October 2010 - tz - lists.iana.org

FW: Samoa
by Olson, Arthur David (NIH/NCI) [E] Oct. 13, 2010

Oct. 13, 2010

I'm forwarding this message from Raymond Hughes, who is not on the time zone mailing list. Those of you who are on the list, please direct replies appropriately. (Right now http://www.mcil.gov.ws is unavailable--at least from the system I'm using--but Googling "03/04/11 adjust clocks" and using the "Cache" link provides the content). --ado -----Original Message----- From: Raymond Hughes [mailto:raymond@hughes.ws] Sent: Thursday, October 07, 2010 9:36 To: tz(a)lecserver.nci.nih.gov Subject: Re: Samoa the start of DST here is Samoa was OK from what I've seen, but looking at my zdump I see the ending does not have the correct times. Sun Sep 26 10:59:59 2010 UTC = Sat Sep 25 23:59:59 2010 WST isdst=0 Sun Sep 26 11:00:00 2010 UTC = Sun Sep 26 01:00:00 2010 WSDT isdst=1 Sun Apr 3 09:59:59 2011 UTC = Sat Apr 2 23:59:59 2011 WSDT isdst=1 Sun Apr 3 10:00:00 2011 UTC = Sat Apr 2 23:00:00 2011 WST isdst=0 Please see www.mcil.gov.ws, the Ministry of Commerce, Industry and Labour (sideframe) "Last Sunday September 2010 (26/09/10) - adjust clocks forward from 12:00 midnight to 01:00am and First Sunday April 2011 (03/04/11) - adjust clocks backwards from 1:00am to 12:00am" PS. Thanks Maintainers Regards Ray

3 3

Re Fiji Timezone
by Franck Martin Oct. 13, 2010

Oct. 13, 2010

Seems the TZ file is not updated with the info: http://www.fiji.gov.fj/index.php?option=com_content&view=article&id=2438:da… Fiji will switch back to daylight saving time from 24th October, 2010 at 2am. This means members of the public will be required to switch their clocks one hour ahead to 3am. It is expected to end in March 2011, a date is yet to be confirmed. Franck Martin http://www.avonsys.com/ http://www.facebook.com/Avonsys twitter: FranckMartin Avonsys Check your domain reputation: http://gurl.im/b69d4o

1 0

Australia has changed the first day of daylight?
by Ahmed Hassan (ahmhassa) Oct. 13, 2010

Oct. 13, 2010

Dear, Since last year, Australia has changed the first day of daylight saving to the first Sunday of October, and ending first Sunday of April. Here is the URL for your reference for states specific changes: http://www.bom.gov.au/climate/averages/tables/dst_times.shtml, which version does capture this change? Best regards, Ahmed

1 0

A question about Amsterdam in 1900
by David Patte Oct. 10, 2010

Oct. 10, 2010

I was reviewing some historical zones & rules, and I can across a date for Amsterdam that has me a bit confused. I'm not sure if I'm misinterpreting the records, or whether a record is missing. Lets take Jan 1, 1900 for Amsterdam. The Zone table for Europe/Amsterdam shows: a GMT offset of 0:19:32 (and no rule) until 1835, then a GMT offset of 0:19:32 (and Neth rule) until July 1 1937 - using '%s' for letters So for Jan 1, 1900 I am assuming I should be looking for a rule in the Neth rules records But the first 'Neth' Rule starts 1916. I can easily easily understand that this means that there is no special save amount to apply on Jan 1st 1900, but what letters should I be using for that date given that the zone says the letters should be %s? Since the zone record says '%s' for that date, should there not be a rule record starting in 1835 that gives a save amount of 0, and provides the letters 'AMT'? Thanks

2 1

code question
by Zoidiasoft Tech Oct. 9, 2010

Oct. 9, 2010

Hi, I am a Delphi programmer and am working on a time zone problem. In the Olson database I'm having trouble differentiating between times that are strictly standard and those that include summer time / daylight savings. Is there a standard in the code for differentiating these? Most of it is fairly simple to understand but I am mystified by some of the abbreviations in the rule changes. For instance: # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Finland 1942 only - Apr 3 0:00 1:00 S Rule Finland 1942 only - Oct 3 0:00 0 - Rule Finland 1981 1982 - Mar lastSun 2:00 1:00 S Rule Finland 1981 1982 - Sep lastSun 3:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Helsinki 1:39:52 - LMT 1878 May 31 1:39:52 - HMT 1921 May # Helsinki Mean Time 2:00 Finland EE%sT 1983 2:00 EU EE%sT What is the % symbol I keep seeing in the code for? For instance, in "2:00 EU EE%sT" is "EE%sT" Eastern European Standard Time? I take this to mean that Finland is currently on EU rules with a UTC offset of +2. Every now and then I also find something like this: "2:00 1:00 EEST 1989 Sep 24 2:00s" Does this line mean that I add 1 hour to the 2:00 hours UTC? Or is this correction already included in the 2 hours? Thanks! Sincerely, Curtis Manwaring

1 0

Samoa
by Olson, Arthur David (NIH/NCI) [E] Oct. 8, 2010

Oct. 8, 2010

Can anyone with boots on the ground tell us what actually happened with clock changes over the weekend in Samoa? Thanks. --ado

3 3

Valgrind error "Conditional jump or move depends on uninitialised value" on 64-bit
by Ravindra Oct. 5, 2010

Oct. 5, 2010

[Sending this directly to you because POSTing it on website does not seem to be working for me. This is about reproducing valgrind issue and a compilation warning.] Arthur and Paul, I'm seeing two problems in localtime.c: 1. When I compile localtime.c using gcc 4.1.2 for 32-bit, I get following warning: $ gcc -m32 -g -c -o localtime.o localtime.c localtime.c: In function âdiffer_by_repeatâ: localtime.c:326: warning: comparison is always false due to limited range of data type 2. valgrind reports following error for some timezones (only when localtime.c is compiled as 64-bit binary, it does not happen when compiled as 32-bit binary): ==26671== Conditional jump or move depends on uninitialised value(s) ==26671== at 0x401CDA: typesequiv (localtime.c:588) ==26671== by 0x401B85: tzload (localtime.c:562) ==26671== by 0x405553: tz_validate (localtime.c:2017) ==26671== by 0x40560D: main (localtime.c:2040) One can reproduce it using following valgrind command and patch: valgrind -v --tool=memcheck --leak-check=full --leak-resolution=med --show-reachable=yes --error-limit=no --workaround-gcc296-bugs=yes --run-libc-freeres=no --log-file=valgrind-tzvalid ./tzvalid /<path/to/my/zoneinfo/EST5EDT> NOTE: I can consistently reproduce it for EST5EDT timezone and I don't get this error for GMT. Patch for localtime.c for reproducing above error: --- org/localtime.c 2010-08-16 16:27:05.000000000 +0530 +++ localtime.c 2010-10-01 17:34:45.999089000 +0530 @@ -2004,6 +2004,47 @@ #endif /* defined CMUCS */ +static int +tz_validate(const char * name) +{ + struct state state = {0}; + time_t now; + struct tm tm; + + time(&now); + tzload(gmt, &state, TRUE); + if (0 == tzload(name, &state, TRUE)) { + localsub(&now, 0L, &tm); + return TRUE; + } + if (0 == tzparse(name, &state, FALSE)) { + localsub(&now, 0L, &tm); + return TRUE; + } + return FALSE; +} + +/* + * Compile it simply as "$gcc -m64 -g -o tzvalid localtime.c" + */ +int main(int argc, char ** argv) { + + if (argc != 2) { + printf("Usage: ./tzvalid <timezone_file_path>\n"); + return 2; + } + + char * tz = argv[1]; + + if (tz_validate(tz)) { + printf("timezone %s is valid\n", tz); + return 0; + } else { + printf("timezone %s is not valid\n", tz); + return 1; + } +} + /* ** XXX--is the below the right way to conditionalize?? */ Thanks in advance, Ravindra

1 0

Valgrind error "Conditional jump or move depends on uninitialised value" on 64-bit
by Ravindra Oct. 5, 2010

Oct. 5, 2010

I'm seeing two problems in localtime.c: 1. When I compile localtime.c using gcc 4.1.2 for 32-bit, I get following warning: $ gcc -m32 -g -c -o localtime.o localtime.c localtime.c: In function âdiffer_by_repeatâ: localtime.c:326: warning: comparison is always false due to limited range of data type 2. valgrind reports following error for some timezones (only when localtime.c is compiled as 64-bit binary, it does not happen when compiled as 32-bit binary): ==26671== Conditional jump or move depends on uninitialised value(s) ==26671== at 0x401CDA: typesequiv (localtime.c:588) ==26671== by 0x401B85: tzload (localtime.c:562) ==26671== by 0x405553: tz_validate (localtime.c:2017) ==26671== by 0x40560D: main (localtime.c:2040) One can reproduce it using following valgrind command and patch: valgrind -v --tool=memcheck --leak-check=full --leak-resolution=med --show- reachable=yes --error-limit=no --workaround-gcc296-bugs=yes --run-libc-freeres=no --log- file=valgrind-tzvalid ./tzvalid /<path/to/my/zoneinfo/EST5EDT> NOTE: I can consistently reproduce it for EST5EDT timezone and I don't get this error for GMT. Patch for localtime.c for reproducing above error: --- org/localtime.c 2010-08-16 16:27:05.000000000 +0530 +++ localtime.c 2010-10-01 17:34:45.999089000 +0530 @@ -2004,6 +2004,47 @@ #endif /* defined CMUCS */ +static int +tz_validate(const char * name) +{ + struct state state = {0}; + time_t now; + struct tm tm; + + time(&now); + tzload(gmt, &state, TRUE); + if (0 == tzload(name, &state, TRUE)) { + localsub(&now, 0L, &tm); + return TRUE; + } + if (0 == tzparse(name, &state, FALSE)) { + localsub(&now, 0L, &tm); + return TRUE; + } + return FALSE; +} + +/* + * Compile it simply as "$gcc -m64 -g -o tzvalid localtime.c" + */ +int main(int argc, char ** argv) { + + if (argc != 2) { + printf("Usage: ./tzvalid <timezone_file_path>\n"); + return 2; + } + + char * tz = argv[1]; + + if (tz_validate(tz)) { + printf("timezone %s is valid\n", tz); + return 0; + } else { + printf("timezone %s is not valid\n", tz); + return 1; + } +} + /* ** XXX--is the below the right way to conditionalize?? */ Thanks in advance, Ravindra

1 0

FW: About China time zones
by Olson, Arthur David (NIH/NCI) [E] Oct. 4, 2010

Oct. 4, 2010

I'm forwarding this message from Tom Korte, who is not on the time zone mailing list. Those of you who are on the list, please direct replies appropriately. (While all of China is now officially in one time zone there were different zones until 1980; that's why there are multiple entries in the database. The database started out life with an entry for Chungking; the entry was later changed to use the Pinyin spelling Chongqing, but a Chungking link was kept for backward compatibility. And time zones are as a rule named for the city in the zone with the highest population--at the time the zone was named if not at present.) --ado From: Walter Korte [mailto:Walter.Korte@tomtom.com] Sent: Thursday, September 23, 2010 11:15 To: tz(a)lecserver.nci.nih.gov Subject: About China time zones To the maintainers of the tz database, We are using the TZ database in our navigation SW, and we have had feedback from our users in China that the time zone city names included do not make sense to local users. Neither Harbin, Shanghai nor Chungking (this is a very strange spelling, it is spelled Chongqing in Pinyin) is used in China to indicate a time zone. It would make most sense to remove these 3 entries and replace them with a single Beijing entry in the database. As HongKong and Macao are special administrative regions (SAR) in China, listing these makes sense. For the cities listed that are located in Xinjiang (Kashgar and Urumqi), these should not be listed as China only has one official time zone. But in practice Xinjiang uses its own time zone so keeping these would be better. Best regards, Walter Korte TomTom Asia

7 9

Cond jump or move depends on uninit'd value error on 64-bit Linux
by Ravindra Oct. 1, 2010

Oct. 1, 2010

I'm seeing two problems in localtime.c: 1. When I compile localtime.c using gcc 4.1.2 for 32-bit, I get following warning: $ gcc -m32 -g -c -o localtime.o localtime.c localtime.c: In function âdiffer_by_repeatâ: localtime.c:326: warning: comparison is always false due to limited range of data type 2. valgrind reports following error for some timezones (only when localtime.c is compiled as 64-bit binary, it does not happen when compiled as 32-bit binary): ==26671== Conditional jump or move depends on uninitialised value(s) ==26671== at 0x401CDA: typesequiv (localtime.c:588) ==26671== by 0x401B85: tzload (localtime.c:562) ==26671== by 0x405553: tz_validate (localtime.c:2017) ==26671== by 0x40560D: main (localtime.c:2040) One can reproduce it using following valgrind command and patch: valgrind -v --tool=memcheck --leak-check=full --leak-resolution=med --show- reachable=yes --error-limit=no --workaround-gcc296-bugs=yes --run-libc- freeres=no --log-file=valgrind-tzvalid ./tzvalid /<path/to/my/zoneinfo/EST5EDT> NOTE: I can consistently reproduce it for EST5EDT timezone and I don't get this error for GMT. Patch for localtime.c for reproducing above error: --- org/localtime.c 2010-08-16 16:27:05.000000000 +0530 +++ localtime.c 2010-10-01 17:34:45.999089000 +0530 @@ -2004,6 +2004,47 @@ #endif /* defined CMUCS */ +static int +tz_validate(const char * name) +{ + struct state state = {0}; + time_t now; + struct tm tm; + + time(&now); + tzload(gmt, &state, TRUE); + if (0 == tzload(name, &state, TRUE)) { + localsub(&now, 0L, &tm); + return TRUE; + } + if (0 == tzparse(name, &state, FALSE)) { + localsub(&now, 0L, &tm); + return TRUE; + } + return FALSE; +} + +/* + * Compile it simply as "$gcc -m64 -g -o tzvalid localtime.c" + */ +int main(int argc, char ** argv) { + + if (argc != 2) { + printf("Usage: ./tzvalid <timezone_file_path>\n"); + return 2; + } + + char * tz = argv[1]; + + if (tz_validate(tz)) { + printf("timezone %s is valid\n", tz); + return 0; + } else { + printf("timezone %s is not valid\n", tz); + return 1; + } +} + /* ** XXX--is the below the right way to conditionalize?? */ Thanks in advance, Ravindra

1 0