Martin Simmons wrote:
>>>>>> On Wed, 22 Aug 2007 14:07:26 +0200, Kern Sibbald said:
>> On Wednesday 22 August 2007 13:58, Martin Simmons wrote:
>>>>>>>> On Tue, 21 Aug 2007 21:07:49 +0200, Kern Sibbald said:
>>>> Hello again,
>>>>
>>>> By the way, in the bug report, please show the date generated by both the
>>>> mail (or Mail) and bsmtp commands. On my machine the dates seem to be
>>>> correct. Perhaps postfixing (CEST) confuses someone's email program as
>>>> that is the only difference between mail and bsmtp (bsmtp has an extra
>>>> (CEST) added).
>>> Try setting TZ=EST5EDT. That should put -0400 as the offset, but on Linux
>>> I always get the local offset including dst.
>>>
>>> Also, on FreeBSD I always get +0000 regardless of TZ (probably the local
>>> offset without dst).
>>>
>>> The basic problem is that gettimeofday() doesn't set the tz_minuteswest
>>> from the TZ variable.
>> If that is the case (tz_minuteswest not set), then the bug would seem to be
>> in
>> your OS' gettimeofday() implementation.
>
> Mine and yours (if you are running Linux) :-)
>
> BSD-like OS's store this info in the tm.tm_gmtoff field set by localtime().
>
> I think the tz_minuteswest was always historically set to the "kernel"
> timezone anyway, because gettimeofday() is a system call. The complexity of
> TZ and the zoneinfo databases is only handled in the userspace by functions
> such as localtime().
>
> __Martin
The same is true for solaris It is reflected in the prototype
int gettimeofday(struct timeval *tp, void *tzp);
Looking at the different timezone handling among Solaris, FreeBSD and Linux
i would say the most portable solution for getting the timezone offset is
along the lines of the code below:
#include <stdio.h>
#include <time.h>
/* Return the offset west from localtime to UTC in minutes
* Same as timezone.tz_minuteswest
*/
int tz_offset ()
{
time_t now = time (NULL);
struct tm tm_utc, tm_lcl;
(void) localtime_r (&now, &tm_lcl);
(void) gmtime_r (&now, &tm_utc);
tm_utc.tm_isdst = tm_lcl.tm_isdst;
return difftime (mktime (&tm_utc), now) / 60;
}
This code is tested, one should add proper error handling though.
Attila
>
>> Can you send me a copy of the FreeBSD gettimeofday() man page?
>>
>>> __Martin
>>>
>>>> On Tuesday 21 August 2007 19:59, Dan Langille wrote:
>>>>> FYI, a proposed solution for the date problem.
>>>>>
>>>>> ------- Forwarded message follows -------
>>>>> From: "Dan Langille" <[EMAIL PROTECTED]>
>>>>> To: Ryan Novosielski <[EMAIL PROTECTED]>
>>>>> Date sent: Tue, 21 Aug 2007 13:55:03 -0400
>>>>> Priority: normal
>>>>> Copies to: [EMAIL PROTECTED]
>>>>> Subject: Re: [Bacula-users] wrong date in mail report with
>>>>> 2.2.0
>>>>>
>>>>> [ Double-click this line for list subscription options ]
>>>>>
>>>>> On 21 Aug 2007 at 12:47, Ryan Novosielski wrote:
>>>>>> Dan Langille wrote:
>>>>>>> On 20 Aug 2007 at 0:40, Ryan Novosielski wrote:
>>>>>>>> Dan Langille wrote:
>>>>>>>>> On 18 Aug 2007 at 11:00, Ryan Novosielski wrote:
>>>>>>>>>> Dan Langille wrote:
>>>>>>>>>>> On 18 Aug 2007 at 2:00, Per olof Ljungmark wrote:
>>>>>>>>>>>> Since upgrading to 2.2.0 bsmtp sets the wrong Date: header in
>>>>>>>>>>>> the mail: Date: Sat, 18 Aug 2007 01:48:56 +0000 (CEST)
>>>>>>>>>>>>
>>>>>>>>>>>> For the recipient the above message appears to have been
>>>>>>>>>>>> receieved at 03:48, two hours ahead.
>>>>>>>>>>>>
>>>>>>>>>>>> The correct timezone is UTC +2, the above header should have
>>>>>>>>>>>> read Date: Sat, 18 Aug 2007 01:48:56 +0200 (CEST)
>>>>>>>>>>>> ^
>>>>>>>>>>>> Again, this might be FreeBSD specific just as my last report.
>>>>>>>>>>> This just in (at 23:22 local time):
>>>>>>>>>>>
>>>>>>>>>>> Date: Fri, 17 Aug 2007 23:19:12 +0000 (EDT)
>>>>>>>>>>>
>>>>>>>>>>> Compare that to an email generated via: echo 'test' | mail dan
>>>>>>>>>>>
>>>>>>>>>>> Fri, 17 Aug 2007 23:21:39 -0400 (EDT)
>>>>>>>>>>>
>>>>>>>>>>> It's getting the timezone wrong. In my case, it should be
>>>>>>>>>>> -0400. I won't have time to look at this until Saturday
>>>>>>>>>>> afternoon.
>>>>>>>>>> I'd like to chime in that it is NOT just FreeBSD, but at least
>>>>>>>>>> also Solaris. I've had this problem with Solaris and Horde/IMP
>>>>>>>>>> before as well, so I never reported it on the Bacula list
>>>>>>>>>> figuring it was really not their problem. My Bacula director is
>>>>>>>>>> Solaris 9. I don't know how 10 would change this.
>>>>>>>>> Good. Good to know. Thank you.
>>>>>>>>>
>>>>>>>>> Perhaps it was this commit:
>>>>>>>>>
>>>>>>>>> http://bacula.svn.sourceforge.net/viewvc/bacula/trunk/bacula/src/
>>>>>>>>> tool s /bsmtp.c?view=diff&r1=4789&r2=479
>>>>>>>>>
>>>>>>>>> If you look at older versions of bsmtp, and I think that's where
>>>>>>>>> the problem is, you see this (look around line 353):
>>>>>>>>>
>>>>>>>>> http://bacula.svn.sourceforge.net/viewvc/bacula/trunk/bacula/src/
>>>>>>>>> tool s /bsmtp.c?view=diff&r1=2835&r2=283
>>>>>>>>>
>>>>>>>>> The code used to do:
>>>>>>>>>
>>>>>>>>> strftime(buf, sizeof(buf), "%a, %d %b %Y %H:%M:%S %z", &tm)
>>>>>>>>>
>>>>>>>>> Where the %z (time zone offset, e.g. -04) is the time zone. For
>>>>>>>>> some reason, this is now done as a different step (see the first
>>>>>>>>> URL) and uses %Z (time zone name, eg. EST)
>>>>>>>>>
>>>>>>>>> This patch fixes the problem for me, but I suspect there were
>>>>>>>>> some special issues attempted by the original code that my fix
>>>>>>>>> does not address.
>>>>>>>>>
>>>>>>>>> --- src/tools/bsmtp.c~ Sat Aug 18 16:33:26 2007
>>>>>>>>> +++ src/tools/bsmtp.c Sat Aug 18 16:33:26 2007
>>>>>>>>> @@ -194,11 +194,7 @@
>>>>>>>>> gettimeofday(&tv, &tz);
>>>>>>>>> my_timezone = tz.tz_minuteswest; /* timezone offset in mins
>>>>>>>>> */ #endif
>>>>>>>>> - strftime(buf, buf_len, "%a, %d %b %Y %H:%M:%S", &tm);
>>>>>>>>> - sprintf(tzbuf, " %+2.2ld%2.2u", -my_timezone / 60,
>>>>>>>>> abs(my_timezone) % 60);
>>>>>>>>> - strcat(buf, tzbuf); /* add +0100 */
>>>>>>>>> - strftime(tzbuf, sizeof(tzbuf), " (%Z)", &tm);
>>>>>>>>> - strcat(buf, tzbuf); /* add (CEST) */
>>>>>>>>> + strftime(buf, buf_len, "%a, %d %b %Y %H:%M:%S %z %Z", &tm);
>>>>>>>>> }
>>>>>>>> I haven't had a lot of time to look this over, but I think that
>>>>>>>> the date in my headers from bstmp will be interesting to you and
>>>>>>>> others who may be having this problem and have had more time to
>>>>>>>> look at it:
>>>>>>>>
>>>>>>>> Date: Sun, 08 Jul 2007 23:57:00 +0000 (%z)
>>>>>>>>
>>>>>>>> ...that don't look right to me. :) I suspect %z and %Z may not be
>>>>>>>> handled properly on non-Linux? Is that even possible? I thought
>>>>>>>> most of this stuff kinda stuck to POSIX for this kind of thing,
>>>>>>>> whatever it might have to say about this. <shrug>
>>>>>>> Ryan: please check man strftime and compare to what I found here
>>>>>>> for %z: http://bama.ua.edu/cgi-bin/man-cgi?00+00
>>>>>>>
>>>>>>> In short, %Z is supported on Solaris, but not %z.
>>>>>> Changing this %z to %Z (actually, on 2.0.3 I think it was changing
>>>>>> '%Z (%z)' to just %Z) rectifies the problem on Solaris 9, which is
>>>>>> nice, because I was really sick of doing the math every time my
>>>>>> backups finished.
>>>>> I am tempted to change the code to use %Z. It seems to be more
>>>>> portable and exists on all the man strftime pages that I've checked.
>>>>>
>>>>> Either that or add an autoconf test. Which I don't know how to do.
>>>>> If someone does that work, I'll be sure to add it in.
>>>>>
>>>>> If no objections, I'll do the %Z patch tomorrow. And, FYI, I'll
>>>>> patch the 2.2.0 code in the Bacula port.
>>>>>
>>>>> --
>>>>> Dan Langille - http://www.langille.org/
>>>>> Available for hire: http://www.freebsddiary.org/dan_langille.php
>>>>>
>>>>>
>>>>>
>>>>> ----------------------------------------------------------------------
>>>>> ---
>>>>> This SF.net email is sponsored by: Splunk Inc.
>>>>> Still grepping through log files to find problems? Stop.
>>>>> Now Search log events and configuration files using AJAX and a
>>>>> browser.
>>>>> Download your FREE copy of Splunk now >> http://get.splunk.com/
>>>>> _______________________________________________
>>>>> Bacula-users mailing list
>>>>> [EMAIL PROTECTED]
>>>>> https://lists.sourceforge.net/lists/listinfo/bacula-users
>>>>>
>>>>>
>>>>> ------- End of forwarded message -------
>>>> -------------------------------------------------------------------------
>>>> This SF.net email is sponsored by: Splunk Inc.
>>>> Still grepping through log files to find problems? Stop.
>>>> Now Search log events and configuration files using AJAX and a browser.
>>>> Download your FREE copy of Splunk now >> http://get.splunk.com/
>>>> _______________________________________________
>>>> Bacula-devel mailing list
>>>> [email protected]
>>>> https://lists.sourceforge.net/lists/listinfo/bacula-devel
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems? Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >> http://get.splunk.com/
> _______________________________________________
> Bacula-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/bacula-devel
>
>
-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bacula-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bacula-devel
