FW: Question about: TWiki.org Service: Date and Time Gateway

Neither Peter nor Krystian is on the time zone mailing list; direct replies appropriately. --ado -----Original Message----- From: Peter Thoeny [mailto:peter.thoeny@attglobal.net] Sent: Sunday, February 13, 2005 6:23 PM To: krystian@mindspring.com; tz@lecserver.nci.nih.gov Subject: Re: Question about: TWiki.org Service: Date and Time Gateway Dear Krystian and tz team, The program on TWiki.org takes the data from the Olson tz database, http://www.twinsun.com/tz/tz-link.htm I am not sure I can explain the +/- GMT question, someone from the Olson tz database team is probably better in that. More info on GMT at http://en.wikipedia.org/wiki/GMT Info on timezones at http://wwp.greenwichmeantime.com/info/timezone.htm Interstingly, above site shows GMT times in reverse to the Olson database! E.g. for me in California, the Olson database tells me I am in Etc/GMT+8, and greenwichmeantime.com tells me I am GMT-8. Regards, Confused Peter krystian@mindspring.com wrote:
This question pertains to the data within:
TWiki.org Service: Date and Time Gateway http://twiki.org/cgi-bin/xtra/tzdate
Which I believe is under a revision by your penmanship.
Q. Are the GMT offsets mixed in this website? When I think of GMT-5 I think of Eastern Standard Time, however your site lists GMT-5 as where I always think GMT+5 is.
Greenwich Mean Time: (GMT/UCT/UTC/Universal/Zulu; positive numbers count hours west of GMT
What is the cause of having positive numbers count west of GMT? Aren't positive numbers, by definition, to be east of GMT? (like GMT+5 is add 5 hours to the time at GMT?)
I've seen lots of confusion about this on the web and I don't understand why this is so. I have always thought GMT-5 was Eastern Standard Time, not some time in the European longitudes.
Can you please speculate on this? I'm also trying to find the difference between UTC and GMT (as I've read they are the same but not exactly). Maybe this is how they differ.
I'm terribly confused. Please offer an explanation to your readers in the form of a link or direct them to such a place.
Regards, Roland Krystian Alberciak.
-- * Peter Thoeny Peter@Thoeny.com * Is your team already TWiki enabled? http://TWiki.org * This e-mail is: (_) public (x) ask first (_) private

This confusion basically stems from twiki's adoption of the tz convention, which in turn comes from POSIX, which in turn is backwards from the normal. I'll add something like the following text to tz-link.htm. I don't see any simple fix to the twiki.org page, other than a footnote containing the following info (which would probably cause more confusion than it'd cure). <li>Numeric time zone abbreviations typically count hours east of UTC, e.g., <code>+0900</code> for Japan and <code>-1000</code> for Hawaii. However, the <abbr>POSIX</abbr> <code>TZ</code> environment variable uses the reverse convention. For example, one might use <code>TZ="JST-9"</code> and <code>TZ="HST10"</code> for Japan and Hawaii, respectively. One should never set <code>TZ</code> to a value like <code>"GMT+5"</code>, since this would mean local time is five hours behind UTC and the time zone is called "GMT". If the <code>tz</code> database is available, it is usually better to use settings like <code>TZ="Asia/Tokyo"</code> and <code>TZ="Pacific/Honolulu"</code> instead, as this should avoid confusion, handle old timestamps better, and insulate you better from any future changes to the rules.</li>

Date: Mon, 14 Feb 2005 09:18:10 -0500 From: "Olson, Arthur David (NIH/NCI)" <olsona@dc37a.nci.nih.gov> Message-ID: <75DDD376F2B6B546B722398AC161106C740377@nihexchange2.nih.gov> To expand a little on Paul Eggert's response... | I am not sure I can explain the +/- GMT question, someone from the Olson tz | database team is probably better in that. There is no absolute right or wrong here, there are two different times, local time (wherever) and UTC (GMT). To reference the difference between local time and UTC we subtract one from the other. Which one is subtracted from makes no real difference to anything. Given there are two ways to do a thing, and which way it is done makes no real difference to anything, different people, naturally, choose different ways.
From 30 years ago (or more), unix picked UTC-local as the way to get the offset. If you do it that way, in the Eastern part of the US, during winter, you get a +5 hour difference.
The International time standards (notational standards) picked the other way (whether that's from before or after unix picked its way I have no idea) - they use local-UTC. The Eastern time zone in the US is then -5 hours from UTC. Most things these days use international time notation, or something loosely based upon it (so the time zone in the Date field of this message should be +0700 if nothing alters it - I'm currently 7 hours ahead of UTC). But the time zone in the system I'm running is set at -420 (420 minutes different from UTC, with UTC behind local time). That's because it is a unix system, which does things the other way. The Olson tz database was developed for unix systems, the names it gives to zones when represented externally, are generally the unix version of the time - on a unix system to set US EST, you'd set GMT+5, which would then cause an e-mail program to put -0500 in the e-mail headers! For any of this to make sense, you simply must know the convention used. If times are being displayed in some non-OS specific context, I would certainly not be showing anyone the unix (now posix) specified format, that should be confined to where it is needed. Display timezones always in the international format (if you don't have a suitable time zone name to display, get the numeric value, and display it yourself, don't use the timezone name from the olson database). kre

The Bell Labs people who originally developed UNIX were in the US and were thinking in terms convenient to North American locations (i.e., a positive integer value representing minutes west of GMT made sense because all of the US is west of Greenwich). By the time anyone dreamed that Bell Labs' modest creation would give rise to a family of OS'es that would one day span the entire globe, the original time zone sign convention was cast in concrete, and no one dared try to change it to something that made more logical sense for fear of breaking countless existing applications that depended on that early decision made by a handful of American developers in the early 1970's. Traces of the confusion which UNIX created over the time zone sign issue can probably also be seen in RFC 822 (the 1982 Internet e-mail format standard), which allowed the conventional one-letter military time zone designators but mixed up their meanings by getting the signs wrong. Rich Wales richw@richw.org http://www.richw.org

Date: Tue, 15 Feb 2005 11:36:02 -0800 (PST) From: Rich Wales <richw@richw.org> Message-ID: <20050215191431.L09900.richw@whodunit.richw.org> | and no | one dared try to change it to something that made more logical sense Neither convention makes more logical sense than the other. They're just different. | Traces of the confusion which UNIX created over the time zone sign | issue can probably also be seen in RFC 822 (the 1982 Internet e-mail | format standard), Nonsense, the e-mail standard writers weren't unix users (at the time) - that was a simple error that was totally unrelated. kre

Dear Robert and Paul, thank you for the clarification. TWiki.org's gateway already states "The names are extensions to POSIX defined by the Olson tz database" and "Greenwich Mean Time: (GMT/UCT/UTC/Universal/Zulu; positive numbers count hours west of GMT)" Which is enough explanation; to avoid confusion I do not add a note about difference of International time standards and POSIX. Best regards, Peter Robert Elz wrote:
Date: Mon, 14 Feb 2005 09:18:10 -0500 From: "Olson, Arthur David (NIH/NCI)" <olsona@dc37a.nci.nih.gov> Message-ID: <75DDD376F2B6B546B722398AC161106C740377@nihexchange2.nih.gov>
To expand a little on Paul Eggert's response...
| I am not sure I can explain the +/- GMT question, someone from the Olson tz | database team is probably better in that.
There is no absolute right or wrong here, there are two different times, local time (wherever) and UTC (GMT). To reference the difference between local time and UTC we subtract one from the other. Which one is subtracted from makes no real difference to anything. Given there are two ways to do a thing, and which way it is done makes no real difference to anything, different people, naturally, choose different ways.
From 30 years ago (or more), unix picked UTC-local as the way to get the offset. If you do it that way, in the Eastern part of the US, during winter, you get a +5 hour difference.
The International time standards (notational standards) picked the other way (whether that's from before or after unix picked its way I have no idea) - they use local-UTC. The Eastern time zone in the US is then -5 hours from UTC.
Most things these days use international time notation, or something loosely based upon it (so the time zone in the Date field of this message should be +0700 if nothing alters it - I'm currently 7 hours ahead of UTC).
But the time zone in the system I'm running is set at -420 (420 minutes different from UTC, with UTC behind local time). That's because it is a unix system, which does things the other way.
The Olson tz database was developed for unix systems, the names it gives to zones when represented externally, are generally the unix version of the time - on a unix system to set US EST, you'd set GMT+5, which would then cause an e-mail program to put -0500 in the e-mail headers!
For any of this to make sense, you simply must know the convention used.
If times are being displayed in some non-OS specific context, I would certainly not be showing anyone the unix (now posix) specified format, that should be confined to where it is needed.
Display timezones always in the international format (if you don't have a suitable time zone name to display, get the numeric value, and display it yourself, don't use the timezone name from the olson database).
kre
-- * Peter Thoeny Peter@Thoeny.com * Is your team already TWiki enabled? http://TWiki.org * This e-mail is: (_) public (x) ask first (_) private
participants (5)
-
Olson, Arthur David (NIH/NCI)
-
Paul Eggert
-
Peter Thoeny
-
Rich Wales
-
Robert Elz