137018 - [4xp] Incorrect Time Display For Posts While Daylight Savings Time

Status: VERIFIED FIXED

(SeaMonkey :: MailNews: Message Display, defect)

Description

[Felix Miata]

2002041109 OS/2

Steps to reproduce:
1-View posts in Mozilla
2-View sames posts in Netscape 3 or 4

Actual behavior:
1-Time is correctly displayed in Netscape 3 or 4
2-Time is one hour earlier than it should in Mozilla

Expected behavior:
1-Time shows the same regardless of app used to display

Compare to bug 118899. Could fix be related?

Comment 1

[Felix Miata]

Attachment: Screenshot of Netscape & Mozilla Display of Same Message Headers

Comment 2

[Felix Miata]

CONFIG.SYS contained 'SET TZ=EST5EDT' for the original creation of the
screenshot displayed posts.

Comment 3

[Felix Miata]

I also filed bug 137037, which I'm sure is related. But, I don't think they
represent the exact same problem. When email or news is sent using Mozilla,
viewing the sent folder or the post in the newsgroup, Mozilla shows the correct
time for the posts you sent with it, but only because it is making the same
error again, which cancels the apparent effect. Because of bug 118899 effect,
you can't tell about time display or actual zone corrector errors of other posts
without using some other email/news reader.

Comment 4

[Mike Kaply [:mkaply]]

*** Bug 137037 has been marked as a duplicate of this bug. ***

Comment 5

[Mike Kaply [:mkaply]]

This is definitely a bug in VACPP handling of DST.

We have opened a bug against the compiler folks.

There is no easy workaround at this time.

Comment 6

[Mike Kaply [:mkaply]]

We found a workaround but I would like to emphasize that I am not accepting this 
as the fix. We are still pushing the compiler team.

See:

http://www2.hursley.ibm.com/goserve/$GoSer50.htm

If you specify explicitly the starting days/times for DST, it supposedly should 
work. I haven't tested this.

Comment 7

[Mike Kaply [:mkaply]]

We got the fixes DLL from the compiler team and have integrated it into all 
our builds.

Comment 8

[Felix Miata]

Verified fixed in trunk 2002053008.

Comment 9

[Martin Sch]

Same problem now with emx builds: my TZ is
ECT-1CEST,3,-1,0,7200,10,4,0,10800,3600 but mails are sent with +0000

Comment 10

[knut st. osmundsen]

>Same problem now with emx builds: my TZ is
>ECT-1CEST,3,-1,0,7200,10,4,0,10800,3600 but mails are sent with +0000
4 character timezones were broken in EMX, LIBC01 and LIBC02. 
It should be fixed in LIBC03 which we released last week. You should re-check
test problem with a build which is using LIBC03 - such a build is not yet
available and I don't know when Mike and his team will do one, but I'm kind of
hoping 1.5rc1 will be done using that LIBC and GCC release.

Comment 11

[Martin Sch]

Seems to be fixed in 1.5rc1.
WFM

Comment 12

[Marco Alberoni]

My TZ is: CET-1CDT,3,-1,0,10800,10,-1,0,7200,3600 (created with TZ Calculator).
Everything worked fine for me until 1.5b; now, with 1.5rc1, every mail sent by
me has the "Date:" field one hour forward (using every possible mail reader,
1.5rc1 included).

Comment 13

[knut st. osmundsen]

LIBC03/mozilla should work fine with this TZ. Previous releases of LIBC (ie.
LIBC02, LIBC01 & LIBC00) probably didn't.

BTW. The right one for Italy is supposed to be (according to some other Italian
guy):
  CET-1CDT,3,-1,0,7200,10,-1,0,10800,3600
You mixed the start and end time of the daylight saving.

For the other with TZ issues: There are some a bug left in the code which
prevents TZ values like PST8PDT. We're fixing it.

Comment 14

[Marco Alberoni]

LIBC04 and Mozilla 1.5 "GA"... the problem still persists...

Comment 15

[knut st. osmundsen]

We've fixed the problem in GCC 3.2.2 Beta 3 which was released a couple of days ago.
The updated LIBC can be downloaded at:
http://download.innotek.de/gccos2/runtime/libc04fix1.zip

Comment 16

[Felix Miata]

Since this vacpp bug was marked fixed over a year ago, isn't the right place for
discussing recent gcc problems bug 220850?

Comment 17

[Felix Miata]

Not fixed. Time local is 09:56 -0400.

Comment 18

[Mike Kaply [:mkaply]]

Why did you reopen this?

GCC problems are to be discussed in bug 220850

This bug was about VACPP.

Comment 19

[Felix Miata]

I found this first, with a patch, and comment 15 discussing the supposed gcc
fix. So, I didn't look for another.

Comment 20

[Mike Kaply [:mkaply]]

For posterity, I'm posting the testcases we sent to VACPP:

Test case #1

#include <time.h>
#include <stdio.h>
#include <stdlib.h>

main(int argc, char *argv[], char *envp[])
{
   time_t Secs;
   time_t Secs2;
   struct tm *LocalTime;
   struct tm *SecondsTime;

   /*_putenv("TZ=CST6CDT,4,1,0,3600,10,-1,0,7200,3600"); /*Central*/
   /*_putenv("TZ=EST5EDT,4,1,0,3600,10,-1,0,7200,3600");*/ /*Eastern*/
   /*_putenv("TZ=PST8PDT,4,1,0,3600,10,-1,0,7200,3600");*/ /*Pacific*/
   _tzset();     

   time(&Secs);
   LocalTime = localtime(&Secs);

   if (LocalTime->tm_isdst) {
      printf("DST is on\n");
   } else {
      printf("DST is off\n");
   }

   Secs2 = 1018894369;
   SecondsTime = localtime(&Secs2);
   printf("Hour = %d\n", LocalTime->tm_hour);
   printf("Hour should be 13 if we are in DST"); /*13 for CDT only*/

   return 0;
}

Test case #2

 #include <time.h>
 #include <stdio.h>
 #include <stdlib.h>
 
 int main(){
     time_t t;
     struct tm *pTime;
 
     /*_putenv("TZ=EST5EDT,4,1,03600,10,-1,0,7200,3600");*/
     _tzset();
 
     time(&t);
     pTime = localtime(&t);
 
     printf("pTime->tm_isdst (DST indicator) = %d ==> ");
     printf("%s\n", (pTime->tm_isdst > 0) ? "on" : "off");
     printf("_daylight  = %d\n",_daylight);
     printf("_timezone  = %ld\n",_timezone);
     printf("_tzname[0] = %s\n",_tzname[0]);
     printf("_tzname[0] = %s\n",_tzname[1]);
 
     return 0;
 }