Re: ISO 15924 draft fixes

From: Philippe Verdy (verdy_p@wanadoo.fr)
Date: Thu May 20 2004 - 18:26:25 CDT

  • Next message: Philippe Verdy: "Re: ISO 15924 draft fixes"

    From: "Michael Everson" <everson@evertype.com>
    > I could use a little help rendering this into French, lest I
    > embarrass myself....
    >
    > "The Property Value Alias is defined as part of the Unicode Standard
    > and is provided informatively in the tables here to show how entries
    > in the ISO 15924 code table relate to script names defined in
    > Unicode."

    Tip: French translation is:
        "Le synonyme de valeur de propriété est défini au sein du Standard Unicode
    et est fourni ici de façon informative dans les tables, afin de montrer comment
    les entrées des tables de codets ISO 15924 correspondent aux noms de scripts
    définis dans Unicode."
        (there should be a reference to the PropertyValueAliases.txt file in the
    UCD, and the section in the UTS or its annexes that describes this UCD text
    file.)

    It's true that the PropertyValueAliases.txt file in the UCD already contains
    long aliases for the shorter ISO-15924 codes:

    (...)
    sc ; Arab ; Arabic
    sc ; Armn ; Armenian
    (...)
    sc ; Zyyy ; Common
    (...)

    It's true that this same file does not list all possible values (the long value
    "Inherited" has no other alias defined in that file).
    May be this file in the UCD could list also the ISO-15924 numeric codes, but
    there's no obligation to add them there. Simply the existence of the "sc: ..."
    lines are enough to indicate that the prefered alias is the ISO-15924 code when
    it exists, so that "Arab" is prefered to "Arabic", or "Linb" is prefered to
    "Linear_B".

    With regards to semantics however, there's no difference between "Arab" and
    "Arabic", or between "Linb" and "Linear_B", meaning that these values are in the
    same value space. That's a good reason to not pollute that value space with new
    long uneeded aliases. The long aliases only exist for legacy reasons, also in
    Unicode, and the "ID" column in ISO-15924 tables is mostly informative, and
    should not be normative.

    This ID column in ISO-15924 already has the semantics of a "Unicode Script
    Property Value Alias", but it could be any other alias needed for some other
    legacy applications. I just wonder why this column was placed there, before the
    Date column that is required, given that there may possibly exist several legacy
    aliases to list in ISO-15924, and defined in other standards than Unicode.

    If you want to keep a master table for the long term, I would either drop this
    ID column, or put it at end of the row, after the Date field (so that more than
    1 alias could be added to each code; For example, there are some numeric script
    ids defined in OpenType and that could be listed as "X_OT_17", if they are bound
    directly to standard script codes)



    This archive was generated by hypermail 2.1.5 : Thu May 20 2004 - 18:27:13 CDT