).
I think there may also be no daylight savings in India, but I'm not sure.


Also, I think enabling the zoneinfo style names only on GNU/Linux and MS
DOS is too conservative. All BSD based systems use the zoneinfo
database, as do most SYSV ones AFAIK. It may be better to use those by
default, and enable the legacy ones on a case by case basis,
particularly since there may be problems with the legacy posix ones'
handling of DST on many platforms other than Windows (which we know
always uses US switchover dates).

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eli Zaretskii]

> Date: Sat, 27 Oct 2007 15:09:11 +0100
> From: Jason Rumney <jasonr@gnu.org>
> Cc: offby1@blarg.net, emacs-pretest-bug@gnu.org, apple@kanis.eu
> 
> BST0BDT should be GMT0BST
> JST-9JDT should be JST-9 (no daylight savings in Japan).
> I think there may also be no daylight savings in India, but I'm not sure.

Thanks, I fixed these (you were right about India).

> Also, I think enabling the zoneinfo style names only on GNU/Linux and MS
> DOS is too conservative.

I simply don't know exactly which ones support zoneinfo, and don't
have access to any other system besides those two.  The tzcode package
says that HP, Solaris, and SGI also support it, but don't provide the
details, and at least SGI Irix 6.5 didn't support zoneinfo back when I
worked on it 3 years ago, so the support could be version-specific.

Please feel free to modify the list as appropriate...

> It may be better to use those by
> default, and enable the legacy ones on a case by case basis,

...or make this change, if it's deemed better.

It strikes me that the best would be to test the support at run time,
and do TRT accordingly.  If someone knows how to test this, maybe such
a test should be part of `set-time-zone-rule', and then
`display-time-world' could use it.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Eli Zaretskii]

> From: Eric Hanchrow <offby1@blarg.net>
> Date: Sat, 27 Oct 2007 06:49:18 -0700
> Cc: emacs-devel@gnu.org
> 
> >>>>> "Eli" == Eli Zaretskii <eliz@gnu.org> writes:
> 
>     Eli> I've fixed time.el so that it supports both zoneinfo-style
>     Eli> time zones and the legacy time zones.  Please see if the
>     Eli> change does TRT.
> 
> Works for me.  Thanks!

Thanks for testing.

Re: 23.0.50; on Windows, M-x display-time-world shows the same, incorrect, time for all cities

[Jason Rumney]

Eli Zaretskii wrote:
>> Date: Sat, 27 Oct 2007 15:09:11 +0100
>> From: Jason Rumney <jasonr@gnu.org>
>>     
>> Also, I think enabling the zoneinfo style names only on GNU/Linux and MS
>> DOS is too conservative.
> It strikes me that the best would be to test the support at run time,
> and do TRT accordingly.  If someone knows how to test this, maybe such
> a test should be part of `set-time-zone-rule', and then
> `display-time-world' could use it.
>   

I added such a test to the initialization of display-time-world-list;
comparing the results of setting the timezone to "America/New York" and
"Europe/London". If these both give the same result, then we know that
zoneinfo style is not supported.