[Bug libc/4033] New: mktime bug in Europe/Lisbon timezone

[Bug libc/4033] New: mktime bug in Europe/Lisbon timezone

[cvalente at spambob dot com]

Consider the following C program:

----------*time.c*------------
#include <time.h>
#include <stdio.h>

int main(int argc, char* argv[]){
        struct tm t1;
        struct tm t2;

        /* midnight 26/SET/1976*/
        t1.tm_sec  = 0;
        t1.tm_min  = 0;
        t1.tm_hour = 0;
        t1.tm_mday = 26;
        t1.tm_mon  = 8;
        t1.tm_year = 76;

        /* midnight 25/SET/1976*/
        t2.tm_sec  = 0;
        t2.tm_min  = 0;
        t2.tm_hour = 0;
        t2.tm_mday = 25;
        t2.tm_mon  = 8;
        t2.tm_year = 76;

        printf("%li\n", mktime(&t1)-mktime(&t2));
        printf("%li\n", mktime(&t1)-mktime(&t2));

        return 0;
}

--------*time.c*------------

Compile and run it to get the output:
90000
86400

That is, the same function mktime (shouldn't have any side-effects) returns 
different values the second time it's called.

My timezone is Europe/Lisbon
My system is Linux/gentoo/x86
My glibc version is 2.4-r4

This behaviour has been reproduced in other Linux distributions.
In case this bug is not related with libc, but some other distributor of 
timezone files, please indicate where I should submit the bug.

Thank you.

-- 
           Summary: mktime bug in Europe/Lisbon timezone
           Product: glibc
           Version: 2.4
            Status: NEW
          Severity: minor
          Priority: P2
         Component: libc
        AssignedTo: drepper at redhat dot com
        ReportedBy: cvalente at spambob dot com
                CC: glibc-bugs at sources dot redhat dot com
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[cvalente at spambob dot com]

------- Additional Comments From cvalente at spambob dot com  2007-02-13 14:18 -------
Created an attachment (id=1551)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1551&action=view)
Example program


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-02-13 14:24 -------
this is also visible in python 
with WET timezone 
>>> import time
>>> time.mktime((1976,9,26,0,0,0,0,0,0))
212544000.0
>>> time.mktime((1976,9,26,0,0,0,0,0,0))-time.mktime((1976,9,25,0,0,0,0,0,0))
90000.0
>>> time.mktime((1976,9,26,0,0,0,0,0,0))-time.mktime((1976,9,25,0,0,0,0,0,0))
86400.0

with NewYork time zone 
>>> import time
>>> time.mktime((1976,9,26,0,0,0,0,0,0))-time.mktime((1976,9,25,0,0,0,0,0,0))
86400.0

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sergio at sergiomb dot no-ip
                   |                            |dot org


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2007-02-17 06:43 -------
There is nothing wrong here.  This is the daylight saving time switch.  The
mktime is intelligent in that it gives an answer based on the context which is
more correct than always returning the old or new time.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WONTFIX


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-02-17 16:44 -------
(In reply to comment #3)
> There is nothing wrong here.  This is the daylight saving time switch.  The
> mktime is intelligent in that it gives an answer based on the context which is
> more correct than always returning the old or new time.

what context ? 
>>> import time
>>> time.mktime((1976,9,26,0,0,0,0,0,0))
212544000.0
>>> time.mktime((1976,9,25,0,0,0,0,0,0))
212454000.0
>>> time.mktime((1976,9,26,0,0,0,0,0,0))
212540400.0

I just want to know how I convert strtotime and after timetostr and get the same
initial string, if it a valid string date. I do this to validate the strings.

I verified that 01-jan-1970 after strtotime and timetostring give me 31-dez-1969
23.00 
which I think this is not correct !  

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-02-21 19:13 -------
don't follow the explanation 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WONTFIX                     |


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2007-02-21 19:15 -------
Stop reopening bugs.  This is no place to get an education.  There is no problem
so I close the bug.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |WORKSFORME


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-02-22 05:19 -------
ok this is a bug in bug in Europe/Lisbon timezone because if I change to
timezone of New York, you don't see it. And only when date is before 1976/9/26
I will try to prove this here so don't close the bug with see all comments .
First python
set your system timezone in WET (Western Europe Time) 

>>> import time,datetime
>>> time.mktime((1976, 9, 26, 0, 0, 0, 0, 0, 0))
212544000.0
>>> time.mktime((1976, 9, 26, 0, 0, 0, 0, 0, 1))
212540400.0
>>> time.mktime((1976, 9, 27, 0, 0, 0, 0, 0, 0))
212630400.0
>>> time.mktime((1976, 9, 27, 0, 0, 0, 0, 0, 1))
212626800.0
#until here fine
>>> time.mktime((1976, 9, 26, 0, 0, 0, 0, 0, 0))
212540400.0
>>> time.mktime((1976, 9, 26, 0, 0, 0, 0, 0, 1))
212540400.0
# same time with or without dst
>>> time.mktime((1970, 1, 1, 0, 0, 0, 0, 0, 1))
-3600.0
>>> time.mktime((1970, 1, 1, 0, 0, 0, 0, 0, 0))
-3600.0
# Zero time give me always -1 hour !
>>> time.mktime((1976, 9, 27, 0, 0, 0, 0, 0, 0))
212630400.0
>>> time.mktime((1976, 9, 27, 0, 0, 0, 0, 0, 1))
212626800.0
# 27 of Dez correct no problem again  

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-02-22 05:26 -------
Created an attachment (id=1572)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1572&action=view)
demo2 

after compile with cc time.c
./a.out with New York timezone works perfectly 
and if I change to WET time zone , the results are wrong!

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-02-22 05:29 -------
My results:
sergio@segulix:~#date
Thu Feb 22 00:03:48 EST 2007
#EST time
sergio@segulix:~#./a.out
count for 27
wi dst 212644800
wo dst 212648400
difference between -3600

count 26
wi dst 212472000
wo dst 212475600
difference between -3600
count 27
wi dst 212558400
wo dst 212562000
difference between -3600
count 28
wi dst 212644800
wo dst 212648400
difference between -3600
count 29
wi dst 212731200
wo dst 212734800
difference between -3600
wi dst 212731200
wo dst 212734800
differrence between -3600
# perfect results 

sergio@segulix:~#date
Thu Feb 22 05:06:59 WET 2007
# WET 
sergio@segulix:~#./a.out
count for 27
wi dst 212626800
wo dst 212630400
difference between -3600
# correct
count 26
wi dst 212536800
wo dst 212454000
difference between 82800
#wrong none sense 
count 27
wi dst 212626800
wo dst 212544000
difference between 82800
count 28
wi dst 212713200
wo dst 212630400
difference between 82800
count 29
wi dst 212799600
wo dst 212716800
difference between 82800
wi dst 212799600
wo dst 212716800
differrence between 82800
# all results are wrong ! 

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2007-02-22 06:06 -------
This is no bug anywhere but in your expectation.  Stop reopening.  The time you
select is the DST switch time for that timezone.  Other timezones switch at
different times.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |WORKSFORME


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-02-22 15:04 -------
this is a bug and don't close the bug until this is fixing, I think you don't
read all the things 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2007-02-22 15:17 -------
I said stop reopening.  You have no udea about the code.  None at all. So stop
jumping to conclusions.  If I say there is no bug there is nothing else here for
you to do.  You can write your own (incorrect) changes and use them. But that's
all.  There will be no change to accomodate your wrong perception in glibc.  Period.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |WORKSFORME


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[jakub at redhat dot com]

------- Additional Comments From jakub at redhat dot com  2007-02-22 15:52 -------
FYI, that time isn't actually a DST switch, but just switching the GMT offset
of the timezone for non-DST time:
Europe/Lisbon  Sun Oct  3 01:59:59 1965 UTC = Sun Oct  3 02:59:59 1965 WEST
isdst=1 gmtoff=3600
Europe/Lisbon  Sun Oct  3 02:00:00 1965 UTC = Sun Oct  3 02:00:00 1965 WET
isdst=0 gmtoff=0
Europe/Lisbon  Sun Apr  3 01:59:59 1966 UTC = Sun Apr  3 01:59:59 1966 WET
isdst=0 gmtoff=0
Europe/Lisbon  Sun Apr  3 02:00:00 1966 UTC = Sun Apr  3 03:00:00 1966 WET
isdst=0 gmtoff=3600
Europe/Lisbon  Sat Sep 25 23:59:59 1976 UTC = Sun Sep 26 00:59:59 1976 WET
isdst=0 gmtoff=3600
Europe/Lisbon  Sun Sep 26 00:00:00 1976 UTC = Sun Sep 26 00:00:00 1976 WET
isdst=0 gmtoff=0
Europe/Lisbon  Sat Mar 26 23:59:59 1977 UTC = Sat Mar 26 23:59:59 1977 WET
isdst=0 gmtoff=0
Europe/Lisbon  Sun Mar 27 00:00:00 1977 UTC = Sun Mar 27 01:00:00 1977 WEST
isdst=1 gmtoff=3600

As such, you can't rely on tm_isdst here to pick one or the other time,
Sep 26 00:30:00 1976 WET has ambiguous translation into seconds since epoch.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-03-30 14:26 -------
>As such, you can't rely on tm_isdst here to pick one or the other time,
>Sep 26 00:30:00 1976 WET has ambiguous translation into seconds since epoch.

but delta time of two hours shouldn't be ambiguous, at least, in four
differentes days, like I prove in my demo.  
This was the deepest that I can go , may be is just a problem is Portuguese time
zone.
I had around the problem using 
return calendar.timegm(gmt) + __extract_tzd(m) - time.timezone
instead of 
return time.mktime(gmt) + __extract_tzd(m) - time.timezone

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2007-03-30 15:19 -------
Stop reopening.  The behavior is exactly as designed.  If you don't like it
write your own code.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |WORKSFORME


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2007-03-30 15:47 -------
fuck Drepper this is a bug , the delta time between two days can't be 82800
seconds in my time zone ok 


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2007-03-30 15:49 -------
It is no bug whatsoever.  Stop reopening.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |WORKSFORME


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2008-02-15 17:31 -------
export TZ=WET; date
Fri Feb 15 17:20:17 WET 2008
python
>>> import time
>>> time.mktime((1970, 1, 1, 0, 0, 0, 3, 1, 0))
0.0

(right !) 

unset TZ; date (this is the usual case) 
Fri Feb 15 17:21:55 WET 2008
(see same TZ, I am in WET timezone )
python
>>> import time
>>> time.mktime((1970, 1, 1, 0, 0, 0, 3, 1, 0))
-3600.0

So can you explain when TZ is not set , PORTUGAL is one hour delayed to the rest
of the world ? 

Many sites and blogs, when date are 0 from epoch, in Portugal, appears 
31 Dec 1969 at 23 hours (1969, 12, 31, 23, 0, 0, 2, 365, 0).
when system use Linux of course ! 





-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2008-03-05 02:52 -------
ok as you write the bug could be I my expectations :

as datetest.py
import time
print time.mktime((1970, 1, 1, 0, 0, 0, 3, 1, 0))

TZ='WET' python -u datetest.py
0.0
TZ='WEST' python -u datetest.py
0.0
TZ='Europe/London' python -u datetest.py
-3600.0
TZ='Europe/Lisbon' python -u datetest.py
-3600.0
TZ='America/Caracas' python -u datetest.py
14400.0
TZ='America/Caracas' date
Tue Mar  4 22:01:33 VET 2008
TZ='VET' python -u datetest.py
0.0
TZ='Atlantic/Cape_Verde' date
Wed Mar  5 01:33:25 CVT 2008
TZ='CVT' date
Wed Mar  5 02:34:00 CVT 2008

so setting TZ with country is different to setting TZ with 3 letters of the same
timezone ?
This is the bug or a very odd feature, this is not, a result as I expect, and
not a bug specifically in my timezone ...


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/4033] mktime bug in Europe/Lisbon timezone

[sergio at sergiomb dot no-ip dot org]

------- Additional Comments From sergio at sergiomb dot no-ip dot org  2008-03-29 03:21 -------
ok , seems that 'Europe/lisbon' and WET timezone have different results with 
mktime , so others 'Americas/etc' also have different results to 3 Letters
timezones, which, for me, is clearly a bug, but glibc is consistent in is results.
So I am closing this bug because that was all about of the inconsistence of
results. Maybe I will open other bug, about the problem of "'Europe/Lisbon' and
WET timezone have different results with mktime, I really don't know if this is
the right place for do it. 
  

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |WONTFIX


http://sourceware.org/bugzilla/show_bug.cgi?id=4033

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.