Re: conditional case mappings

From: Mark Davis (
Date: Thu Apr 29 2004 - 10:53:47 EDT

  • Next message: Michael Everson: "Re: New contribution"

    comments below.

    ► शिष्यादिच्छेत्पराजयम् ◄

    ----- Original Message -----
    From: "Theo Veenker" <>
    To: "unicode" <>
    Sent: Thu, 2004 Apr 29 00:02
    Subject: conditional case mappings

    > I have two questions about the special case mappings.

    for others:

    > Some conditional case mappings map to a zero length code point string.
    > Is it valid to assume this will never happen for non-conditional mappings?

    I can't imagine a case like that.

    > What are the values for the Special_Case_Condition; is is the contexts,
    > or the contexts plus the locales?

    Most people actually evaluate this in code, and we have not expressed it clearly
    in terms of a property. It would be something like:

    code point =>
            (lowercase, (locale, condition, result), (locale, condition,
            (titlecase, (locale, condition, result), (locale, condition,
            (uppercase, (locale, condition, result), (locale, condition,

    Now, in practice, we don't really have cases where a character maps differently
    (from the default) according to two different locales or conditions, or more
    than one casing condition.

    So faced with lines like the following:

    0049; 0131; 0049; 0049; tr Not_Before_Dot; # LATIN CAPITAL LETTER I
    0049; 0131; 0049; 0049; az Not_Before_Dot; # LATIN CAPITAL LETTER I

    the value, if treated as a single property value is really collapsed to:

    U+0049 -> [uppercasing, tr OR az, Not_Before_Dot, U+0131]

    > Theo

    This archive was generated by hypermail 2.1.5 : Thu Apr 29 2004 - 11:32:03 EDT