- tz - lists.iana.org

East-Saskatchewan -> Saskatchewan [forwarded with permission]
by ado Oct. 28, 1992

Oct. 28, 1992

> Date: Tue, 27 Oct 92 13:04:32 PST > From: eggert(a)twinsun.com (Paul Eggert) > Message-Id: <9210272104.AA11841(a)farside.twinsun.com> > To: ado(a)ncifcrf.gov > Subject: East-Saskatchewan -> Saskatchewan > > W. Jones <jones(a)skdad.usask.ca> writes that all of Saskatchewan is on CST year > round. I'll enclose his comments at the end of this message. So I propose > the following patch to the tz package: > > *** northamerica Thu Apr 23 10:34:33 1992 > --- northamerica.fix Tue Oct 27 12:58:38 1992 > *************** > *** 215,220 **** > --- 215,224 ---- > # be same as Pacific. Stick with rules posted in 1988 until more authoritative > # information is available. > > + # From W. Jones (October 27, 1992): > + # Since (I believe) around 1964, all of Saskatchewan is on CST year round, > + # by law [the U.S. Naval Observatory notwithstanding]. > + > # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S > Rule Canada 1969 max - Oct lastSun 2:00 0 S > Rule Canada 1969 1986 - Apr lastSun 2:00 1:00 D > *************** > *** 226,232 **** > Zone Canada/Atlantic -4:00 Canada A%sT > Zone Canada/Eastern -5:00 Canada E%sT > Zone Canada/Central -6:00 Canada C%sT > ! Zone Canada/East-Saskatchewan -6:00 - CST # No DST as of 1987 > Zone Canada/Mountain -7:00 Canada M%sT > Zone Canada/Pacific -8:00 Canada P%sT > Zone Canada/Yukon -9:00 Canada Y%sT > --- 230,236 ---- > Zone Canada/Atlantic -4:00 Canada A%sT > Zone Canada/Eastern -5:00 Canada E%sT > Zone Canada/Central -6:00 Canada C%sT > ! Zone Canada/Saskatchewan -6:00 - CST > Zone Canada/Mountain -7:00 Canada M%sT > Zone Canada/Pacific -8:00 Canada P%sT > Zone Canada/Yukon -9:00 Canada Y%sT > > > Here are Jones's comments: > > Date: Tue, 27 Oct 92 14:37:34 CST > From: jones(a)skdad.usask.ca (W. Jones) > To: eggert(a)twinsun.com > Cc: wright(a)sask.usask.ca > Organization: University of Saskatchewan, Saskatoon, Canada > > ... for this site, in order to follow local time correctly > requires choosing the ruleset named "East-Saskatchewan". Sounds > plausible, until you consider that the measure of places for which a > ruleset "Saskatchewan" or "West-Saskatchewan" applies is zero. > > How did this come about? Since (I believe) around 1964, all of > Saskatchewan is on CST year round, by law; but since that distinction > makes us rebels to North American convention, it tends to get left out of > discussions for simplicity. Thus when the timezone package appeared in > 1987, lo and behold we were lumped with the other Central places, as > implied in this table in the documentation: > > # From U. S. Naval Observatory (January 19, 1989): > # CANADA NEW FDL 3.5H BEHIND UTC ST.JOHN'S > # CANADA NEW FDL 1.5H BEHIND UTC APR 3 - OCT 29 > # CANADA ATLANTIC 4 H BEHIND UTC HALIFAX > # CANADA ATLANTIC 3 H BEHIND UTC APR 3 - OCT 29 > # CANADA EASTERN 5 H BEHIND UTC TORONTO, MONTREAL, OTTAWA > # CANADA EASTERN 4 H BEHIND UTC APR 3 - OCT 29 > # CANADA CENTRAL 6 H BEHIND UTC REGINA, WINNIPEG > # CANADA CENTRAL 5 H BEHIND UTC APR 3 - OCT 29 > # CANADA MOUNTAIN 7 H BEHIND UTC CALGARY, EDMONTON > # CANADA MOUNTAIN 6 H BEHIND UTC APR 3 - OCT 29 > # CANADA PACIFIC 8 H BEHIND UTC VANCOUVER > # CANADA PACIFIC 7 H BEHIND UTC APR 3 - OCT 29 > # CANADA YUKON SAME AS PACIFIC DAWSON > > I complained, and the eventual result was the adjustment below: > > # Zone NAME GMTOFF RULES/SAVE FORMAT [UNTIL] > # Bob Devine says that DST *is* observed in Newfoundland > Zone Canada/Newfoundland -3:30 Canada N%sT > Zone Canada/Atlantic -4:00 Canada A%sT > Zone Canada/Eastern -5:00 Canada E%sT > Zone Canada/Central -6:00 Canada C%sT > Zone Canada/East-Saskatchewan -6:00 - CST # No DST as of 1987 > Zone Canada/Mountain -7:00 Canada M%sT > Zone Canada/Pacific -8:00 Canada P%sT > > apparently the U.S. Naval Observatory was believed infallible, thus the > fellow from Saskatchewan must be following some kooky (and new) local > practice, thus we'll just throw him a sop....

1 0

US presidential election year politics help cause time zone bugs
by eggert＠twinsun.com Oct. 26, 1992

Oct. 26, 1992

Several people on the west coast of the US reported that their Unix systems failed to switch from daylight savings time to standard time yesterday, 26 October 1992. The reason? When they originally configured their systems, they were asked to choose one of the following time zone rules: US/Alaska US/Central US/Hawaii US/Pacific US/Aleutian US/East-Indiana US/Michigan US/Pacific-New US/Arizona US/Eastern US/Mountain US/Samoa ... Some people chose `US/Pacific-New' instead of `US/Pacific'. After all, who wants the old version when you can have the new version? Unfortunately, `US/Pacific-New' stands for ``Pacific Presidential Election Time'', which was passed by the House in April 1989 but never signed into law. In presidential election years, this rule would have delayed the PDT-to-PST switchover until after the election, to lessen the effect of broadcast news election projections on last-minute west-coast voters. Thus, US/Pacific-New and US/Pacific have always been identical -- until yesterday. This problem comes from combining Arthur David Olson's deservedly popular time zone software (which you can FTP from elsie.nci.nih.gov in pub/tz92b.tar.Z) with some overly terse vendor-supplied installation procedures. No doubt Olson did not use a more informative name like `US/Pacific-Presidential-Election' because of the 14-character file name length limit in many Unix file systems. In view of yesterday's experience, though, it seems unwise to make the hypothetical choice available under any name, since it gives free rein to Murphy's Law.

1 0

Re: strtotm
by guy＠auspex.com Sept. 16, 1992

Sept. 16, 1992

> strptime() [at least] falls into this category -- I believed Guy intended > it as the inverse of strptime() [or at least as close as you can get]. "Inverse of 'strftime()'". Yes, that was the basic idea. It does take a format parameter; however, it hasn't any magic way of figuring out whether 1/2/92 is January 2, 1992 or February 1, 1992 - you have to tell it how to interpret dates such as that, by passing it the right format string.

1 0

Re: strtotm
by Bradley White Sept. 16, 1992

Sept. 16, 1992

>> * Prior Art >> >> func who where >> ----------------------------------------------------- >> dparsetime() RAND/UCI MH >> getabsdate() Moraes C News >> getdate() USL Sys V Rel 4 >> getdate() Bellovin/Salz/Berets B News >> parsedate() Hamey/Accetta Mach >> partime() Harrenstien/Eggert RCS >> strptime() Harris SunOS 4.1[.x] > Are there any routines available that take a "format" parameter so that they > can scan dates that are in ambiguous formats. strptime() [at least] falls into this category -- I believed Guy intended it as the inverse of strptime() [or at least as close as you can get]. So, ... char * strptime(buf, fmt, tm) char *buf; char *fmt; struct tm *tm; strptime() converts the character string pointed to by buf to a time value, which is stored in the tm structure pointed to by tm, using the format specified by fmt. It is available under SunOS. > I would like to be able to scan dates in any of the following formats: > > M-D-Y T D-M-Y T There is no getting around that those D/M-numeric forms are just plain ambiguous. The best you can do is to use some ``internationalization'' feature like setlocale(LC_TIME, ...), or to just prefer one over the other. > I would like to be able to scan dates in any of the following formats: > > M-D-Y T D-M-Y T Y-M-D T > M/D/Y T D/M/Y T Y/M/D T > M D Y T D M Y T Y M D T > M.D.Y T D.M.Y T Y.M.D T Opinion: Any generalized date parser is overkill for these restricted situations. Brad

1 0

Re: strtotm
by Monty Solomon Sept. 16, 1992

Sept. 16, 1992

> Date: Mon, 22 Jun 92 22:42:18 -0400 (EDT) > From: Bradley White <bww+(a)CS.CMU.EDU> > > * Prior Art > > Perhaps a reasonable candidate already exists, or can be grown from > existing code. These routines (in alphabetical order) have been > mentioned: > > func who where > ----------------------------------------------------- > dparsetime() RAND/UCI MH > getabsdate() Moraes C News > getdate() USL Sys V Rel 4 > getdate() Bellovin/Salz/Berets B News > parsedate() Hamey/Accetta Mach > partime() Harrenstien/Eggert RCS > strptime() Harris SunOS 4.1[.x] Where are these various routines available? Are there any routines available that take a "format" parameter so that they can scan dates that are in ambiguous formats. I would like to be able to scan dates in any of the following formats: M-D-Y T D-M-Y T Y-M-D T M/D/Y T D/M/Y T Y/M/D T M D Y T D M Y T Y M D T M.D.Y T D.M.Y T Y.M.D T Where: M is the month expressed as 3, Mar, MAR, or March Y is the year expressed as 92 or 1992 T is optional and is the time (12- or 24-hour format) Thanks. Monty --- # Monty Solomon / PO Box 2486 / Framingham, MA 01701-0405 # monty%roscom(a)think.com

1 0

problem with ltzset()
by Joel.Tornatore＠Eng.Sun.COM Sept. 9, 1992

Sept. 9, 1992

I am seeing a problem with _ltzset near daylight savings time when non-POSIX TZ values are used. Here is the output of a program that locates discontinuities in the time (seconds since epoch are increased by 3600; check if hour increased by one): gap % tzcheck TZ=PST8PDT 720005582: 9/25/92 1:33 733919582: 3/4/93 3:33 752059982: 9/31/93 1:33 ^C gap % setenv TZ US/Pacific gap % tzcheck TZ=US/Pacific 720034366: 9/25/92 9:32 733948366: 3/4/93 11:32 752088766: 9/31/93 9:32 It's off by the GMT offset, 8 hours, in every case. If EST5EDT and US/Easter are used, it's off by 5 hours. Has anyone seen or fixed this problem? This is on SVR4. joel tornatore jrt(a)eng.sun.com

2 1

test
by Glenn.Wright＠UK.Sun.COM Sept. 9, 1992

Sept. 9, 1992

please ignore.

1 0

Re: strtotm
by ado July 7, 1992

July 7, 1992

Two general questions about the high-test end of the string-to-time function spectrum: 1. What's the right way to handle the duration of events (such as "Lent" or "high tea")? 2. What calendar-keeping software out in the field--be it UNIX-, DOS-, Mac-, or whatever-else-based, can be used to get an idea of the capabilities that might be useful in a high-test function? --ado

2 1

Re: functions to convert ASCII time
by abe＠mace.cc.purdue.edu June 23, 1992

June 23, 1992

Brad, I've been reading this discussion with interest, because I had to create a function some time ago to convert mail header date/time strings to time_t values. While my function represents another possible entry in your list, I don't think it answers all the criteria that this discussion has established. For example, it returns a time_t, not a struct tm, and it doesn't return a character pointer to the end ot its scan (although it could). The major contribution of this function might be the list of time zone abbreviations that it understands. Paul Eggert has helped me assemble (and purify) it. :-) I will be glad to contribute my atotm() function to the list of prior art, for whatever value it might have. Vic Abell

1 0

1st of 2 forwarded messages from Bradley White: time2posix/posix2time
by ado June 23, 1992

June 23, 1992

The attached comes from Bradley White, bww+(a)cs.cmu.edu. Your comments will be appreciated! --ado I have found [time2posix() and posix2time()] invaluable when dealing with things like external time formats, or programs that want to wake up every minute (as opposed to every 60 seconds), on systems that allow for leap seconds.... /* ** IEEE Std 1003.1-1988 (POSIX) legislates that 536457599 ** shall correspond to "Wed Dec 31 23:59:59 GMT 1986", which ** is not the case if we are accounting for leap seconds. ** So, we provide the following conversion routines for use ** when exchanging timestamps with POSIX conforming systems. */ static long leapcorr(timep) time_t * timep; { register struct state * sp; register struct lsinfo * lp; register int i; if (!lcl_is_set) (void) tzset(); sp = lclptr; i = sp->leapcnt; while (--i >= 0) { lp = &sp->lsis[i]; if (*timep >= lp->ls_trans) return lp->ls_corr; } return 0L; } time_t time2posix(t) time_t t; { return t - leapcorr(&t); } time_t posix2time(t) time_t t; { time_t x, y; /* ** For a positive leap second hit, the result ** is not unique. For a negative leap second ** hit, the corresponding time doesn't exist, ** so we return an adjacent second. */ x = t + leapcorr(&t); y = x - leapcorr(&x); if (y < t) { do { x++; y = x - leapcorr(&x); } while (y < t); if (t != y) return x-1; } else if (y > t) { do { --x; y = x - leapcorr(&x); } while (y > t); if (t != y) return x+1; } return x; }

2 1