Re: Why are the double-part Indic vowel signs decomposable

From: Shriramana Sharma (
Date: Thu Sep 03 2009 - 19:18:38 CDT

  • Next message: Otto Stolz: "Re: Support of U+00AD SOFT HYPHEN in current software"

    On 2009-Sep-03 23:27, Ed Trager wrote:
    > What about the case of preparing a primary school reader or text book
    > where one wants to be able to show how the decomposed parts are fit
    > together to write words on paper? For example, what if I want an
    > illustration using the following string:

    Right, I thought about this, but does this actually warrant *encoding*
    the right part separately? (Mostly the left part is already needed for
    something else.) Seeing as how this gives rise to decomposition and all
    that, I would think that such a rare use would only warrant a
    recommendation to put a copy of the glyph in the PUA.

    (Hey, if you are a publisher, then get your font designer to tweak it.
    Or else, do what I often do, create a white-filled box in your
    wordprocessor and cover up what you don't want to be seen...)

    Has noone considered the IDN problem of the following two looking the same:

    KA + E + AA கொ
    KA + O கொ

    (I actually composed them that way now.)

    Only because Unicode provides decomposition, makers of rendering engine
    allow the two matra-s E and AA to be stuck onto the single consonanant.
    If Unicode did not provide decomposition, a good rendering-engine maker
    (if I were one) would allow only a single matra to be attached to a
    single consonant.

    If at all any unexpected need arises to stick two matra-s to a single
    consonant, like in:

    க ு ் -> கு்
    (short vowel U, very very old style of writing)

    the rendering engine maker can allow the use of ZWJ to stick together
    things that would normally not be stuck together. (Like what I got on
    this same list when I complained that I could not place a visarga after
    a digit. -- digit + ZWJ + visarga = २‍ः )

    Shriramana Sharma

    This archive was generated by hypermail 2.1.5 : Thu Sep 03 2009 - 19:22:19 CDT