Message-ID: <4CCC12E7.5060807@cornell.edu>
Date: Sat, 30 Oct 2010 08:43:19 -0400
From: Ken Brown <kbrown@cornell.edu>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2.11) Gecko/20101013 Thunderbird/3.1.5
MIME-Version: 1.0
To: cygwin@cygwin.com
Subject: Re: localtime and TZ
References: <4CCB4025.6070405@cornell.edu> <4CCB42AF.2090907@redhat.com> <4CCB439B.7060602@redhat.com> <4CCB4676.8080609@cornell.edu> <4CCB47A6.9010600@redhat.com> <4CCB561D.5070304@cornell.edu>
In-Reply-To: <4CCB561D.5070304@cornell.edu>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Mailing-List: contact cygwin-help@cygwin.com; run by ezmlm
Sender: cygwin-owner@cygwin.com
Mail-Followup-To: cygwin@cygwin.com

On 10/29/2010 7:17 PM, Ken Brown wrote:
> On 10/29/2010 6:16 PM, Eric Blake wrote:
>> On 10/29/2010 04:11 PM, Ken Brown wrote:
>>>
>>> Thanks, Eric.  I didn't know about any of this.  (I was using a modification of a configure test from the emacs sources.)
>>
>> Probably worth pointing it out to the emacs upstream, then :)
>>
>>>    But I get the same behavior with the following revised test case:
>>>
>>> #include<time.h>
>>> #include<stdio.h>
>>>
>>> int
>>> main (void)
>>> {
>>>     time_t now = time ((time_t *) 0);
>>>     printf ("TZ is initially unset; hour = %d\n", localtime (&now)->tm_hour);
>>>     putenv ("TZ=GMT0");
>>>     printf ("TZ=GMT0; hour = %d\n", localtime (&now)->tm_hour);
>>>     unsetenv("TZ");
>>>     printf ("TZ unset; hour = %d\n", localtime (&now)->tm_hour);
>>>     putenv ("TZ=PST8");
>>>     printf ("TZ=PST8; hour = %d\n", localtime (&now)->tm_hour);
>>>     unsetenv("TZ");
>>>     printf ("TZ unset again; hour = %d\n", localtime (&now)->tm_hour);
>>> }
>>>
>>> So the question remains whether this difference between Cygwin and Linux is a bug or by design.
>>
>> Apparently by design.  POSIX requires:
>>
>> http://www.opengroup.org/onlinepubs/9699919799/functions/localtime.html
>>
>> Local timezone information is used as though localtime() calls tzset().
>>
>> http://www.opengroup.org/onlinepubs/9699919799/functions/tzset.html
>>
>> The tzset() function shall use the value of the environment variable TZ
>> to set time conversion information used by ctime , localtime , mktime ,
>> and strftime . If TZ is absent from the environment,
>> implementation-defined default timezone information shall be used.
>
> Wouldn't you interpret this as meaning that the implementation-defined
> default timezone information should be the same every time localtime is
> called with TZ unset?  If not, what should a program do to get the
> "standard" default timezone information that it would get if TZ had
> never been set in the first place?

I've looked at Cygwin's localtime.cc, and the behavior I'm complaining 
about is caused by the following code at the beginning of tzset:

	const char *	name = getenv("TZ");

	if (name == NULL) {
		if (!lcl_is_set)
			tzsetwall();
		goto out;
	}

So getting rid of 'if (!lcl_is_set)' would solve the problem.  But this 
would be inefficient, because it would mean that tzsetwall gets called 
every time tzset is called if TZ is never set.  To get around that, one 
could have tzsetwall set TZ.

It seems that tzset and tzsetwall used to behave the way I'm proposing 
before the following two changes were made:

2007-12-11  Corinna Vinschen  <corinna@vinschen.de>

	* localtime.cc (tzset): Call tzsetwall only if it hasn't been
	called before.

2007-08-01  Corinna Vinschen  <corinna@vinschen.de>

	* localtime.cc (tzsetwall): Don't set TZ.

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple