Re: New property for reordrant dependent vowels reordering?

From: Vinod Kumar (
Date: Tue Sep 20 2005 - 05:19:24 CDT

  • Next message: Vinod Kumar: "Re: New property for reordrant dependent vowels reordering?"

    > From: Eric Muller
    > Date: Tue Sep 06 2005 - 16:30:13 CDT
    > 1b) Should the reph of an initial ra be re-ordered across a virama +
    > ZWNJ boundary?

    First I examine how the Reph is reordered using the same principles
    as for I_vs reordering.

    The argument that <TTA-Glyph Virama-Glyph I_vs-Glyph MA-Glyph>
    will be pronounced properly, but not
    <I_vs-Glyph TTA-Glyph Virama-Glyph MA-Glyph>
    is not as convincing as the one for the Reph-Glyph. Consider text
    <930 94d 91f 94d 92e> (=<Ra Halant TTA Halant MA>). TUS4
    specifies Reph rendering in rules R2-R4 (chap 9.1).
    These rules specify what happens when the logical syllable is of
    the form <Ra Halant <consonant cluster>>.

    R3 handles the case when the <consonant cluster> takes form 1 (single
    conjunct glyph). The Reph mark applies to the single conjunct glyph.

    R4, handles the case when the <consonant cluster> is displayed in
    form 2 (<half-conj-glyph full-conj-glyph>).
    The reph mark applies to the full-conj-glyph.
    The original TUS4 R4 specifies half-consonant-form and
    I have replaced it with the more correct half-conjunct-form.
    (See how <Ra Halant Ka halant Ssa halant Pa> will be displayed.

    R2, handles the case when the consonant cluster is displayed
    in the third (explicit halant) form
    <full-conj1-glyph halantGlyph full-conj2-glyph>.
    The reph attaches to the first full-conj1-glyph.

    These three rules can be replaced by rules in terms of orthogonal
    glyph clusters:

    Reph sign rule: When a syllable is of the form
      <Ra Halant <consonant cluster>> then the <Ra Halant> is visually
      represented by a Reph sign.

    Reph reordering: The Reph sign is always written to the right of
      the first orthogonal glyph cluster formed from the <consonant cluster>.

    These rules ensure that when I enter the name of my pet doggy "Rtt-ma"
    as <930 94d 91f 94d 92e>, it is displayed as
    <TTA_Glyph HalantGlyph RephGlyph> <Ma-Glyph> pronounced as "RTT-MA"
    and not as
    <TTA_Glyph HalantGlyph> <Ma-Glyph RephGlyph>
    which will be pronounced as "TT-R-MA".

    The current TUS4 rules (R2-R4) ensure the correct rightward repositioning
    of the Reph sign by explicitly listing the three forms of consonant
    cluster. Similar rules for the leftward repositioning I_vs are
    required in TUS4.

    It is not necessary for Unicode to specify whether I_vs or Reph-Sign
    should cross ZWNJ. Unicode has specified the action of ZWJ/ZWNJ
    and how the signs should be reordered. Now it is for the user
    of the standard to put two and two together and work out the rendering.

    The function of ZWJ and ZWNJ is to provide a mechanism to the user
    to override the default forms. Depending on the selection of ZWJ/ZWNJ
    and the richness of the font, the consonant cluster would be
    displayed in one of the three forms. The I_vs-Glyph (the Reph-Glyph)
    should move left (right) of the previous (next) orthogonal glyph cluster
    formed. The reordering specification is dependent only on the
    orthogonal clusters. ZWJ/ZWNJ and the richness of the font decide
    what orthogonal clusters are formed.

    Using ZWNJ to prevent a I_vs from crossing consonant with explicit halant
    <TTA_Glyph HalantGlyph> is needed only to rectify the incorrect behaviour.
    First one makes a mistake, then uses ZWNJ to correct it
    it and then complains that Unicode has not specified how ZWNJ interacts
    with I_vs.

    Vinod Kumar
    Project IndiX

    This archive was generated by hypermail 2.1.5 : Tue Sep 20 2005 - 05:20:32 CDT