bug#11807: M-x display-time-world & EDT/DST

bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

Bug confirmed by John Wiegley.  See
http://lists.gnu.org/archive/html/help-gnu-emacs/2012-06/msg00476.html


,---- M-x display-time-world
| Seattle   Wednesday 27 June 01:24 PDT
| New York  Wednesday 27 June 04:24 EDT <===  Off by 1 hr 
| London    Wednesday 27 June 09:24 BST
| Paris     Wednesday 27 June 10:24 CDT
| Bangalore Wednesday 27 June 14:54 IST <=== Local Timezone
| Tokyo     Wednesday 27 June 18:24 JST
`----

,---- http://www.timeanddate.com/worldclock/city.html?n=179
| 
| Current Time|Wednesday, 27 June 2012, 05:27:42 EDT <=== Time now
| 
| | Standard time zone:       | UTC/GMT -5 hours            |
| | Daylight saving time:     | +1 hour                     |
| | Current time zone offset: | UTC/GMT -4 hours            |
| | Time zone abbreviation:   | EDT - Eastern Daylight Time |
`----

As you can see, the time reported by Emacs is off by 1 hr.  Is there a
way to fix it?

Also I could take some help with interpreting the value of this
variable.  Should display-time-world be documented in Emacs reference
manual? 

,----[ C-h v display-time-world-list RET ]
| display-time-world-list is a variable defined in `time.el'.
| Its value is (("PST8PDT" "Seattle")
|  ("EST5EDT" "New York")
|  ("GMT0BST" "London")
|  ("CET-1CDT" "Paris")
|  ("IST-5:30" "Bangalore")  <=== Why this ''minus''. Isn't IST = GMT+5:30?
|  ("JST-9" "Tokyo"))
| 
| 
| Documentation:
| Alist of time zones and places for `display-time-world' to display.
| Each element has the form (TIMEZONE LABEL).
| TIMEZONE should be in a format supported by your system.  See the
| documentation of `zoneinfo-style-world-list' and
| `legacy-style-world-list' for two widely used formats.  LABEL is
| a string to display as the label of that TIMEZONE's time.
| 
| You can customize this variable.
| 
| This variable was introduced, or its default value was changed, in
| version 23.1 of Emacs.
| 
| [back]
`----

ps: I am really interested in local Ottawa time (which is apparently
same as NY's throughout the year)

bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

I have an actual watch (which is more reliable) to look up at time.  So
closing this request.

bug#11807: M-x display-time-world & EDT/DST

[Andreas Schwab]

Jambunathan K <kjambunathan@gmail.com> writes:

> ,---- http://www.timeanddate.com/worldclock/city.html?n=179
> | 
> | Current Time|Wednesday, 27 June 2012, 05:27:42 EDT <=== Time now

Looking at it now I see 04:54, which agrees with what display-time-world
shows.

> |  ("IST-5:30" "Bangalore")  <=== Why this ''minus''. Isn't IST = GMT+5:30?

For the same reason New York is EST5EDT.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

Andreas Schwab <schwab@linux-m68k.org> writes:

> Jambunathan K <kjambunathan@gmail.com> writes:
>
>> ,---- http://www.timeanddate.com/worldclock/city.html?n=179
>> | 
>> | Current Time|Wednesday, 27 June 2012, 05:27:42 EDT <=== Time now
>
> Looking at it now I see 04:54, which agrees with what display-time-world
> shows.

May be it has something to do with locale settings.  

This is what I see *now.

Seattle   Thursday 28 June 00:56 PDT
New York  Thursday 28 June 03:56 EDT <=== This is off by 1-hr.
London    Thursday 28 June 08:56 BST
Paris     Thursday 28 June 09:56 CDT
Bangalore Thursday 28 June 14:26 IST <=== Local time.  This is correct
Tokyo     Thursday 28 June 17:56 JST

>> |  ("IST-5:30" "Bangalore")  <=== Why this ''minus''. Isn't IST = GMT+5:30?
>
> For the same reason New York is EST5EDT.

It is consistent but counter-intuitive. It is a gotcha.

> Andreas.

bug#11807: M-x display-time-world & EDT/DST

[Andreas Schwab]

Jambunathan K <kjambunathan@gmail.com> writes:

> This is what I see *now.
>
> Seattle   Thursday 28 June 00:56 PDT
> New York  Thursday 28 June 03:56 EDT <=== This is off by 1-hr.
> London    Thursday 28 June 08:56 BST
> Paris     Thursday 28 June 09:56 CDT
> Bangalore Thursday 28 June 14:26 IST <=== Local time.  This is correct
> Tokyo     Thursday 28 June 17:56 JST

You are using the wrong timezone (GMT+6:30 instead of GMT+5:30).

>>> |  ("IST-5:30" "Bangalore")  <=== Why this ''minus''. Isn't IST = GMT+5:30?
>>
>> For the same reason New York is EST5EDT.
>
> It is consistent but counter-intuitive. It is a gotcha.

It is a fact of life and perfectly correct.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

Andreas Schwab <schwab@linux-m68k.org> writes:

> Jambunathan K <kjambunathan@gmail.com> writes:
>
>> This is what I see *now.
>>
>> Seattle   Thursday 28 June 00:56 PDT
>> New York  Thursday 28 June 03:56 EDT <=== This is off by 1-hr.
>> London    Thursday 28 June 08:56 BST
>> Paris     Thursday 28 June 09:56 CDT
>> Bangalore Thursday 28 June 14:26 IST <=== Local time.  This is correct
>> Tokyo     Thursday 28 June 17:56 JST
>
> You are using the wrong timezone (GMT+6:30 instead of GMT+5:30).

Let me see:

M-: (getenv "TZ")            
  => nil

M-: (current-time-zone)      
  => (19800 "India Standard Time")

M-: (format-time-string display-time-world-time-format nil)
  => "Thursday 28 June 15:23 India Standard Time"

Here is the math for this time of the year.

IST = GMT + 5:30
NYT = GMT - 4:00

NYT = IST - 5:30 - 4:00 = IST - 9:30

and I expect to see 6:07EDT instead I am seeing 04:55 instead (adjust
for typing minutes)

M-: (setenv "TZ" "EST5EDT")
M-: (format-time-string display-time-world-time-format nil)
 => "Thursday 28 June 04:55 EDT"

bug#11807: M-x display-time-world & EDT/DST

[Andreas Schwab]

What do you get from (format-time-string "%H:%M" nil t)?

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

Andreas Schwab <schwab@linux-m68k.org> writes:

> What do you get from (format-time-string "%H:%M" nil t)?

=> "10:50"

M-: (format-time-string "%Y-%m-%dT%H:%M:%S%z")
  => "2012-06-28T16:19:18+0530"

> Andreas.

bug#11807: M-x display-time-world & EDT/DST

[Andreas Schwab]

What about (progn (require 'cal-dst) (calendar-dst-find-data))?

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

Andreas Schwab <schwab@linux-m68k.org> writes:

> What about (progn (require 'cal-dst) (calendar-dst-find-data))?

(330 0 "India Standard Time" "India Standard Time" nil nil 0 0)


> Andreas.

bug#11807: M-x display-time-world & EDT/DST

[Andreas Schwab]

What about (progn (setenv "TZ" "CET-1CEST") (calendar-dst-find-data))?

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

> What about (progn (setenv "TZ" "CET-1CEST") (calendar-dst-find-data))?

(60 0 "CES" "CES" nil nil 0 0)

bug#11807: M-x display-time-world & EDT/DST

[Andreas Schwab]

Jambunathan K <kjambunathan@gmail.com> writes:

>> What about (progn (setenv "TZ" "CET-1CEST") (calendar-dst-find-data))?
>
> (60 0 "CES" "CES" nil nil 0 0)

That's your problem.  You have a faulty timezone implementation.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#11807: (Windows XP issue?) Re: bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

Andreas Schwab <schwab@linux-m68k.org> writes:

> Jambunathan K <kjambunathan@gmail.com> writes:
>
>>> What about (progn (setenv "TZ" "CET-1CEST") (calendar-dst-find-data))?
>>
>> (60 0 "CES" "CES" nil nil 0 0)
>
> That's your problem.  You have a faulty timezone implementation.

Let me replay the same questions and collect data from your system.

What do the below forms report for you.  I assume the TZ works right on
your system.

(format-time-string "%H:%M" nil t)

(progn (require 'cal-dst) (calendar-dst-find-data))

(progn (setenv "TZ" "CET-1CEST") (calendar-dst-find-data))

With my machine re-configured to use a remote timezone (as below)

,---- (Windows XP/Cygwin) Control Panel->"Date and Time" 
| 
|  (GMT-5:00) Eastern Time (US and Canada) and 
|  Automatically adjust clock for daylight saving changes turned ON
| 
|  I also have my "Internet Time" automatically synchronize with
|  time.windows.com time server.
`----

I see that the correct remote time is reported.

It is difficult to digest an argument that a widely used enterprise
system such as Windows XP has an obviously buggy implementation.  It is
quite possible that Emacs is using the platform APIs in wrong manner.

What library call (dll library) that Emacs uses to fetch time info?  

bug#11807: (Windows XP issue?) Re: bug#11807: M-x display-time-world & EDT/DST

[Eli Zaretskii]

> From: Jambunathan K <kjambunathan@gmail.com>
> Date: Fri, 29 Jun 2012 14:56:10 +0530
> Cc: 11807@debbugs.gnu.org
> 
> Andreas Schwab <schwab@linux-m68k.org> writes:
> 
> > Jambunathan K <kjambunathan@gmail.com> writes:
> >
> >>> What about (progn (setenv "TZ" "CET-1CEST") (calendar-dst-find-data))?
> >>
> >> (60 0 "CES" "CES" nil nil 0 0)
> >
> > That's your problem.  You have a faulty timezone implementation.

I don't think the MS-Windows version of tzset supports TZ values such
as CET-1CEST.  The time-zone strings it supports need to be 3-letter,
not 4-letter.  See this thread for more:

  http://lists.gnu.org/archive/html/emacs-devel/2007-10/msg00991.html

Since the original report didn't mention CET-1CEST, the results above
might not be relevant to the original problem at all.

FWIW, times displayed by display-time-world on my XP SP3 box coincide
with what http://www.timeanddate.com/worldclock/city.html?n=179 shows
for the respective cities.  This is with today's trunk.  So maybe your
system's MSVCRT.DLL needs to be updated wrt the timezone information
(see below).

> With my machine re-configured to use a remote timezone (as below)
> 
> ,---- (Windows XP/Cygwin) Control Panel->"Date and Time" 
> | 
> |  (GMT-5:00) Eastern Time (US and Canada) and 
> |  Automatically adjust clock for daylight saving changes turned ON
> | 
> |  I also have my "Internet Time" automatically synchronize with
> |  time.windows.com time server.
> `----

In what build of Emacs was the original problem detected?  What is the
value of 'system-configuration' and 'system-configuration-options' in
that build?  And which Emacs version is that?  You didn't say all that
in your original report, AFAICT.

If the problem is not in the Windows native (MinGW or MSVC) build of
Emacs, then all I write here is irrelevant.

> It is difficult to digest an argument that a widely used enterprise
> system such as Windows XP has an obviously buggy implementation.  It is
> quite possible that Emacs is using the platform APIs in wrong manner.
> 
> What library call (dll library) that Emacs uses to fetch time info?  

If you are using a Cygwin build of Emacs, the information comes from
newlib, and you should ask this question on the Cygwin list; disregard
all I said in that case.

If you are using the native Windows build, then the information comes
from MSVCRT.DLL, the Microsoft C runtime library.  It is known that
_tzset and related time functions in MSVCRT implement only a subset of
possible time-zone names, and they do the time-zone calculations in a
simplistic manner, separately from a much more complete implementation
in the Win32 APIs, such as GetTimeZoneInformation, which is used by
the OS itself for displaying local time.  There's an entry in
etc/PROBLEMS about that, search for "timezone".  In addition,
depending on the system updates installed on your machine, it is
possible that the latest updates for MSVCRT need to be installed to
account for the change in US timezones a few years ago, when they
switched from fixed dates for DST change to rule-based dates.

Volunteers are welcome to write a custom implementation of _tzset that
uses those APIs.

bug#11807: (Windows XP issue?) Re: bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

Eli Zaretskii <eliz@gnu.org> writes:

> In what build of Emacs was the original problem detected?  What is the
> value of 'system-configuration' and 'system-configuration-options' in
> that build?  And which Emacs version is that?  You didn't say all that
> in your original report, AFAICT.
>
> If the problem is not in the Windows native (MinGW or MSVC) build of
> Emacs, then all I write here is irrelevant.

Sorry about this.  I didn't anticipate that this problem will turn out
to be platform-specific.  

Here is what report-emacs-bug generates.

ps: I haven't looked at the contents of your and other referenced posts
yet.

In GNU Emacs 24.1.1 (i386-mingw-nt5.1.2600)
 of 2012-06-02 on MARVIN
Windowing system distributor `Microsoft Corp.', version 5.1.2600
Configured using:
 `configure --with-gcc (4.6) --no-opt --enable-checking --cflags
 -ID:/devel/emacs/libs/libXpm-3.5.8/include
 -ID:/devel/emacs/libs/libXpm-3.5.8/src
 -ID:/devel/emacs/libs/libpng-dev_1.4.3-1/include
 -ID:/devel/emacs/libs/zlib-dev_1.2.5-2/include
 -ID:/devel/emacs/libs/giflib-4.1.4-1/include
 -ID:/devel/emacs/libs/jpeg-6b-4/include
 -ID:/devel/emacs/libs/tiff-3.8.2-1/include
 -ID:/devel/emacs/libs/gnutls-3.0.9/include'

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: ENG
  value of $XMODIFIERS: nil
  locale-coding-system: cp1252
  default enable-multibyte-characters: t

bug#11807: (Windows XP issue?) Re: bug#11807: M-x display-time-world & EDT/DST

[Jason Rumney]

Eli Zaretskii <eliz@gnu.org> writes:

> I don't think the MS-Windows version of tzset supports TZ values such
> as CET-1CEST.  The time-zone strings it supports need to be 3-letter,
> not 4-letter.  See this thread for more:

That may be an issue for that particular timezone string, but I think
the issue is deeper than that. From my recollection, Windows always
applies the timezone rules configured in the system control panel,
it only partially supports TZ environment variables for overriding
those.  More precisely, Windows itself does not support the TZ variable
at all, the C runtime library does, but retains calls into the Windows
system library for information that is missing from the TZ variable such
as DST start/end dates (which do not exist for K Jambunathan's locale,
so DST is never applied on his machine).

bug#11807: (Windows XP issue?) Re: bug#11807: M-x display-time-world & EDT/DST

[Jambunathan K]

Jason Rumney <jasonr@gnu.org> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> I don't think the MS-Windows version of tzset supports TZ values such
>> as CET-1CEST.  The time-zone strings it supports need to be 3-letter,
>> not 4-letter.  See this thread for more:
>
> That may be an issue for that particular timezone string, but I think
> the issue is deeper than that. From my recollection, Windows always
> applies the timezone rules configured in the system control panel,
> it only partially supports TZ environment variables for overriding
> those.  More precisely, Windows itself does not support the TZ variable
> at all, the C runtime library does, but retains calls into the Windows
> system library for information that is missing from the TZ variable such
> as DST start/end dates (which do not exist for K Jambunathan's locale,
> so DST is never applied on his machine).

Till such time as a right fix is found, the command has to be either
selectively disabled or a warning issued to the user saying that some
sort of DST adjustment might have to be made by the user himself.

Having a command report incorrect values is not a good idea.

bug#11807: (Windows XP issue?) Re: bug#11807: M-x display-time-world & EDT/DST

[Eli Zaretskii]

> From: Jambunathan K <kjambunathan@gmail.com>
> Cc: Eli Zaretskii <eliz@gnu.org>,  schwab@linux-m68k.org,  11807@debbugs.gnu.org
> Date: Sat, 30 Jun 2012 11:13:08 +0530
> 
> Till such time as a right fix is found, the command has to be either
> selectively disabled or a warning issued to the user saying that some
> sort of DST adjustment might have to be made by the user himself.

It works fine on my XP system.  If you can find out why it doesn't on
yours, or confirm Jason's suspicion that it misbehaves on any Windows
box set for a timezone without DST rules, please do.  Failing that,
you are free not to use that command.  It's a toy command, anyway,
IMO.

But disabling a command that works for many users sounds too drastic
to me.

> Having a command report incorrect values is not a good idea.

Patches are more than welcome.  Thanks in advance.