- tz - lists.iana.org

Reentrant versions of localtime and gmtime.
by J.T. Conklin Oct. 31, 1995

Oct. 31, 1995

Thread-safe programming needs reentrant versions of some of the time functions. The following patch adds localtime_r and gmtime_r a'la Solaris (The Solaris manpage mentiones that these are taken from a POSIX draft, but I don't have a copy to double check). Although it is possible to write localtime and gmtime in terms of the new functions, I decided not to because they are so trival. The Solaris manpage also mentions ctime_r and asctime_r. I'll submit those later tonight. --jtc *** OLD/localtime.c Mon Oct 30 17:00:52 1995 --- localtime.c Mon Oct 30 17:03:48 1995 *************** *** 1026,1031 **** --- 1026,1041 ---- } struct tm * + localtime_r(timep, res) + const time_t * const timep; + struct tm * tm; + { + tzset(); + localsub(timep, 0L, tm); + return tm; + } + + struct tm * localtime(timep) const time_t * const timep; { *************** *** 1072,1077 **** --- 1082,1096 ---- #endif /* State Farm */ } #endif /* defined TM_ZONE */ + } + + struct tm * + gmtime_r(timep, tm) + const time_t * const timep; + struct tm * tm; + { + gmtsub(timep, 0L, tm); + return tm; } struct tm *

2 1

Question about reentrant ctime and asctime
by J.T. Conklin Oct. 31, 1995

Oct. 31, 1995

In my previous message, I mentioned that I'd submit versions of asctime_r and ctime_r later this evening. However, I discovered that most systems declare these as: int asctime_r (const struct tm *, char *, int); int ctime_r (const time_t *, char *, int); while Solaris provides: char *asctime_r (const struct tm *, char *, int); char *ctime_r (const time_t *, char *, int); *AND* char *asctime_r (const struct tm *, char *) char *ctime_r (const time_t *, char *); Naturally I'd like to implement the interface standardized by whatever POSIX standard covers this (realtime?, threads?). If someone with the appropriate standards document could point out the correct interface, please let me know. --jtc

1 0

Re: tz 8-bit-clean proposed patch
by ado Oct. 30, 1995

Oct. 30, 1995

The (slightly reformatted) "8-bit-clean proposed patch" and the "tz minor patch: US/Eastern -> America/New_York" have been incorporated into... -rw-r--r-- 1 ado 69206 Oct 30 10:40 tzcode95g.tar.gz -rw-r--r-- 1 ado 69838 Oct 30 10:40 tzdata95k.tar.gz in elsie.nci.nih.gov's ~ftp/pub directory, along with a snapshot of the USNO's latest data ("usno1995") and some other minor cleanups. --ado

1 0

tz minor patch: US/Eastern -> America/New_York
by Paul Eggert Oct. 30, 1995

Oct. 30, 1995

Some tz comments refer to US/Eastern, but it is present only if the `backward' file is used. Here is a proposed patch. =================================================================== RCS file: RCS/Theory,v retrieving revision 1995.3 retrieving revision 1995.3.1.1 diff -c -r1995.3 -r1995.3.1.1 *** Theory 1993/11/22 20:23:18 1995.3 --- Theory 1995/10/30 01:50:47 1995.3.1.1 *************** *** 47,53 **** abbreviations are used. It was recognized that allowing the "TZ" environment variable to ! take on values such as "US/Eastern" might cause "old" programs (that expect "TZ" to have a certain form) to operate incorrectly; consideration was given to using some other environment variable (for example, "TIMEZONE") to hold the string used to generate the --- 47,53 ---- abbreviations are used. It was recognized that allowing the "TZ" environment variable to ! take on values such as "America/New_York" might cause "old" programs (that expect "TZ" to have a certain form) to operate incorrectly; consideration was given to using some other environment variable (for example, "TIMEZONE") to hold the string used to generate the =================================================================== RCS file: RCS/date.c,v retrieving revision 1995.3.1.1 retrieving revision 1995.3.1.2 diff -c -r1995.3.1.1 -r1995.3.1.2 *** date.c 1995/10/30 00:43:48 1995.3.1.1 --- date.c 1995/10/30 01:50:47 1995.3.1.2 *************** *** 207,215 **** if ((cp = strchr(value, '.')) != NULL) { /* ** Ensure that the failure of ! ** TZ=US/Eastern date 8712312359.60 ** doesn't get misdiagnosed. (It was ! ** TZ=US/Eastern date 8712311859.60 ** when the leap second was inserted.) ** The normal check won't work since ** the given time is valid in GMT. --- 207,215 ---- if ((cp = strchr(value, '.')) != NULL) { /* ** Ensure that the failure of ! ** TZ=America/New_York date 8712312359.60 ** doesn't get misdiagnosed. (It was ! ** TZ=America/New_York date 8712311859.60 ** when the leap second was inserted.) ** The normal check won't work since ** the given time is valid in GMT. =================================================================== RCS file: RCS/zic.8,v retrieving revision 1995.3 retrieving revision 1995.3.1.1 diff -c -r1995.3 -r1995.3.1.1 *** zic.8 1995/03/11 17:55:51 1995.3 --- zic.8 1995/10/30 01:50:47 1995.3.1.1 *************** *** 324,336 **** .nf .ti +.5i .if t .ta \w'Link\0\0'u +\w'LINK-FROM\0\0'u ! .if n .ta \w'Link\0\0'u +\w'US/Eastern\0\0'u Link LINK-FROM LINK-TO .sp For example: .sp .ti +.5i ! Link US/Eastern EST5EDT .sp .fi The --- 324,336 ---- .nf .ti +.5i .if t .ta \w'Link\0\0'u +\w'LINK-FROM\0\0'u ! .if n .ta \w'Link\0\0'u +\w'America/New_York\0\0'u Link LINK-FROM LINK-TO .sp For example: .sp .ti +.5i ! Link America/New_York US/Eastern .sp .fi The

1 0

tz 8-bit-clean proposed patch
by Paul Eggert Oct. 30, 1995

Oct. 30, 1995

In a few places the tz code assumes that characters are nonnegative. The tz code uses `isascii' to avoid this problem sometimes, but not always, and anyway `isascii' isn't portable to some hosts that have strict ANSI or Posix checking turned on, since isascii is not required by either the C Standard or Posix. Here is a proposed patch that handles top-bit-on characters correctly in all the places where I found this problem. The patch is portable to hosts that don't define isascii. The proposed patch also removes a few unnecessary tests, and changes one call to ciequal to invoke strcmp, since the argument string has no alphabetic characters in it. =================================================================== RCS file: RCS/private.h,v retrieving revision 1995.3 retrieving revision 1995.3.1.1 diff -c -r1995.3 -r1995.3.1.1 *** private.h 1995/03/11 17:55:53 1995.3 --- private.h 1995/10/30 00:43:48 1995.3.1.1 *************** *** 47,53 **** #include "sys/types.h" /* for time_t */ #include "stdio.h" - #include "ctype.h" #include "errno.h" #include "string.h" #include "limits.h" /* for CHAR_BIT */ --- 47,52 ---- *************** *** 66,71 **** --- 65,73 ---- #define R_OK 4 #endif /* !defined R_OK */ #endif /* !(HAVE_UNISTD_H - 0) */ + + /* Unlike <ctype.h>'s isdigit, this also works if c < 0 | c > UCHAR_MAX. */ + #define is_digit(c) ((unsigned)(c) - '0' <= 9) /* ** Workarounds for compilers/systems. =================================================================== RCS file: RCS/date.c,v retrieving revision 1995.3 retrieving revision 1995.3.1.1 diff -c -r1995.3 -r1995.3.1.1 *** date.c 1995/03/11 17:55:47 1995.3 --- date.c 1995/10/30 00:43:48 1995.3.1.1 *************** *** 459,465 **** nondigit(cp) register const char * cp; { ! while (isdigit(*cp)) ++cp; return cp; } --- 459,465 ---- nondigit(cp) register const char * cp; { ! while (is_digit(*cp)) ++cp; return cp; } *************** *** 592,598 **** dotp = strchr(value, '.'); for (cp = value; *cp != '\0'; ++cp) ! if (!isdigit(*cp) && cp != dotp) wildinput("time", value, "contains a nondigit"); if (dotp == NULL) --- 592,598 ---- dotp = strchr(value, '.'); for (cp = value; *cp != '\0'; ++cp) ! if (!is_digit(*cp) && cp != dotp) wildinput("time", value, "contains a nondigit"); if (dotp == NULL) =================================================================== RCS file: RCS/localtime.c,v retrieving revision 1995.6 retrieving revision 1995.6.1.1 diff -c -r1995.6 -r1995.6.1.1 *** localtime.c 1995/10/28 16:09:02 1995.6 --- localtime.c 1995/10/30 00:43:48 1995.6.1.1 *************** *** 415,421 **** { register char c; ! while ((c = *strp) != '\0' && !isdigit(c) && c != ',' && c != '-' && c != '+') ++strp; return strp; --- 415,421 ---- { register char c; ! while ((c = *strp) != '\0' && !is_digit(c) && c != ',' && c != '-' && c != '+') ++strp; return strp; *************** *** 438,452 **** register char c; register int num; ! if (strp == NULL || !isdigit(*strp)) return NULL; num = 0; ! while ((c = *strp) != '\0' && isdigit(c)) { num = num * 10 + (c - '0'); if (num > max) return NULL; /* illegal value */ ! ++strp; ! } if (num < min) return NULL; /* illegal value */ *nump = num; --- 438,452 ---- register char c; register int num; ! if (strp == NULL || !is_digit(c = *strp)) return NULL; num = 0; ! do { num = num * 10 + (c - '0'); if (num > max) return NULL; /* illegal value */ ! c = *++strp; ! } while (is_digit(c)); if (num < min) return NULL; /* illegal value */ *nump = num; *************** *** 508,521 **** register const char * strp; long * const offsetp; { ! register int neg; if (*strp == '-') { neg = 1; ++strp; ! } else if (isdigit(*strp) || *strp++ == '+') ! neg = 0; ! else return NULL; /* illegal offset */ strp = getsecs(strp, offsetp); if (strp == NULL) return NULL; /* illegal time */ --- 508,520 ---- register const char * strp; long * const offsetp; { ! register int neg = 0; if (*strp == '-') { neg = 1; ++strp; ! } else if (*strp == '+') ! ++strp; strp = getsecs(strp, offsetp); if (strp == NULL) return NULL; /* illegal time */ *************** *** 560,566 **** if (*strp++ != '.') return NULL; strp = getnum(strp, &rulep->r_day, 0, DAYSPERWEEK - 1); ! } else if (isdigit(*strp)) { /* ** Day of year. */ --- 559,565 ---- if (*strp++ != '.') return NULL; strp = getnum(strp, &rulep->r_day, 0, DAYSPERWEEK - 1); ! } else if (is_digit(*strp)) { /* ** Day of year. */ =================================================================== RCS file: RCS/scheck.c,v retrieving revision 1994.8 retrieving revision 1994.8.1.1 diff -c -r1994.8 -r1994.8.1.1 *** scheck.c 1994/12/09 04:33:18 1994.8 --- scheck.c 1995/10/30 00:43:48 1994.8.1.1 *************** *** 42,48 **** *tp++ = '*'; if (*fp == '*') ++fp; ! while (isascii(*fp) && isdigit(*fp)) *tp++ = *fp++; if (*fp == 'l' || *fp == 'h') *tp++ = *fp++; --- 42,48 ---- *tp++ = '*'; if (*fp == '*') ++fp; ! while (is_digit(*fp)) *tp++ = *fp++; if (*fp == 'l' || *fp == 'h') *tp++ = *fp++; =================================================================== RCS file: RCS/zic.c,v retrieving revision 1995.6 retrieving revision 1995.6.1.1 diff -c -r1995.6 -r1995.6.1.1 *** zic.c 1995/10/28 20:51:12 1995.6 --- zic.c 1995/10/30 00:43:48 1995.6.1.1 *************** *** 10,15 **** --- 10,28 ---- #include "sys/stat.h" /* for umask manifest constants */ #endif /* defined unix */ + /* + ** On some ancient hosts, predicates like `isspace(C)' are defined + ** only if isascii(C) || C == EOF. Modern hosts obey the C Standard, + ** which says they are defined only if C == (unsigned char)C || C == EOF. + ** Neither the C Standard nor Posix require that `isascii' exist. + ** For portability, we check both ancient and modern requirements. + ** If isascii is not defined, the isascii check succeeds trivially. + */ + #include "ctype.h" + #ifndef isascii + #define isascii(x) 1 + #endif + struct rule { const char * r_filename; int r_linenum; *************** *** 721,727 **** while (fields[nfields] != NULL) { static char nada; ! if (ciequal(fields[nfields], "-")) fields[nfields] = &nada; ++nfields; } --- 734,740 ---- while (fields[nfields] != NULL) { static char nada; ! if (strcmp(fields[nfields], "-") == 0) fields[nfields] = &nada; ++nfields; } *************** *** 1702,1709 **** static int lowerit(a) ! const int a; { return (isascii(a) && isupper(a)) ? tolower(a) : a; } --- 1715,1723 ---- static int lowerit(a) ! int a; { + a = (unsigned char)a; return (isascii(a) && isupper(a)) ? tolower(a) : a; } *************** *** 1775,1781 **** emalloc((int) ((strlen(cp) + 1) * sizeof *array)); nsubs = 0; for ( ; ; ) { ! while (isascii(*cp) && isspace(*cp)) ++cp; if (*cp == '\0' || *cp == '#') break; --- 1789,1795 ---- emalloc((int) ((strlen(cp) + 1) * sizeof *array)); nsubs = 0; for ( ; ; ) { ! while (isascii(*cp) && isspace((unsigned char)*cp)) ++cp; if (*cp == '\0' || *cp == '#') break; *************** *** 1788,1795 **** ++dp; else error("Odd number of quotation marks"); } while (*cp != '\0' && *cp != '#' && ! (!isascii(*cp) || !isspace(*cp))); ! if (isascii(*cp) && isspace(*cp)) ++cp; *dp = '\0'; } --- 1802,1809 ---- ++dp; else error("Odd number of quotation marks"); } while (*cp != '\0' && *cp != '#' && ! (!isascii(*cp) || !isspace((unsigned char)*cp))); ! if (isascii(*cp) && isspace((unsigned char)*cp)) ++cp; *dp = '\0'; } *************** *** 1952,1959 **** /* ** DOS drive specifier? */ ! if (strlen(name) == 2 && isascii(name[0]) && ! isalpha(name[0]) && name[1] == ':') { *cp = '/'; continue; } --- 1966,1973 ---- /* ** DOS drive specifier? */ ! if (isalpha((unsigned char)name[0]) ! && name[1] == ':' && !name[2]) { *cp = '/'; continue; }

1 0

United States Naval Observatory time zone data available via browsers
by ado Oct. 28, 1995

Oct. 28, 1995

The URL is http://tycho.usno.navy.mil/tzones.html --ado

1 0

tzcode95f.tar.gz and tzdata95j.tar.gz
by ado Oct. 28, 1995

Oct. 28, 1995

These new files... -rw-r--r-- 1 ado 65981 Oct 28 16:57 tzcode95f.tar.gz -rw-r--r-- 1 ado 69781 Oct 28 16:57 tzdata95j.tar.gz ...are available by anonymous ftp from elsie.nci.nih.gov (165.112.73.1) in the ~ftp/pub directory. The files incorporate European cleanups circulated earlier on the time zone mailing list; they also provide support for 64-bit time_t's and optimization in localtime.c --ado

1 0

Re: old time zones (europe)
by Dik.Winter＠cwi.nl Oct. 25, 1995

Oct. 25, 1995

> There are some inconsistencies in the file: Indeed, but others than you are seeing. > Also, according to the Germany entry and to my vague memory, Germany > introduced DST in 1980 rather than 1981, which put pressure on all the > small ES countries to come to some kind of harmonisation. The file correctly notes that Germany introduced it in 1980 and Switzerland, Austria and Liechtenstein in 1981 (at least in version 95i). However there is a inconsistency in one of the lines for Luxembourg. It states: 1:00 Belgium MET%s 1979 while a similar line for Brussels states 1977. Should be changed in Luxembourg (the BeNeLux switched to "EC" rules in 1977). Further with respect to France. The file states that Shanks has no data on summertime in France in 1975 but that there is a note that states that summertime has been used. I just checked the Dutch railway timetable for summer 1975 and it says nothing about summer- time in France, so I presume no summertime. (In a later year, when it was unsure whether Germany would already switch to summer time until a very late stage the Dutch railway timetable came with double tables when Germany was involved. One for when summertime would be introduced and one when it would not be introduced.) The timetables are pretty good to establish in what years summertime was used; not for the exact starting and ending dates. dik -- dik t. winter, cwi, kruislaan 413, 1098 sj amsterdam, nederland, +31205924098 home: bovenover 215, 1025 jn amsterdam, nederland; e-mail: dik(a)cwi.nl

1 0

Re: old time zones (europe)
by bs＠cs.uni-sb.de Oct. 25, 1995

Oct. 25, 1995

Rolf-Martin Mantel writes: [...] > Also, according to the Germany entry and to my vague memory, Germany > introduced DST in 1980 rather than 1981, which put pressure on all the > small ES countries to come to some kind of harmonisation. That's correct. I have here a paper from the PTB (Physikalisch-Technische Bundesanstalt, Braunschweig - the responsible "time authority" for Germany) which states: 1980: Apr. 6 to Sept. 28 (!!) 1981: Mar. 29 to Sept. 27 ... Bernard Sieloff U of Saarbruecken, Germany

1 0

old time zones (europe)
by Mr. Rolf-Martin Mantel Oct. 25, 1995

Oct. 25, 1995

There are some inconsistencies in the file: 1.) # ... # The only [current nonmember using EU rules] I can speak for is Estonia, # which uses EU dates but not at 01:00 GMT, they use midnight GMT. I don't # think they know yet what they will do from 1996 onwards. # ... In the individual entries, it is clearly pointed out that Switzerland, Liechtenstein and Norway use EU rules since 1981 or 82. # ...the European time rules are...standardized since 1981, when # most European coun[tr]ies started DST. Before that year, only # a few countries (UK, France, Italy) had DST, each according # to own national rules. In 1981, however, DST started on # 'Apr firstSun', and not on 'Mar lastSun' as in the following # years... # But also since 1981 there are some more national exceptions # than listed in 'europe': Switzerland, for example, joined DST # one year later, Denmark ended DST on 'Oct 1' instead of 'Sep # lastSun' in 1981---I don't know how they handle now. # Also, according to the Germany entry and to my vague memory, Germany introduced DST in 1980 rather than 1981, which put pressure on all the small ES countries to come to some kind of harmonisation. Rolf

1 0

Re: Irish timezone error?
by ado Oct. 23, 1995

Oct. 23, 1995

> In Ireland, we change from BST to GMT on the 4th Sunday in October... > Is there a...version of these timezone files available, or a program > which would generate them properly? (I'm using Linux 1.3) The files available from anonymous ftp from elsie.nci.nih.gov (165.112.73.1) in the ~ftp/pub directory... -rw-r--r-- 1 ado 65805 Oct 13 13:24 tzcode95e.tar.gz -rw-r--r-- 1 ado 69404 Sep 26 10:44 tzdata95i.tar.gz ...reflect a fourth-Sunday-in-October switch in Dublin. --ado

1 0

Irish timezone error?
by Colm Buckley Oct. 23, 1995

Oct. 23, 1995

Hmm - I wonder have you seen this before? In Ireland, we change from BST to GMT on the 4th Sunday in October - ie: an appropriate TZ would be GMT0BST-1,M3.5.0/1,M10.4.0/1 (I'm not 100% certain that the changeover time is 1:00am; it could be 2:00). Anyway, the standard "Eire", "GB-Eire" and "Europe/Dublin" all seem to use M10.5.0 as the change-back date. Is there a newer version of these timezone files available, or a program which would generate them properly? (I'm using Linux 1.3) Thanks in advance, Colm. -- Colm Buckley | EMail: Colm.Buckley(a)tcd.ie Phone: (+353 1) 6082336 Computer Science Dept | WWW: http://isg.cs.tcd.ie/cbuckley/ Trinity College | finger cbuckley(a)isg.cs.tcd.ie for more info. Dublin 2, Ireland | "Too many windows, not enough brain."

1 0

Re: timezone stuff on 64 bit machines
by ado Oct. 23, 1995

Oct. 23, 1995

>> The fix is to use the fact that--at least for now--the calendar operates on >> a grand 2800-year cycle: any set of 2800 consecutive years contains exactly >> 1,022,679 days. A bit of modular arithmetic will get us where we want to be. > > To me this calls attention to a fundamental design flaw: any > computation of daylight savings time that involves looping a > billion times, or even a thousand, cannot be right. Under what > circumstances does this happen? The attached might be used to avoid thousands of loops. --ado SCCS/s.localtime.c: 7.45 vs. 7.48 *** 7.45/localtime.c Mon Oct 23 10:48:43 1995 --- 7.48/localtime.c Mon Oct 23 10:48:45 1995 *************** *** 1,6 **** #ifndef lint #ifndef NOID ! static char elsieid[] = "@(#)localtime.c 7.45"; #endif /* !defined NOID */ #endif /* !defined lint */ --- 1,6 ---- #ifndef lint #ifndef NOID ! static char elsieid[] = "@(#)localtime.c 7.48"; #endif /* !defined NOID */ #endif /* !defined lint */ *************** *** 1173,1191 **** if (tmp->tm_wday < 0) tmp->tm_wday += DAYSPERWEEK; y = EPOCH_YEAR; ! if (days >= 0) ! for ( ; ; ) { ! yleap = isleap(y); ! if (days < (long) year_lengths[yleap]) ! break; ! ++y; ! days = days - (long) year_lengths[yleap]; ! } ! else do { ! --y; ! yleap = isleap(y); ! days = days + (long) year_lengths[yleap]; ! } while (days < 0); tmp->tm_year = y - TM_YEAR_BASE; tmp->tm_yday = (int) days; ip = mon_lengths[yleap]; --- 1173,1190 ---- if (tmp->tm_wday < 0) tmp->tm_wday += DAYSPERWEEK; y = EPOCH_YEAR; ! #define LEAPS_THRU_END_OF(y) ((y) / 4 - (y) / 100 + (y) / 400) ! while (days < 0 || days >= (long) year_lengths[yleap = isleap(y)]) { ! register int newy; ! ! newy = y + days / DAYSPERNYEAR; ! if (days < 0) ! --newy; ! days -= (newy - y) * DAYSPERNYEAR + ! LEAPS_THRU_END_OF(newy - 1) - ! LEAPS_THRU_END_OF(y - 1); ! y = newy; ! } tmp->tm_year = y - TM_YEAR_BASE; tmp->tm_yday = (int) days; ip = mon_lengths[yleap];

1 0

Re: proposed fixes for Chile etc.
by Dik.Winter＠cwi.nl Oct. 22, 1995

Oct. 22, 1995

+ # Europa + # uninhabited + I find this amusing. Perhaps it is better to make it Ile Europe? dik -- dik t. winter, cwi, kruislaan 413, 1098 sj amsterdam, nederland, +31205924098 home: bovenover 215, 1025 jn amsterdam, nederland; e-mail: dik(a)cwi.nl

2 1

proposed zic.c change to cope with "32 < sizeof (time_t)"
by ado Oct. 22, 1995

Oct. 22, 1995

Attached is a proposed change to let zic.c cope gracefully when 32 < sizeof (time_t) I'll make it part of the ftpable stuff if nobody squawks in the next week. --ado SCCS/s.zic.c: 7.50 vs. 7.51 *** 7.50/zic.c Sat Oct 21 05:58:46 1995 --- 7.51/zic.c Sat Oct 21 05:58:47 1995 *************** *** 1,6 **** #ifndef lint #ifndef NOID ! static char elsieid[] = "@(#)zic.c 7.50"; #endif /* !defined NOID */ #endif /* !defined lint */ --- 1,6 ---- #ifndef lint #ifndef NOID ! static char elsieid[] = "@(#)zic.c 7.51"; #endif /* !defined NOID */ #endif /* !defined lint */ *************** *** 567,594 **** ifree(toname); } static void setboundaries P((void)) { register time_t bit; ! register int bii; for (bit = 1; bit > 0; bit <<= 1) ! continue; ! if (bit == 0) { /* time_t is an unsigned type */ ! tt_signed = FALSE; ! min_time = 0; ! max_time = ~(time_t) 0; if (sflag) ! max_time >>= 1; } else { ! tt_signed = TRUE; ! min_time = bit; ! max_time = bit; ! ++max_time; ! max_time = -max_time; if (sflag) ! min_time = 0; } min_year = TM_YEAR_BASE + gmtime(&min_time)->tm_year; max_year = TM_YEAR_BASE + gmtime(&max_time)->tm_year; --- 567,613 ---- ifree(toname); } + #define MAX_BITS_IN_FILE 32 + static void setboundaries P((void)) { register time_t bit; ! register int nbits; ! register int bii; + nbits = 0; for (bit = 1; bit > 0; bit <<= 1) ! ++nbits; ! tt_signed = (bit != 0); ! if (tt_signed) ! ++nbits; ! if (tt_signed) { ! if (nbits <= MAX_BITS_IN_FILE) { ! min_time = bit; ! max_time = bit; ! ++max_time; ! max_time = -max_time; ! } else { ! max_time = 1; ! max_time <<= (MAX_BITS_IN_FILE - 1); ! --max_time; ! min_time = max_time; ! --min_time; ! } if (sflag) ! min_time = 0; } else { ! min_time = 0; ! if (nbits <= MAX_BITS_IN_FILE) ! max_time = ~(time_t) 0; ! else { ! max_time = 1; ! max_time <<= MAX_BITS_IN_FILE; ! --max_time; ! } if (sflag) ! max_time >>= 1; } min_year = TM_YEAR_BASE + gmtime(&min_time)->tm_year; max_year = TM_YEAR_BASE + gmtime(&max_time)->tm_year;

2 1

proposed fixes for Chile etc.
by Paul Eggert Oct. 22, 1995

Oct. 22, 1995

I checked with Shanks, and he agrees with Eduardo Krell that Chile switches on the second Saturday at 24:00. I had transcribed Shanks's Chilean data incorrectly -- sorry about that. Here is a proposed patch that fixes the above bug, plus makes the following political changes that I have been accumulating: * Add an entry (or at least a comment) for every area that the ISO gives a country code to, even the uninhabited ones like Bassas da India. Report which areas are uninhabited (and thus whose time zone rules are pretty much undefined). * The Pacific Island Trust Territories have split up. Add entries for Pacific/Saipan (capital of N Mariana Is) and Pacific/Kosrae (now part of Micronesia). * ``Samoa'' is really two areas: American Samoa (US territory) and Western Samoa (an independent country). Add new entries Pacific/Pago_Pago and Pacific/Apia, one for each Samoa. Link the former to Pacific/Samoa for backward compatibility. Move the Samoan entries from northamerica to australasia, where they really belong. * Similarly, ``The Virgin Islands'' are really two possessions: the American Virgins and the British Virgins. Add new entries America/St_Thomas and America/Tortola, one for each possession. Link the former to America/Virgin for backward compatibility. * Aruba is considered to be a separate part of the Dutch realm from the Netherlands Antilles, so add an entry America/Aruba. * The main town in Mayotte is Mamoutzou, not Dzaoudzi. =================================================================== RCS file: RCS/africa,v retrieving revision 1995.8 retrieving revision 1995.8.0.1 diff -c -r1995.8 -r1995.8.0.1 *** africa 1995/09/21 15:26:25 1995.8 --- africa 1995/10/22 04:34:01 1995.8.0.1 *************** *** 118,123 **** --- 118,126 ---- 0:52 - LMT 1911 May 26 # Luanda Mean Time 1:00 - CAT + # Bassas da India + # uninhabited + # Benin # Whitman says they switched to 1:00 in 1946, not 1934; go with Shanks. # Zone NAME GMTOFF RULES FORMAT [UNTIL] *************** *** 248,253 **** --- 251,259 ---- 2:35 - AAMT 1936 May 5 # Addis Ababa MT 3:00 - EAT + # Europa + # uninhabited + # Gabon # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Africa/Libreville 0:37:48 - LMT 1912 *************** *** 273,278 **** --- 279,287 ---- Zone Africa/Accra -0:00:52 - LMT 1918 0:00 Ghana %s + # Glorioso Is + # uninhabited + # Guinea # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Africa/Conakry -0:54:52 - LMT 1912 *************** *** 286,291 **** --- 295,303 ---- 1:00 - CAT 1975 0:00 - WAT + # Juan de Nova + # uninhabited + # Kenya # From Paul Eggert <eggert(a)twinsun.com> (November 18, 1993): # Shanks says the transition to 2:45 was in 1940, but it must have been 1948. *************** *** 396,402 **** # Mayotte # Zone NAME GMTOFF RULES FORMAT [UNTIL] ! Zone Indian/Mayotte 3:01:08 - LMT 1911 Jul # Dzaoudzi 3:00 - EAT # Morocco --- 408,414 ---- # Mayotte # Zone NAME GMTOFF RULES FORMAT [UNTIL] ! Zone Indian/Mayotte 3:00:56 - LMT 1911 Jul # Mamoutzou 3:00 - EAT # Morocco *************** *** 452,458 **** # Reunion # Zone NAME GMTOFF RULES FORMAT [UNTIL] ! Zone Indian/Reunion 3:41:52 - LMT 1911 Jun # St Denis 4:00 - SMT # Rwanda --- 464,470 ---- # Reunion # Zone NAME GMTOFF RULES FORMAT [UNTIL] ! Zone Indian/Reunion 3:41:52 - LMT 1911 Jun # Saint-Denis 4:00 - SMT # Rwanda *************** *** 550,555 **** --- 562,570 ---- # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Africa/Lome 0:04:52 - LMT 1893 0:00 - WAT + + # Tromelin + # uninhabited # Tunisia # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S =================================================================== RCS file: RCS/antarctica,v retrieving revision 1995.9 retrieving revision 1995.9.0.1 diff -c -r1995.9 -r1995.9.0.1 *** antarctica 1993/11/22 20:23:15 1995.9 --- antarctica 1995/10/22 04:34:01 1995.9.0.1 *************** *** 9,19 **** # South Orkney Is # South Shetland Is - # Amsterdam Island # Bouvet ! # Crozet Is ! # Heard and McDonald Is ! # Kerguelen Is ! # St Paul Island # Peter I Island # Scott Island --- 9,25 ---- # South Orkney Is # South Shetland Is # Bouvet ! # uninhabited ! ! # French Southern and Antarctic Lands (no indigenous inhabitants) include ! # Amsterdam Island ! # Crozet Is ! # Kerguelen Is ! # St Paul Island ! ! # Heard Island, McDonald Is ! # uninhabited ! # Peter I Island # Scott Island =================================================================== RCS file: RCS/asia,v retrieving revision 1995.8 retrieving revision 1995.8.0.1 diff -c -r1995.8 -r1995.8.0.1 *** asia 1995/09/21 15:26:27 1995.8 --- asia 1995/10/22 04:34:01 1995.8.0.1 *************** *** 723,728 **** --- 723,731 ---- 2:00 Zion I%sT # This will undoubtedly change soon. + # Paracel Is + # no information + # Philippines # Howse writes (p 162) that until 1844 the Philippines kept American date. # The rest of this data is from Shanks. *************** *** 762,767 **** --- 765,773 ---- 7:20 - SGT 1950 7:30 - SGT 1982 May 8:00 - SGT + + # Spratly Is + # no information # Sri Lanka # Zone NAME GMTOFF RULES FORMAT [UNTIL] =================================================================== RCS file: RCS/australasia,v retrieving revision 1995.6 retrieving revision 1995.6.0.1 diff -c -r1995.6 -r1995.6.0.1 *** australasia 1995/08/29 00:46:56 1995.6 --- australasia 1995/10/22 04:34:01 1995.6.0.1 *************** *** 143,150 **** --- 143,155 ---- 7:00 - JVT # # Ashmore Is, Cartier + # no indigenous inhabitants; only seasonal caretakers # no information; probably like Australia/Perth # + # Coral Sea Is + # no indigenous inhabitants; only meteorologists + # no information + # # Macquarie, Manihiki, Penrhyn, Rakehanga # no information *************** *** 184,197 **** 10:00 - GST # Howland, Baker ! # no information; probably like Pacific/Samoa # Jarvis ! # no information; probably like Pacific/Kiritimati # Johnston # no information; probably like Pacific/Honolulu # Kiribati # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Pacific/Tarawa 11:32:04 - LMT 1901 # Bairiki --- 189,207 ---- 10:00 - GST # Howland, Baker ! # uninhabited since World War II ! # no information; was probably like Pacific/Pago_Pago # Jarvis ! # uninhabited since 1958 ! # no information; was probably like Pacific/Kiritimati # Johnston # no information; probably like Pacific/Honolulu + # Kingman + # uninhabited + # Kiribati # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Pacific/Tarawa 11:32:04 - LMT 1901 # Bairiki *************** *** 203,208 **** --- 213,248 ---- -10:40 - LIT 1979 Oct # Line Is Time -10:00 - THT + # N Mariana Is + # Zone NAME GMTOFF RULES FORMAT [UNTIL] + Zone Pacific/Saipan 9:43:00 - LMT 1901 + 9:00 - PLT 1969 Oct + 10:00 - GST + + # Marshall Is + # Zone NAME GMTOFF RULES FORMAT [UNTIL] + Zone Pacific/Majuro 11:24:48 - LMT 1901 + 11:00 - NCST 1969 Oct + 12:00 - NZST + Zone Pacific/Kwajalein 11:09:20 - LMT 1901 + 11:00 - NCST 1969 Oct + -12:00 - KJT 1993 Aug 20 + 12:00 - NZST + + # Micronesia + # Zone NAME GMTOFF RULES FORMAT [UNTIL] + Zone Pacific/Yap 9:12:32 - LMT 1901 # Colonia + 9:00 - PLT 1969 Oct + 10:00 - GST + Zone Pacific/Truk 10:07:08 - LMT 1901 + 10:00 - GST 1978 Oct + 11:00 - NCST + Zone Pacific/Ponape 10:32:52 - LMT 1901 # Kolonia + 11:00 - NCST + Zone Pacific/Kosrae 10:51:56 - LMT 1901 + 11:00 - NCST 1969 Oct + 12:00 - NZST + # Nauru # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Pacific/Nauru 11:07:40 - LMT 1921 Jan 15 # Uaobe *************** *** 273,303 **** 11:12 - NMT 1951 11:30 - NRFT - # Pacific Islands Trust Territories - # Zone NAME GMTOFF RULES FORMAT [UNTIL] - Zone Pacific/Majuro 11:24:48 - LMT 1901 - 11:00 - NCST 1969 Oct - 12:00 - NZST - Zone Pacific/Kwajalein 11:09:20 - LMT 1901 - 11:00 - NCST 1969 Oct - -12:00 - KJT 1993 Aug 20 - 12:00 - NZST - Zone Pacific/Truk 10:07:08 - LMT 1901 - 10:00 - GST 1978 Oct - 11:00 - NCST - Zone Pacific/Ponape 10:33:00 - LMT 1901 - 11:00 - NCST - Zone Pacific/Yap 9:12:24 - LMT 1901 - 9:00 - PLT 1969 Oct - 10:00 - GST - # Palau # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Pacific/Palau 8:57:56 - LMT 1901 # Koror 9:00 - PLT # Palmyra ! # no information; probably like Pacific/Kiritmati # Papua New Guinea # Zone NAME GMTOFF RULES FORMAT [UNTIL] --- 313,325 ---- 11:12 - NMT 1951 11:30 - NRFT # Palau # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Pacific/Palau 8:57:56 - LMT 1901 # Koror 9:00 - PLT # Palmyra ! # uninhabited since World War II; was probably like Pacific/Kiritmati # Papua New Guinea # Zone NAME GMTOFF RULES FORMAT [UNTIL] *************** *** 310,315 **** --- 332,351 ---- Zone Pacific/Pitcairn -8:40:20 - LMT 1901 # Adamstown -8:30 - PIT + # American Samoa + Zone Pacific/Pago_Pago 12:37:12 - LMT 1879 Jul 5 + -11:22:48 - LMT 1911 + -11:30 - SST 1950 + -11:00 - NST 1967 Apr # N=Nome + -11:00 - BST 1983 Nov 30 # B=Bering + -11:00 - SST # S=Samoa + + # W Samoa + Zone Pacific/Apia 12:33:04 - LMT 1879 Jul 5 + -11:26:56 - LMT 1911 + -11:30 - SST 1950 + -11:00 - SST + # Solomon Is # excludes Bougainville, for which see Papua New Guinea # Zone NAME GMTOFF RULES FORMAT [UNTIL] *************** *** 353,361 **** Zone Pacific/Wallis 12:15:20 - LMT 1901 12:00 - NZST - # Western Samoa - # See Pacific/Samoa in the `northamerica' file, of all places. - ############################################################################### # NOTES --- 389,394 ---- *************** *** 405,411 **** # - 9:00 GBT Gambier* # - 8:30 PIT Pitcairn* # ! # See the `northamerica' file for Hawaii and Samoa. # See the `southamerica' file for Easter I and the Galapagos Is. # # See the `africa' file for Zone naming conventions. --- 438,444 ---- # - 9:00 GBT Gambier* # - 8:30 PIT Pitcairn* # ! # See the `northamerica' file for Hawaii. # See the `southamerica' file for Easter I and the Galapagos Is. # # See the `africa' file for Zone naming conventions. *************** *** 780,788 **** # respect to the International Date Line, to rejoin its fellow islands, # going from 11:59 p.m. Thursday to 12:00 m. Saturday in a blink. ! # Pacific Islands Trust Territories # Howse writes (p 162) ``The Spaniards, on the other hand, reached the # Philippines and the Ladrones from America,'' and implies that the Ladrones # (now called the Marianas) kept American date for quite some time. # Ignore this for now, as we have no hard data. See also Asia/Manila. --- 813,828 ---- # respect to the International Date Line, to rejoin its fellow islands, # going from 11:59 p.m. Thursday to 12:00 m. Saturday in a blink. ! # N Mariana Is, Guam # Howse writes (p 162) ``The Spaniards, on the other hand, reached the # Philippines and the Ladrones from America,'' and implies that the Ladrones # (now called the Marianas) kept American date for quite some time. # Ignore this for now, as we have no hard data. See also Asia/Manila. + + # Samoa + + # Howse writes that in 1879 the King of Samoa decided to change + # ``the date in his kingdom from the Antipodean to the American system, + # ordaining -- by a masterpiece of diplomatic flattery -- that + # the Fourth of July should be celebrated twice in that year.'' =================================================================== RCS file: RCS/backward,v retrieving revision 1995.3 retrieving revision 1995.3.0.1 diff -c -r1995.3 -r1995.3.0.1 *** backward 1995/07/02 23:19:25 1995.3 --- backward 1995/10/22 04:34:01 1995.3.0.1 *************** *** 3,8 **** --- 3,9 ---- # This file provides links between current names for time zones # and their old names. Many names changed in late 1993. + Link America/St_Thomas America/Virgin Link Australia/Canberra Australia/ACT Link Australia/Lord_Howe Australia/LHI Link Australia/Sydney Australia/NSW *************** *** 51,56 **** --- 52,58 ---- Link America/Shiprock Navajo Link Pacific/Auckland NZ Link Pacific/Chatham NZ-CHAT + Link Pacific/Pago_Pago Pacific/Samoa Link Asia/Shanghai PRC Link Europe/Warsaw Poland Link Europe/Lisbon Portugal *************** *** 70,76 **** Link America/Detroit US/Michigan Link America/Denver US/Mountain Link America/Los_Angeles US/Pacific ! Link Pacific/Samoa US/Samoa Link Etc/UTC UTC Link Etc/Universal Universal Link Europe/Moscow W-SU --- 72,78 ---- Link America/Detroit US/Michigan Link America/Denver US/Mountain Link America/Los_Angeles US/Pacific ! Link Pacific/Pago_Pago US/Samoa Link Etc/UTC UTC Link Etc/Universal Universal Link Europe/Moscow W-SU =================================================================== RCS file: RCS/northamerica,v retrieving revision 1995.9 retrieving revision 1995.9.0.1 diff -c -r1995.9 -r1995.9.0.1 *** northamerica 1995/09/26 14:43:26 1995.9 --- northamerica 1995/10/22 04:34:01 1995.9.0.1 *************** *** 157,176 **** -5:00 - EST 1973 -5:00 US E%sT - # Samoa just changes names. No DST, per Naval Observatory. - # - # Howse writes that in 1879 the King of Samoa decided to change - # ``the date in his kingdom from the Antipodean to the American system, - # ordaining -- by a masterpiece of diplomatic flattery -- that - # the Fourth of July should be celebrated twice in that year.'' - - Zone Pacific/Samoa 12:37:12 - LMT 1879 Jul 5 - -11:22:48 - LMT 1911 - -11:30 - SST 1950 - -11:00 - NST 1967 Apr # N=Nome - -11:00 - BST 1983 Nov 30 # B=Bering - -11:00 - SST # S=Samoa - Zone Pacific/Midway -11:49:28 - LMT 1901 -11:00 - NST 1967 Apr # N=Nome -11:00 - BST 1983 Nov 30 # B=Bering --- 157,162 ---- *************** *** 193,199 **** -10:00 - HST # Navassa ! # no information; probably like America/New_York # Old names, for S5 users --- 179,185 ---- -10:00 - HST # Navassa ! # uninhabited # Old names, for S5 users *************** *** 730,736 **** -5:00 - EST # Clipperton ! # no information # Costa Rica # Shanks gives some very odd dates for 1991, and stops there. --- 716,722 ---- -5:00 - EST # Clipperton ! # uninhabited # Costa Rica # Shanks gives some very odd dates for 1991, and stops there. *************** *** 949,955 **** -5:00 - EST 1979 Apr 29 2:00 -5:00 US E%sT ! # Virgin Is (British and US) # Zone NAME GMTOFF RULES FORMAT [UNTIL] ! Zone America/Virgin -4:19:44 - LMT 1911 Jul # Charlotte Amalie -4:00 - AST --- 935,946 ---- -5:00 - EST 1979 Apr 29 2:00 -5:00 US E%sT ! # British Virgin Is ! # Zone NAME GMTOFF RULES FORMAT [UNTIL] ! Zone America/Tortola -4:18:28 - LMT 1911 Jul # Road Town ! -4:00 - AST ! ! # Virgin Is # Zone NAME GMTOFF RULES FORMAT [UNTIL] ! Zone America/St_Thomas -4:19:44 - LMT 1911 Jul # Charlotte Amalie -4:00 - AST =================================================================== RCS file: RCS/southamerica,v retrieving revision 1995.4 retrieving revision 1995.4.0.1 diff -c -r1995.4 -r1995.4.0.1 *** southamerica 1995/07/03 17:26:22 1995.4 --- southamerica 1995/10/22 04:34:01 1995.4.0.1 *************** *** 107,112 **** --- 107,118 ---- -4:00 Arg A%sT 1969 Oct 5 -3:00 Arg AR%sT + # Aruba + # Zone NAME GMTOFF RULES FORMAT [UNTIL] + Zone America/Aruba -4:40:24 - LMT 1912 Feb 12 # Oranjestad + -4:30 - NAST 1965 # Netherlands Antilles + -4:00 - AST + # Bolivia # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/La_Paz -4:32:36 - LMT 1890 *************** *** 199,219 **** # Chile ! # From Guy Harris: ! # The OAG lists October 11, 1987 and March 12, 1988 as the starting and ! # ending dates, giving them as "estimated date(s) based on previous year." ! ! # From Bob Devine (January 28, 1988): ! # Chile has had 2nd Sunday in October to 2nd Sunday in March DST since 1977. ! # Switch is at midnight. OAG is right. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Chile 1918 only - Sep 1 0:00 1:00 D Rule Chile 1919 only - Jul 2 0:00 0 S Rule Chile 1927 1931 - Sep 1 0:00 1:00 D Rule Chile 1928 1932 - Apr 1 0:00 0 S ! Rule Chile 1969 max - Oct Sun>=8 0:00 1:00 D ! Rule Chile 1970 max - Mar Sun>=8 0:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Santiago -4:42:40 - LMT 1890 -4:43 - SMT 1910 # Santiago Mean Time --- 205,221 ---- # Chile ! # From Eduardo Krell (1995-10-19): ! # The law says to switch to DST at midnight [24:00] on the second SATURDAY ! # of October.... The law is the same for March and October. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Chile 1918 only - Sep 1 0:00 1:00 D Rule Chile 1919 only - Jul 2 0:00 0 S Rule Chile 1927 1931 - Sep 1 0:00 1:00 D Rule Chile 1928 1932 - Apr 1 0:00 0 S ! Rule Chile 1969 max - Oct Sun>=9 0:00 1:00 D ! Rule Chile 1970 max - Mar Sun>=9 0:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Santiago -4:42:40 - LMT 1890 -4:43 - SMT 1910 # Santiago Mean Time *************** *** 334,340 **** -2:00 - FST # South Sandwich Is ! # no information # Suriname # Zone NAME GMTOFF RULES FORMAT [UNTIL] --- 336,342 ---- -2:00 - FST # South Sandwich Is ! # uninhabited # Suriname # Zone NAME GMTOFF RULES FORMAT [UNTIL]

1 0

Chile's time zone data wrong
by ekrell＠tandem.cl Oct. 19, 1995

Oct. 19, 1995

Hi, I don't know if you're the right person, but your name is in Sun's /usr/share/lib/zoneinfo/southamerica file under Solaris 2.4 I wanted you to know that the rule for Chile is incorrect. This year, we switched to DST at midnight on Saturday, October 14th, but the rules in that file made our Suns switch to DST a week earlier. The reason for the discrepancy is that the rules in the file specify to switch to DST at 0:00 the second sunday of October. This year, October 1st was a sunday, so the second Sunday was October 8th. That's when our Suns switched to DST. However, the law says to switch to DST at midnight on the second SATURDAY of October, which is not the same as 0:00 on the second sunday of October. In fact, this year the bug showed up because October 1st was a Sunday... May I suggest to change the specification to Sun>=9 ? Eduardo Krell R&D Manager Tandem Chile S.A. Santiago ekrell(a)tandem.cl

2 1

Re: timezone stuff on 64 bit machines
by ado Oct. 18, 1995

Oct. 18, 1995

> It fixed that problem form my point of view -- meaning it comes out > of that loop. Now there's a hang in outzone() in zic.c. At line 1439 > it loops over every year once again. Here the solution may be in the "setboundaries" code in "zic.c". Since time zone files only encode four characters (32 bits) worth of time_t values, min_time and max_time (and min_year and max_year) might be limited. A cheap way to hot-wire things for testing purposes is to add the lines min_time = 0; max_time = 0x7fffffffL; just above the min_year = TM_YEAR_BASE + gmtime(&min_time)->tm_year; max_year = TM_YEAR_BASE + gmtime(&max_time)->tm_year; lines in "setboundaries" This begs the question of whether the format of time zone files would need to be changed to allow for 64-bit time_t values (with attendant interoperability considerations). --ado

1 0

Re: timezone stuff on 64 bit machines
by ado Oct. 18, 1995

Oct. 18, 1995

If someone with a 64-bit machine could check out the attached (and if everyone could check out the logic), I'd love it. --ado SCCS/s.localtime.c: 7.45 vs. 7.46 *** 7.45/localtime.c Wed Oct 18 16:36:18 1995 --- 7.46/localtime.c Wed Oct 18 16:36:20 1995 *************** *** 1,6 **** #ifndef lint #ifndef NOID ! static char elsieid[] = "@(#)localtime.c 7.45"; #endif /* !defined NOID */ #endif /* !defined lint */ --- 1,6 ---- #ifndef lint #ifndef NOID ! static char elsieid[] = "@(#)localtime.c 7.46"; #endif /* !defined NOID */ #endif /* !defined lint */ *************** *** 1173,1178 **** --- 1173,1189 ---- if (tmp->tm_wday < 0) tmp->tm_wday += DAYSPERWEEK; y = EPOCH_YEAR; + #define CYCLE_DAYS 1022679L /* Any consecutive CYCLE_DAYS days... */ + #define CYCLE_YEARS 2800L /* make up exactly CYCLE_YEARS years. */ + if (days >= CYCLE_DAYS || days <= -CYCLE_DAYS) { + register int cycles; + + cycles = (days >= 0) ? + (days / CYCLE_DAYS) : + (-1 - (-1 - days) / CYCLE_DAYS); + days -= cycles * CYCLE_DAYS; + y += cycles * CYCLE_YEARS; + } if (days >= 0) for ( ; ; ) { yleap = isleap(y);

2 1

Re: timezone stuff on 64 bit machines
by ado Oct. 18, 1995

Oct. 18, 1995

> Is there a chance of getting a 64 bit fix for the timezone stuff? The fix is to use the fact that--at least for now--the calendar operates on a grand 2800-year cycle: any set of 2800 consecutive years contains exactly 1,022,679 days. A bit of modular arithmetic will get us where we want to be. --ado

2 1

timezone stuff on 64 bit machines
by Erik Troan Oct. 18, 1995

Oct. 18, 1995

I'm trying to use the timezone stuff you maintain on a 64-bit Alpha running Linux. It builds fine, but the timezone compiler has some trouble (it runs forever). I tracked the problem down to this code from timesub(): if (days >= 0) for ( ; ; ) { yleap = isleap(y); if (days < (long) year_lengths[yleap]) break; ++y; days = days - (long) year_lengths[yleap]; } else do { --y; yleap = isleap(y); days = days + (long) year_lengths[yleap]; } while (days < 0); This works fine when the number of years you can represent is small, as it is on a 32 bit machine. On a 64-bit machine however, you have some 4 billion more years to worry about. That means this loop has to repeat well over 4 billion times, and that takes 1 while :-) Any ideas on how to fix this? Erik ------------------------------------------------------------------------------- "Eggheads unite! You have nothing to lose but your yolks" - Adlai Stevenson Erik Troan = http://sunsite.unc.edu/ewt/ = ewt(a)sunsite.unc.edu

3 4

tzcode95e.tar.gz
by ado Oct. 13, 1995

Oct. 13, 1995

The mktime change to scan from future to past when trying to find time zone offsets has been incorporated into tzcode95e.tar.gz, available from elsie.nci.nih.gov (165.112.73.1) in the ~ftp/pub directory. --ado

1 0

time to retire SystemV?
by eggert＠twinsun.com Oct. 11, 1995

Oct. 11, 1995

I'm forwarding the following message originally posted to comp.os.linux.admin. Given the potential for confusion, perhaps it's time to comment-out the SystemV zones, for the same reason that US/Pacific-PET is commented out? Newsgroups: comp.os.linux.misc,comp.os.linux.admin From: davidsen(a)sixhub.tmr.com (Bill Davidsen) Subject: Who fixes broken timezones? Reply-To: davidsen(a)tmr.com (bill davidsen) Organization: TMR Associates, Schenectady NY Date: Mon, 9 Oct 1995 23:09:50 GMT Message-ID: <DG7F0G.43t(a)sixhub.tmr.com> This spring I posted a note saying the daylight savings time was broken. After about 50 people assured me it wasn't, Peter Dalgaard noted that "SystemV/EST5EDT" is broken and "US/Eastern" is not (guess most people use that). I just checked the 2.3 Slackware release and it's still wrong, so would someone tell me who maintains this so it can get fixed before it bites people again next year? Note: view the settings with zdump -v -c 2000 <filename> | grep 1995 the SystemV version thinks DST starts on the last Sunday of the month, not the first! -- TMR Associates, Inc +1 518 370-5654

1 0

Re: mktime problems when adjusting tm_isdst
by Bradley White Oct. 10, 1995

Oct. 10, 1995

I guess I would benefit from a succinct explanation of what mktime(`TZ=Europe/Paris 1996-01-01 00:00:00 tm_isdst=1') is supposed to mean. Paul's assertion ... > Surely the result should be > `1996-12-31 23:00:00 tm_isdst=0'. ..., while perhaps true, doesn't seem patently obvious. If I took `TZ=Europe/Paris 1995-06-01 00:00:00 tm_isdst=1', for example, and added six months, I think I would prefer to see a `1996-01-01 00:00:00 tm_isdst=0' mktime() result. In any case, a clear definition will tell us what to do with the code. Bradley

2 1

Re: mktime problems when adjusting tm_isdst
by ado Oct. 10, 1995

Oct. 10, 1995

We could run through the transition times from future to past (rather than from past to future as is done now); this doesn't solve the deep problem, but does get the "TZ=Europe/Paris 1996-01-01 00:00:00 tm_isdt=1" case right. It does have the advantage of being a fairly minimal change to the code. --ado *************** *** 1458,1467 **** if (sp == NULL) return WRONG; #endif /* defined ALL_STATE */ ! for (samei = 0; samei < sp->typecnt; ++samei) { if (sp->ttis[samei].tt_isdst != tmp->tm_isdst) continue; ! for (otheri = 0; otheri < sp->typecnt; ++otheri) { if (sp->ttis[otheri].tt_isdst == tmp->tm_isdst) continue; tmp->tm_sec += sp->ttis[otheri].tt_gmtoff - --- 1458,1467 ---- if (sp == NULL) return WRONG; #endif /* defined ALL_STATE */ ! for (samei = sp->typecnt - 1; samei >= 0; --samei) { if (sp->ttis[samei].tt_isdst != tmp->tm_isdst) continue; ! for (otheri = sp->typecnt - 1; otheri >= 0; --otheri) { if (sp->ttis[otheri].tt_isdst == tmp->tm_isdst) continue; tmp->tm_sec += sp->ttis[otheri].tt_gmtoff -

2 1