Grapheme cluster boundaries and left-side spacing dependent vowels

From: James Clark (
Date: Sun Apr 20 2003 - 21:39:33 EDT

  • Next message: souravm: "Query on Unicode equivalent of iso-pub"

    I am confused about whether in Indic scripts there should a grapheme
    cluster boundary between a consonant and a following left-side spacing
    dependent vowel (e.g. 093F, Devenagari vowel sign i).

    As I understand it, grapheme cluster boundaries are supposed to be
    useful for things like cursor movement: a reasonable default action for
    the right arrow key would be to move over one grapheme cluster. For
    this reason, I wouldn't expect there to be a boundary before a left-side
    dependent vowel: the cursor would have to bounce backwards and forwards
    if there was such a boundary.

    According to UAX #29, there will be a boundary only if the left-side
    dependent vowels have the Grapheme_Extend property. Now, in Unicode 3.2,
    left-side dependent vowels did have the Grapheme_Extend property,
    because Grapheme_Extend was computed as (Me + Mn + Mc +
    Other_Grapheme_Extend - Grapheme_Link - CGJ). However, in Unicode 4.0,
    they do not because Grapheme_Extend is computed as (Me + Mn +


    This archive was generated by hypermail 2.1.5 : Sun Apr 20 2003 - 22:32:31 EDT