- tz - lists.iana.org

UK timezone news --- Summary: don't panic
by peter＠aldie.co.uk Jan. 21, 1996

Jan. 21, 1996

UK timezone matters came to a head on Friday 18 January when John Butterfill MP failed to get a second reading in the House of Commons for his British Time (Extra Daylight) Bill. This means nothing will change. This story starts back in 1989 when the Home Office, the department responsible for these things, produced a consultation document on Summer Time. Officially the Government were neutral, but it was widely reported that they wanted to move the UK into the central European timezone, GMT+1 in winter and GMT+2 in summer. Back then the EC (as it was then called) hadn't agreed on a common Summer Time end date so this proposed change was bound up with European harmonisation. There was a wide divergence of opinion in the responses to the document and the Government couldn't decide what to do. I am not aware of any statement being published. The Home Office, if you ask them, say the whole matter `is under consant review'. A friend of mine elsewhere in the Civil Service translates this as `we haven't a clue what to do about this and we wish it would go away'. In the intervening years the EC Commission agreed the seventh Directive on summertime arrangements which finally introduced a common end date. As this date was basically the UK end date, with all the other countries changing from the end of September to the end of October, this removed one of the objections in the UK for European harminisation: that it would lead to darker evenings in October. As the EC Commission has said several times that it is solely interested in harmonising dates and has no interest in what timezone countries are in, the responsibility for any change would now fall squarely on the UK Government and could not be blamed on Europe. In January 1995 a member of the House of Lords (the upper house of our Parliament), Viscount Mountgarret, introduced his Central European Time Bill. In the debate he complained bitterly about the lack of decision since 1989 and said the Government had broken several specific committments to him that things would be done. His Bill proposed that England, Wales and Northern Ireland, but not Scotland, move to CET. This moves the timezone boundary from the English Channel up to the Scottish border. The Scots, because of being in the far north and west of the UK, have been the most vociferous opponents of any change. His Bill was given a second reading, nobody put their names down to speak in the next, committee, stage, so as far as I understand Lords procedure it passed. It had no chance of becoming law however as nobody introduced it in the House of Commons. I think that Viscount Mountgarret introduced the Bill at least partly to put the Government on the spot and force some sort of decision. In that he failed. The action now moves to the House of Commons. Every year when the new session of Parliament starts there is a ballot among MPs for places on the list for introducing Private Members Bills. Only the first few MPs in the list have any hope of getting their Bill (which they might not even have thought of when they enter the ballot) passed into law, and then usually only with help in the form of debating time from the Government. Last year John Butterfill came top of the list, and announced he would introduce a British Time (Extra Daylight) Bill to move the whole of the UK into the CET zone. I think he offered the Scots the option of opting out if they wanted to but nobody really took this seriously. This did put the Government on the spot as they would have to decide whether to give it time. I asked the Home Office last week and all they would say was: `the Government's position was that it was not a party political matter and it would be decided by Parliament on a free vote'. It was widely reported though that they wouldn't give it time, the Scottish Office ministers had won the right to vote against it and all other ministers had been instructed to abstain or have some other engagement so they wouldn't be in the House at all. In the event none of this was really necessary. Last Friday, after a good humoured debate, the Bill fell on a technicality. One hundred MPs were required (I don't know why, some arcane rule) to vote in favour of a motion to close the debate, so the Bill's second reading vote could take place. Only 93 voted in favour and the Bill was lost. The Government escapes one again without having to make any decision. So, nothing changes as a result of all this, but the backers of the Bill, who call themselves the Daylight Extra Campaign, have already said they will be back. There will have to be an election in the UK before the end of April next year and if the current opinion polls are to be believed the Conservative Party are heading for a drubbing of possibly Canadian proportions. The Daylight Extra folk are already talking of bringing the proposal back, possibly as a 5 year experiment, after a change of Government. One of the most depressing things about all this is that many people have been taken in by John Butterfill's `(Extra Daylight)' and somehow think that politicians can conjure the stuff out of thin air. Peter Ilieve peter(a)aldie.co.uk

2 1

Re: Bug#1229
by Alan.Perry＠Eng.Sun.COM Jan. 18, 1996

Jan. 18, 1996

> From guy(a)netapp.com Wed Jan 17 21:58 PST 1996 > > Reported by Herbert Xu <herbert(a)greathan.apana.org.au> > > > > PRC abandoned daylight saving in 1992 or 1993 but currently the files do > > not reflect this change. > > The latest version of the time zone data files appear to reflect this: > > # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S > Rule Shang 1928 only - Jan 1 0:00 0 S > Rule Shang 1940 only - Jun 3 0:00 1:00 D > Rule Shang 1940 1941 - Oct 1 0:00 0 S > Rule Shang 1941 only - Mar 16 0:00 1:00 D > Rule PRC 1949 only - Jan 1 0:00 0 S > Rule PRC 1986 only - May 4 0:00 1:00 D > Rule PRC 1986 1991 - Sep Sun>=11 0:00 0 S > Rule PRC 1987 1991 - Apr Sun>=10 0:00 1:00 D > > although earlier versions didn't, I think (the files in SunOS 5.4 > doesn't, for example, and they're probably based on an earlier version). SunOS 5.4 (Solaris 2.4) zoneinfo files are taken from tzdata94d. There should be a patch available that will update this to tzdata95c (I think). We still need to update our zic to compile some of the latest zoneinfo files (like europe) as well as fix zdump, which we (I) :-( broke in 2.4. alan

1 0

Re: Bug#1229
by Fernando Alegre Jan. 17, 1996

Jan. 17, 1996

Reported by Herbert Xu <herbert(a)greathan.apana.org.au> PRC abandoned daylight saving in 1992 or 1993 but currently the files do not reflect this change.

2 1

Some bugs in timezone reported to the Debian project
by Fernando Alegre Jan. 17, 1996

Jan. 17, 1996

I maintain the timezone utilities for the Debian project. In case you don't know, the Debian project is the result of a volunteer effort to create a high-quality non-commercial Linux distribution which will be supported by the Free Software Foundation. As part of this project we have set up a bug tracking system. Users post bug reports and the Debian maintainers try to fix them or forward them to the developer of a given package. There are some bugs related to the timezone utilities which haven't been fixed for some time. I would like them to be considered for the next release of these packages. Here is the list of outstanding bugs: 1) Reported by Herbert Xu <herbert(a)greathan.apana.org.au> PRC abandoned daylight saving in 1992 or 1993 but currently the files do not reflect this change. 2) Reported by Raul Miller <rdm(a)tad.micro.umn.edu> zdump output from all of the files in the US directory gives GMT times. 3) Reported by Raul Miller <rdm(a)tad.micro.umn.edu> there appears to be a complete duplicate set of files under zoneinfo/right 4) Reported by Raul Miller <rdm(a)tad.micro.umn.edu> either zdump is broken or the zoneinfo files are broken -- they depend on the file name rather than file contents to determine their time zone. This makes "localtime" rather useless, because it conceals the file name. [My interpretation is: zdump needs an option for dumping files] Our bug tracking system needs that a separate e-mail message is sent for each bug, so that all messages regarding the same bug can be automatically archived together. In order for this to function, it is necessary for me to send you a separate message for each bug with a Subject: line reading "Re: Bug#0000". Therefore you will receive 4 more messages in the near future. I am asking you to kindly keep the Subject: line of each message untouched, as well as the Cc: field. If you need more details please feel free to contact me. Thank you for your attention. Regards, Fernando Alegre alegre(a)mars.superlink.net

1 0

tzcode96c.tar.gz
by ado Jan. 16, 1996

Jan. 16, 1996

tzcode96c.tar.gz is now available for anonymous ftp from elsie.nci.nih.gov (165.112.73.1) in the ~ftp/pub directory: -rw-r--r-- 1 ado 73301 Jan 16 16:58 tzcode96c.tar.gz It incorporates the leap-year streamlining and binary-search changes noted on the mailing list, as well as the fix to newctime.3. --ado

1 0

A patch to newctime.3
by J.T. Conklin Jan. 15, 1996

Jan. 15, 1996

The enclosed patch removes text that says that localtime calls tzset if it hasn't been called before. Localtime (as well as asctime, ctime, mktime and strftime) now always call tzset. There should probably be text added that documents the real behavior. --jtc Index: newctime.3 =================================================================== RCS file: /home/cvsroot/tz/tzcode/newctime.3,v retrieving revision 1.1.1.1 diff -c -r1.1.1.1 newctime.3 *** newctime.3 1996/01/15 21:48:02 1.1.1.1 --- newctime.3 1996/01/15 22:00:31 *************** *** 56,68 **** .I Localtime\^ corrects for the time zone and any time zone adjustments (such as Daylight Saving Time in the U.S.A.). - Before doing so, - .I localtime\^ - calls - .I tzset\^ - (if - .I tzset\^ - has not been called in the current process). After filling in the ``tm'' structure, .I localtime sets the --- 56,61 ----

1 0

tzcode96b.tar.gz and ISO8601
by fred＠genesis.demon.co.uk Jan. 15, 1996

Jan. 15, 1996

Hello! I've recently been 'dabbling' in dates and in particular ISO8601 week numbers so I was interested in tzcode96b.tar.gz, in particular the strftime.c file. First however I'll just point out something minor in tzfile.h which is: #define isleap(y) ((((y) % 4) == 0 && ((y) % 100) != 0) || ((y) % 400) == 0) This is not the best way to test for a leap year since if the year is not divisible by 4 we know it is not divisible by 400 (but it tests for this anyway). The more usual form is: #define isleap(y) (((y) % 4) == 0 && (((y) % 100) != 0 || ((y) % 400) == 0)) My main interest is in the following extract from the code for the 'V' (amongst others) conversion specifier: #ifdef XPG4_1994_04_09 if (w == 52 && t->tm_mon == TM_JANUARY) w = 53; #endif /* defined XPG4_1994_04_09 */ Is it true that XPG4 has a slightly different definition of week number than ISO8601? What this appears to be saying is that if week 52 crosses into the new year the days in January become week 53 (but if week 53 crosses the year boundary it stays as week 53). The following is the function I wrote to calculate ISO8601 week numbers. I've added an 'XPG4' option in what I believe to be an equivalent way (this means that points 2 and 6 of the 'rules' aren't quite correct for XPG4). Since I'm submitting something similar for inclusion in the free 'Snippets' library any thoughts would be welcome. /****************************************************************************** * weeknum_ISO8601 January 1996 L.Kirby * * Released to the Public Domain * * Calculates the week number of a day in the year based on the ISO8601 week * numbering scheme. The arguments are: * * t - a pointer to a struct tm in which the following members must be set and * normalised to the standard ranges specified (as standard library * functions gmtime, localtime and mktime should do): * * tm_wday - The day of the week of the day in question: * 0-Sunday -> 6-Saturday * * tm_yday - The day of the year of the day in question: 0 -> 365 * January 1st is day 0. * * tm_year - The year since 1900. * * firstdow - This defines the day of the week on which a week starts: * 0-Sunday -> 6-Saturday. For normal ISO8601 weeks that start on * a Monday this should be 1. * ****************************************************************************** * * The week number is a value between 1 and 53 inclusive defined according to * the following rules: * * 1. The Gregorian calendar is assumed. * * 2. There are always 7 consecutive days with the same week number. * * 3. January 4th is defined to be in week 1. Equivalently week 1 is the * first week of a year which has at least 4 days in that year. * * 4. firstdow defines the day of the week which starts a new week i.e. has * a different week number from the previous day. * * 5. Week numbers increase in sequence from week 1 until the week that is * defined to be week 1 of the following year. * * It follows that: * * 6. Up to January 3rd may be in either week 1 of the current year or in * weeks 52 or 53 of the previous year. * * 7. 29th December onwards may be in either weeks 52 or 53 of the current * year or week 1 of the following year. * ****************************************************************************** */ #include <time.h> #define isleap(year) (!((year) % 4) && (((year) % 100) || !((year) % 400))) int weeknum_ISO8601(const struct tm *t, int firstdow) { const int tmp1 = firstdow - t->tm_wday; const int tmp2 = t->tm_yday + ((tmp1 > 0) ? 3 : 10) + tmp1; const int fourthdaynum = tmp2 % 7; int week = tmp2 / 7; if (week == 0) { #ifdef XPG4_1994_04_09 week = 53; #else const int year = t->tm_year + (1900 % 400) - 1; week = (fourthdaynum + isleap(year) >= 6) ? 53 : 52; #endif } else if (week == 53) { const int year = t->tm_year + (1900 % 400); if (fourthdaynum > isleap(year)) week = 1; } return week; } -- ----------------------------------------- Lawrence Kirby | fred(a)genesis.demon.co.uk Wilts, England | 70734.126(a)compuserve.com -----------------------------------------

1 0

minor tz tuneups and portability fix
by Paul Eggert Jan. 15, 1996

Jan. 15, 1996

Here are some proposed very minor code tuneups for the tz code. While accumulating these I noticed a portability bug in localtime.c; the expression `(t < 0) ? 0 : ((time_t) 1 << bits)' yields undefined results if time_t is unsigned, since `bits' is the number of bits in a time_t in that case. The patch below fixes this bug as part of the tuneups. =================================================================== RCS file: RCS/tzfile.h,v retrieving revision 1995.3 retrieving revision 1995.3.1.1 diff -c -r1995.3 -r1995.3.1.1 *** tzfile.h 1995/03/11 17:55:58 1995.3 --- tzfile.h 1996/01/15 06:22:59 1995.3.1.1 *************** *** 153,159 **** ** that will probably do. */ ! #define isleap(y) ((((y) % 4) == 0 && ((y) % 100) != 0) || ((y) % 400) == 0) #ifndef USG --- 153,159 ---- ** that will probably do. */ ! #define isleap(y) (((y) % 4) == 0 && (((y) % 100) != 0 || ((y) % 400) == 0)) #ifndef USG =================================================================== RCS file: RCS/difftime.c,v retrieving revision 1994.8 retrieving revision 1994.8.1.1 diff -c -r1994.8 -r1994.8.1.1 *** difftime.c 1994/12/09 04:33:32 1994.8 --- difftime.c 1996/01/15 06:22:59 1994.8.1.1 *************** *** 43,51 **** /* ** Repair delta overflow. */ ! hibit = 1; ! while ((hibit <<= 1) > 0) ! continue; /* ** The following expression rounds twice, which means ** the result may not be the closest to the true answer. --- 43,49 ---- /* ** Repair delta overflow. */ ! hibit = (~ (time_t) 0) << (TYPE_BIT(time_t) - 1); /* ** The following expression rounds twice, which means ** the result may not be the closest to the true answer. *************** *** 65,74 **** ** This problem occurs only with very large differences. ** It's too painful to fix this portably. ** We are not alone in this problem; ! ** many C compilers round twice when converting ** large unsigned types to small floating types, ** so if time_t is unsigned the "return delta" above ! ** has the same double-rounding problem. */ return delta - 2 * (long_double) hibit; } --- 63,72 ---- ** This problem occurs only with very large differences. ** It's too painful to fix this portably. ** We are not alone in this problem; ! ** some C compilers round twice when converting ** large unsigned types to small floating types, ** so if time_t is unsigned the "return delta" above ! ** has the same double-rounding problem with those compilers. */ return delta - 2 * (long_double) hibit; } =================================================================== RCS file: RCS/localtime.c,v retrieving revision 1995.7 retrieving revision 1995.7.1.1 diff -c -r1995.7 -r1995.7.1.1 *** localtime.c 1995/10/30 15:32:46 1995.7 --- localtime.c 1996/01/15 06:22:59 1995.7.1.1 *************** *** 1163,1175 **** tmp->tm_hour = (int) (rem / SECSPERHOUR); rem = rem % SECSPERHOUR; tmp->tm_min = (int) (rem / SECSPERMIN); ! tmp->tm_sec = (int) (rem % SECSPERMIN); ! if (hit) ! /* ! ** A positive leap second requires a special ! ** representation. This uses "... ??:59:60" et seq. ! */ ! tmp->tm_sec += hit; tmp->tm_wday = (int) ((EPOCH_WDAY + days) % DAYSPERWEEK); if (tmp->tm_wday < 0) tmp->tm_wday += DAYSPERWEEK; --- 1163,1173 ---- tmp->tm_hour = (int) (rem / SECSPERHOUR); rem = rem % SECSPERHOUR; tmp->tm_min = (int) (rem / SECSPERMIN); ! /* ! ** A positive leap second requires a special ! ** representation. This uses "... ??:59:60" et seq. ! */ ! tmp->tm_sec = (int) (rem % SECSPERMIN) + hit; tmp->tm_wday = (int) ((EPOCH_WDAY + days) % DAYSPERWEEK); if (tmp->tm_wday < 0) tmp->tm_wday += DAYSPERWEEK; *************** *** 1344,1358 **** /* ** Calculate the number of magnitude bits in a time_t ** (this works regardless of whether time_t is ! ** signed or unsigned, though lint complains if unsigned). */ ! for (bits = 0, t = 1; t > 0; ++bits, t <<= 1) ! continue; /* ** If time_t is signed, then 0 is the median value, ! ** if time_t is unsigned, then 1 << bits is median. */ ! t = (t < 0) ? 0 : ((time_t) 1 << bits); for ( ; ; ) { (*funcp)(&t, offset, &mytm); dir = tmcomp(&mytm, &yourtm); --- 1342,1355 ---- /* ** Calculate the number of magnitude bits in a time_t ** (this works regardless of whether time_t is ! ** signed or unsigned). */ ! bits = TYPE_BIT(time_t) - TYPE_SIGNED(time_t); /* ** If time_t is signed, then 0 is the median value, ! ** if time_t is unsigned, then 1 << (bits - 1) is median. */ ! t = TYPE_SIGNED(time_t) ? 0 : ((time_t) 1 << (TYPE_BIT(time_t) - 1)); for ( ; ; ) { (*funcp)(&t, offset, &mytm); dir = tmcomp(&mytm, &yourtm);

2 1

tzcode96b.tar.gz
by ado Jan. 10, 1996

Jan. 10, 1996

The file tzcode96b.tar.gz is available for anonymous ftp from the ~ftp/pub directory of elsie.nci.nih.gov (165.112.73.1): -rw-r--r-- 1 ado 73347 Jan 10 20:43 tzcode96b.tar.gz It contains fixes and enhancements from Paul Eggert, including code that emulates the behavior of recent versions of the SunOS "date" command. --ado

1 0

tzcode96a.tar.gz and tzdata96a.tar.gz
by ado Jan. 6, 1996

Jan. 6, 1996

These files are now available by anonymous ftp from the ~ftp/pub directory on elsie.nci.nih.gov (165.112.73.1): -rw-r--r-- 1 ado 73078 Jan 6 09:08 tzcode96a.tar.gz -rw-r--r-- 1 ado 72790 Jan 6 09:08 tzdata96a.tar.gz The "data" file includes the Israel updates recently circulated on the mailing list; the "code" file contains fixes to strftime.c for correct ISO 8601 week number generation, plus support for two new formats ('G' and 'g') to give ISO 8601 year numbers (which are not necessarily the same as calendar year numbers). --ado

1 0