Re: Variation Selectors

From: Asmus Freytag (
Date: Wed Mar 29 2006 - 13:59:21 CST

  • Next message: Kent Karlsson: "RE: Variation Selectors"

    On 3/28/2006 2:50 PM, David Starner wrote:
    > On 3/27/06, Andrew West <> wrote:
    >> 1. Bad Standardized Variants
    >> For Mathematical operators and CJK ideographs variation selectors are
    >> (or will be) used to select glyph variants that may or may not have
    >> any semantic significance. My view is that if these differences are
    >> not semantically significant they should be expressed using a higher
    >> level protocol, and if they are then they should be encoded as
    >> separate characters.
    > I'm sure that for characters that people want a variation selector
    > for, someone can come up with an example where they can claim that
    > there's a semantic significance in that document. Mathematians and
    > linguists have established semantic differences for characters that
    > are almost indistinguishable at times. For differences that don't
    > matter 99.9% of the time, the variant selectors let it be ignored
    > easily.
    The key aspects of variation selectors for this use (other than
    overriding shaping as in Mongolian
    and Phags-pa) are:

    o ignoring them does not require mapping (only filtering)
    o software can filter ranges of VS characters predictively (forward
    o obviates the need to code large numbers of compatibility characters
    o obviates the need to add mappings
    o obviates the need for software to track updates in mappings

    (the last two points are what gave rise to Normalization stability which
    expressly forbids many
     types of changes for compatibility mappings, so adding compatibility
    characters is not
     necessarily even a possible option).
    > CJK ideograph variation selectors are more important. It has been
    > established to my satisfaction that no one agrees what is a semantic
    > significance in CJK ideographs, and that there are people who want to
    > distinguish ideographs that aren't distinguished in Unicode. A higher
    > level protocol is very clumsy on a letter by letter basis, and the
    > fact that Unicode doesn't support distinctions that are percieved as
    > necessary has been used to make political hay. Variation selectors
    > provide an easy way to support those people.
    This is not in fact so different from the situation in Mathematics. The
    few mathematical
    character shapes for which variation sequences were standardized
    precisely reflect the
    subset for which no agreement on their status (mere style difference vs.
    semantic difference
    in some context) could be reached. Nor did it seem likely that such
    agreement would be
    reached anytime in the future. Ergo, the use of variation sequences.


    This archive was generated by hypermail 2.1.5 : Wed Mar 29 2006 - 14:01:46 CST