Re: Standardize TimeZone ID

From: Antoine Leca (
Date: Tue Apr 27 2004 - 06:03:55 EDT

  • Next message: Ernest Cline: "Defined Private Use was: SSP default ignorable characters"

    On Monday, April 26, 2004 9:27 PM, Frank Yung-Fong Tang va escriure:

    > The problem is the "Olson ID" itself lack of clear specfication in
    > term of RFC kind of format.

    It is perhaps better to use the Windows identifiers, then.

    We have a discussion if this kind about 6 years ago (during the building of
    RFC2445, iCalendar, which specifies formal references to timezones). The
    idea was to have a unique way to reference formally a timezone. Including,
    there were formal contacts between the chairing of the WG and A.D. Olson
    (but I do not know the decisions first hand; there were no discussion at
    this respect on the TZ list then), about creating a RFC to "standardize" the
    rules for naming, or even some kind of agency or registrar (that would be to
    officialize the tz `band'). Writing such RFC should have been pretty easy,
    rules are already pretty clear. I do not know about the various egos there,
    perhaps this is the problem. Fact is, nobody even wrote a draft.
    I should add, for the record and because it might become important, that in
    my record, the Microsoft's guys then did not loudly opposed to the use of
    "Olson ids" to names the timezones, even they usually find the Olson
    database to be a very nice reference. In fact, the RFC did explicitely
    mention the elsie database, a thing that have been added on second thought,
    while the editor of the RFC were someone from Microsoft.

    The end result (for iCalendar) is that we now have two main formats, "Olson
    ids" and the Windows strings (thanks to the omnious presence of the Windows
    tools). Of course, the second outweights the first by ample margin.

    So, since it appears there is no possibility to standardize on "Olson ids",
    perhaps we could switch using _the_ other scheme? Since it is a mere list,
    it should be pretty OK to specify, isn't it?


    This archive was generated by hypermail 2.1.5 : Tue Apr 27 2004 - 07:39:50 EDT