August 1997 - tz - lists.iana.org

Re: bug in mktime() normalization?
by Paul Eggert Aug. 21, 1997

Aug. 21, 1997

Date: Tue, 19 Aug 97 14:13:13 -0400 From: Tom Peterson (USG) <tomp(a)zk3.dec.com> In mktime(), or rather time2(), I noticed that a change was made a while back which removed the following seconds normalization: ! if (yourtm.tm_sec >= SECSPERMIN + 2 || yourtm.tm_sec < 0) ! normalize(&yourtm.tm_min, &yourtm.tm_sec, SECSPERMIN); Instead, these seconds are now saved aside into saved_seconds and only added back in after an appropriate match is found using the remaining normalized data. This was to handle leap seconds properly. On systems that support leap seocnds, it's incorrect to normalize the seconds count first, since you don't know how many seconds there are per minute until you determine which minutes you're talking about. For example, with leap second support, mktime should adjust an input value of ``1997-07-01 00:00:-1'' UTC (i.e. tm_sec == -1) so that it becomes 1997-06-30 23:59:60 UTC, but with the code above, mktime generated 1997-06-30 23:59:59 UTC which is incorrect. Here's what UNIX98 has to say regarding this: >From your quote it appears that UNIX98 says the same thing that the C Standard says, which is to say, not much. The C Standard spec for mktime is terribly ambiguous and you've found one of the ambiguities. The C Standard says that mktime can return -1 if attempted in the ``spring forward gap'', but it doesn't say how mktime should determine whether one is in the gap. Any user program that messes in this area is relying on behavior that is not guaranteed by any standard. For what it's worth, the GNU C library's mktime agrees with the tz mktime on your example, whereas Solaris 2.5.1 and BSD/OS 3.0 have the other reasonable interpretation (i.e. both mktime invocations yield 575002800).

3 2

bug in mktime() normalization?
by Tom Peterson Aug. 19, 1997

Aug. 19, 1997

Hi all, I've been looking over the time zone package and testing various scenerios. In mktime(), or rather time2(), I noticed that a change was made a while back which removed the following seconds normalization: ! if (yourtm.tm_sec >= SECSPERMIN + 2 || yourtm.tm_sec < 0) ! normalize(&yourtm.tm_min, &yourtm.tm_sec, SECSPERMIN); Instead, these seconds are now saved aside into saved_seconds and only added back in after an appropriate match is found using the remaining normalized data. The question I have is whether this could lead to erroneous rejection of certain tm structs. For example: If the values (other than tm_sec) of an incoming tm struct result in no match from time2()'s binary search, such as a time between a dst change, the struct is rejected and the function returns WRONG. However, the value of tm_sec may have been large enough to push the rest of my tm struct into an acceptible range where a match could have occurred. As with other tm struct members, the value of tm_sec before normalization is not restricted to the range specified in <time.h>. Here's what UNIX98 has to say regarding this: > The mktime() function converts the broken-down time, expressed as > local time, in the structure pointed to by timeptr, into a time since > the Epoch value with the same encoding as that of the values returned > by time(). The original values of the tm_wday and tm_yday components > of the structure are ignored, and the original values of the other > components are not restricted to the ranges described in the <time.h> > entry. > > ... > > Upon successful completion, the values of the tm_wday and tm_yday > components of the structure are set appropriately, and the other > components are set to represent the specified time since the Epoch, > but with their values forced to the ranges indicated in the <time.h> > entry; the final value of tm_mday is not set until tm_mon and tm_year > are determined. Here's an example program: #include <stdio.h> #include <time.h> #include <stdlib.h> void display_tm(void); void run_test(); struct tm tm; time_t mktime_returned; #define TZ_value "ABC0DEF,J81,J300" #define TM_year 1988 #define TM_mon 3 #define TM_mday 22 #define TM_hour1 2 #define TM_hour2 4 #define TM_min 0 #define TM_sec1 7200 #define TM_sec2 0 #define TM_wday -1 #define TM_yday -1 #define TM_isdst -1 main() { putenv("TZ=ABC0DEF,J81,J300"); printf("TZ = %s\n\n", getenv("TZ")); /* same time as below, but tm_hour and tm_sec have not yet been * normalized */ tm.tm_year = TM_year - 1900; tm.tm_mon = TM_mon - 1; tm.tm_mday = TM_mday; tm.tm_hour = TM_hour1; /* un-normalized hour */ tm.tm_min = TM_min; tm.tm_sec = TM_sec1; /* un-normalized seconds */ tm.tm_wday = TM_wday; tm.tm_yday = TM_yday; tm.tm_isdst = TM_isdst; run_test(); printf("\n"); /* same time as above, but tm_hour and tm_sec have already been * normalized */ tm.tm_year = TM_year - 1900; tm.tm_mon = TM_mon - 1; tm.tm_mday = TM_mday; tm.tm_hour = TM_hour2; /* pre-normalized hour */ tm.tm_min = TM_min; tm.tm_sec = TM_sec2; /* pre-normalized seconds */ tm.tm_wday = TM_wday; tm.tm_yday = TM_yday; tm.tm_isdst = TM_isdst; run_test(); } void run_test(void) { printf("tm struct set to:\n"); display_tm(); if ((mktime_returned = mktime(&tm)) == (time_t)-1) { (void)puts("mktime returned -1 (-unknown-)"); } else { printf("mktime returned value = %d\n", (int) mktime_returned); } printf("mktime() final tm struct reset to:\n"); display_tm(); } void display_tm(void) { printf("tm_year = %d, tm_mon = %d, tm_mday = %d, tm_hour = %d, \ tm_min = %d, sec = %d,\nwday = %d, yday = %d, isdst = %d\n", tm.tm_year, tm.tm_mon, tm.tm_mday, tm.tm_hour, tm.tm_min, tm.tm_sec, tm.tm_wday, tm.tm_yday, tm.tm_isdst); } output ====== $ a.out TZ = ABC0DEF,J81,J300 tm struct set to: tm_year = 88, tm_mon = 2, tm_mday = 22, tm_hour = 2, tm_min = 0, sec = 7200, wday = -1, yday = -1, isdst = -1 mktime returned -1 (-unknown-) mktime() final tm struct reset to: tm_year = 88, tm_mon = 2, tm_mday = 22, tm_hour = 2, tm_min = 0, sec = 7200, wday = -1, yday = -1, isdst = -1 tm struct set to: tm_year = 88, tm_mon = 2, tm_mday = 22, tm_hour = 4, tm_min = 0, sec = 0, wday = -1, yday = -1, isdst = -1 mktime returned value = 575002800 mktime() final tm struct reset to: tm_year = 88, tm_mon = 2, tm_mday = 22, tm_hour = 4, tm_min = 0, sec = 0, wday = 2, yday = 81, isdst = 1 $ I would very much appreciate any thoughts concerning this matter and any possible solutions. thanks, - Tom ===================================================================== Tom Peterson | DIGITAL UNIX Development Environment Digital Equipment Corporation | Phone:(603)884-7550 110 Spit Brook Road ZKO3-2/W17 | FAX:(603)881-2257 Nashua, NH 03062-2698 | Email: mailto:tomp@zk3.dec.com

1 0

Re: FW: tztab entries (forwarded with permission)
by Dave Kelly Aug. 13, 1997

Aug. 13, 1997

>From : x>eggert(a)twinsun.com To : dkelly(a)lhr-sys.DHL.COM Subject : Re: FW: tztab entries (forwarded with permission) CC : tz(a)elsie.nci.nih.gov Date : 13/08/97 00:46 Hi Paul Thanks for your help on this. > Here's what I think you're asking for. If you're interested only in > times now and in the future, then the Olson database contains too many > names, partly because it goes back to 1970 and thus must track > historical changes, and partly because it insists on at least one name > per country. You want just one distinct name for each time zone > future, regardless of country. Yes, this is exactly what we need. > This script assumes that the only thing that > matters about a time zone is its UTC offset; it ignores the time zone > name (e.g. "EST", "EDT") and whether the time is daylight-saving time. Just to reassure me - I assume that futures with the same UTC offsets but with different daylight savings time change dates are considered to be distinct by the script? Can you offer any guidelines on the allocation of canonical names to these futures? The tztab entry on HP-UX does seem to follow a convention of <zone name><offset><dst zone name>, e.g.: GMT0BST, WET0WETDST, PST8PDT, CST-9:30CDT but there appear to be different variations of these, e.g. for Central Europe I could use CET-2CEST or CET-2CETDST (and that's without considering MEDST, MEST, MESZ, METDST). Also "CST" is used in China/Australia/US. What would you recommend as the most common set? Thanks again for the help Dave Kelly

2 1

FW: tztab entries (forwarded with permission)
by Olson, Arthur David Aug. 13, 1997

Aug. 13, 1997

---------- From: Dave Kelly[SMTP:dkelly@lhr-sys.DHL.COM] Sent: Monday, August 11, 1997 2:30 PM To: ibilling(a)lhr-sys.DHL.COM; ado(a)elsie.nci.nih.gov Subject: tztab entries Hello There I wonder if you might be able to help... We are building a reference database in which one of the attributes of a "site" is the timezone in which the site is located. We decided to use the HP-UX timezone name from the tztab file as a reference to a timezone since it seemed as good as anything else. We have now realised that the entries we have as default in tztab are far from complete. Do you know where we could get hold of a tztab file which has a complete list of entries? We are only interested in modern (future) entries, but we are interested in the whole global set. We do not want to maintain multiple tztab entries for countries which use the same zone - i.e. we want to have a single entry for all the countries which use CET/CEST. I have found the tzcode and tzdata files but these do not link to a particular "name" of a timezone such as GMT0BST or CET-1CEST. It would obviously make sense to use something that already exists. Hope you can help or offer some advice Thank you and regards Dave Kelly DHL Worldwide Express dkelly(a)lhr-sys.dhl.com

3 2

You Simply Must Check This Out!
by iai＠t-1net.com Aug. 12, 1997

Aug. 12, 1997

/////////////////////////////////////////////////////////////////////////////// If you wish to be removed from this advertiser's future mailings, please reply with "Remove" in the Sbject line and this software will automatically block you from their future mailings. //////////////////////////////////////////////////////////////////////////////// Are you interested in a unique home business marketing a Brand Name Mini Digital Satellite Dish System? These are just a few of the benefits: ----------------------------------------------------- 1) Marketing a Brand Name Mini Digital Satellite Dish. 2) Commissions on products purchased via the company Digital Satellite TV Channel. 3) Your Personal Web page for YOUR BUSINESS. 4) Subscribe to the company's ISP. Earn commssions here too! 5) Your customers can purchase over 500 products via satellite TV, internet and mail order. COMING SOON! OVER 15,000 products...Guaranteed lowest price! ------------------------------------------------------------------- This a truly a great opportunity! Do yourself a favor and inquire for all of the details! You simply must check this out! ================================================== For More Infomation ================================================== Click Reply and Type "TPN INFO" info the Subject line then click Send. IMPORTANT: To receive this additional information "TPN INFO" must be found in the Subject line! Have a great day! :-)

1 0

Millennium update: Caroline Islands to be renamed
by Eric Ulevik Aug. 12, 1997

Aug. 12, 1997

>From a story in the Sydney Morning Herald 12 August 1997: * The Government of Kiribati is going to rename the Carolines to Millennium * The Old Royal Observatory in Greenwich says the Millennium arrives first at the Balleny Islands off Antartica - 2h2m before the Carolines Regards, Eric Ulevik

1 0

FW: Eighth EC Directive approved
by Olson, Arthur David Aug. 6, 1997

Aug. 6, 1997

---------- From: peter@aldie.co.uk[SMTP:peter@aldie.co.uk] Sent: Wednesday, August 06, 1997 2:38 PM To: tz(a)elsie.nci.nih.gov Subject: Eighth EC Directive approved I now have a copy of the eighth summer time directive, as published in the Official Journal. It is titled: Eighth Directive 97/44/EC of the European Parliament and of the Council of 22 July 1997 on summer-time arrangements. It runs for 4 years, 1998--2001, and confirms the current rules of last Sunday in March to last Sunday in October, ie: 1998: 29 March to 25 October 1999: 28 March to 31 October 2000: 26 March to 29 October 2001: 25 March to 28 October (it does give the actual dates). The directive does not apply in overseas territories of the Memmber States. It says the Commision should produce a proposal for 2002 and beyond by 1 Jan 2000 and this should be adopted by 1 Jan 2001. I doubt that this will happen though. This Eighth Directive was supposed to be adopted by 1 Jan 1997 and I doubt that summer-time arrangements will be top of the Commission's priority list as the new millennium dawns. There is no mention of the French desire to abandon the whole idea. France has had a change of government recently so maybe it will be quietly dropped. More and more EU stuff is now on the WWW but I can't find this yet. It may be that Directives won't appear in order to make you buy the Official Journal. It may appear at the European Parliament's site (http://www.europarl.eu.int/, select language, Subject Index, Texts Adopted: Provisional edition, and search) but it only goes up to 18 July at present. Peter Ilieve peter(a)aldie.co.uk

1 0

Eighth EC Directive approved
by peter＠aldie.co.uk Aug. 6, 1997

Aug. 6, 1997

I now have a copy of the eighth summer time directive, as published in the Official Journal. It is titled: Eighth Directive 97/44/EC of the European Parliament and of the Council of 22 July 1997 on summer-time arrangements. It runs for 4 years, 1998--2001, and confirms the current rules of last Sunday in March to last Sunday in October, ie: 1998: 29 March to 25 October 1999: 28 March to 31 October 2000: 26 March to 29 October 2001: 25 March to 28 October (it does give the actual dates). The directive does not apply in overseas territories of the Memmber States. It says the Commision should produce a proposal for 2002 and beyond by 1 Jan 2000 and this should be adopted by 1 Jan 2001. I doubt that this will happen though. This Eighth Directive was supposed to be adopted by 1 Jan 1997 and I doubt that summer-time arrangements will be top of the Commission's priority list as the new millennium dawns. There is no mention of the French desire to abandon the whole idea. France has had a change of government recently so maybe it will be quietly dropped. More and more EU stuff is now on the WWW but I can't find this yet. It may be that Directives won't appear in order to make you buy the Official Journal. It may appear at the European Parliament's site (http://www.europarl.eu.int/, select language, Subject Index, Texts Adopted: Provisional edition, and search) but it only goes up to 18 July at present. Peter Ilieve peter(a)aldie.co.uk

1 0

comments on DHCP time zone draft (draft-ietf-dhc-timezone-03.txt)
by Paul Eggert Aug. 1, 1997

Aug. 1, 1997

Here are some comments on your document: DHCP Option for IEEE 1003.1 POSIX Timezone Specifications ftp://ds.internic.net/internet-drafts/draft-ietf-dhc-timezone-03.txt (1997-07-30) * The draft allows only time zone settings that conform to POSIX.1. However, in practice, these settings are not enough to satisfy many real-world applications, for the following reasons: - Due to politics, time zone rules change with time, and POSIX settings cannot capture these changes. For example, last year the time zone in Sri Lanka changed from +0530 to +0630 on May 25, and then to +0600 on Oct 26. POSIX rules cannot represent the time in Sri Lanka last year. - Some countries use periodic rules that cannot be represented by POSIX rules. Examples include Israel (which uses rules based on the Jewish calendar) and Iran (the Persian calendar). Again, POSIX rules cannot represent changes like these, except for one or two transitions at a time. - Many hosts use the Olson extension to POSIX.1 <ftp://elsie/nci/nih/gov/pub/tz/> The draft gives no way to specify time zone rules using the Olson extension, and so these hosts cannot be configured as desired. The Olson time zone settings are upward compatible with POSIX.1. They are widely used in practice: the source code and tables are public domain, and they are supported by many types of hosts, including BSD, HPUX, Irix, Linux, Network Appliance, Sun, and Unixware. Therefore I propose that the draft be changed to also allow time zone settings that conform to the Olson extension to POSIX.1. Perhaps the DHCP server could reply with both an Olson setting (for clients that support the Olson extension) and a POSIX.1 setting (for clients that don't). * There are several discrepancies between the latest POSIX.1 spec for TZ format and the DHCP time zone draft. They should be removed by copying the latest POSIX.1 spec's wording more carefully. Here are some example discrepancies. - The `,' and `[' before `start' should be interchanged. In other words, `stdoffset[dst[offset],[start[/time],end[/time]]]' should be `stdoffset[dst[offset][,start[/time],end[/time]]]'. - The current POSIX spec places an upper bound of {TZNAME_MAX} on the length of a time zone designation like "EST"; the draft does not have this limit. - The current POSIX spec does not allow null characters in a time zone designation. - The current POSIX spec allows hour=24. - The current POSIX spec explicitly says that day 0 is Sunday. * The example for the Eastern USA time zone, 1986 is unrealistic. In practice, people want to avoid changing their time zone settings every year, so they rarely use Julian day rules. Here is a better set of examples (subject to change at any time by political events): Eastern USA from 1987: EST5EDT,M4.1.0,M10.5.0 EST (-0500) during winter EDT (-0400) during summer summer time from the 1st Sunday in April at 02:00 to the last Sunday in October at 02:00 Central European Union from 1996: CET-1CEST,M3.5.0/1,M10.5.0/2 CET (+0100) during winter CEST (+0200) during summer summer time from the last Sunday in March at 01:00 to the last Sunday in October at 02:00 Arizona from 1968: MST7 MST (-0700) all year Tasmania from 1991: EST-10EST,M10.1.0/2,M3.5.0/3 EST (+1000) during winter (``Eastern Standard Time'') EST (+1100) during summer (``Eastern Summer Time'') summer time from the the first Sunday in October at 02:00 to the last Sunday in March at 03:00 Cook Islands from 1979: CKT-10CKST-10:30,M10.5.0/0,M3.1.0/0 CKT (+1000) during winter CKST (+1030) during summer summer time from the first Sunday in October at 00:00 to the first Sunday in March at 00:00 Line Islands from 1995: LINT-14 LINT (+1400) all year If you accept my suggestion for the Olson extension, here are some more examples (which should be valid indefinitely): Eastern USA: America/New_York Central European Union: Europe/Brussels Arizona: America/Phoenix Tasmania: Australia/Hobart Cook Islands: Pacific/Rarotonga Line Islands, Kiribati: Pacific/Kiritimati * Please consider adding the following references: ftp://ds.internic.net/internet-drafts/draft-newman-datetime-01.txt contains a proposed procedure for IANA registration of time zone names, and uses the Olson extension as its basis. ftp://elsie/nci/nih/gov/pub/tz/ contains public-domain code and data for the Olson extension to the POSIX time zone strings. The current version of POSIX.1 is: ISO/IEC 9945-1: 1996 (ANSI/IEEE Std 1003.1, 1996) Information technology -- Portable Operating System Interface (POSIX) -- Part 1: System Application Program Interface (API) [C Language] (1996-07-12), section 8.1.1, Extensions to Time Functions * Terminology. The usual terms are `time zone' (not `timezone') and `daylight-saving time' (not `daylight savings time'). The draft should be reworked to use the usual terms. POSIX.1 currently uses `summer time' (the British term) instead of `daylight-saving time' (the American term). Whichever term is used, the draft should mention its equivalence to the other term.

3 2