Re: Malayalam Zero - an error

From: Philippe Verdy (
Date: Sat Apr 09 2005 - 08:06:57 CST

  • Next message: N. Ganesan: "Encoding Grantha (Indic) script in Unicode"

    From: "John Hudson" <>
    > Philippe VERDY wrote:
    > > But then, isn't the circle form an alternative for this digit, and a
    > > candidate for
    > > adding a variant of it, looking like LETTER TTHA (using a variant
    > > selector)?
    > The world is full of variant glyph forms and most do not have and will not
    > have variant selector sequences. The latter is a mechanism that is only
    > very occasionally necessary or useful. The vast majority of variant glyph
    > forms can be addressed at the glyph processing level, e.g. by selecting an
    > appropriate font or by enabling stylistic alternates through smart font
    > features, either of which would work for possible variant forms of the
    > Malayalam zero.

    Would that still be true if there exists some legacy text in the transition
    period, where a written number sequence like "85(1/2)" that includes the
    one-half glyph ambigously means "85.5" or "850"?

    In that case, this cannot be solved at the glyph processing level, because
    this is subject to human interpretation. In that case, the variant
    explicitly says that the one-half glyph is used, and probably means zero,
    but for accuracy the original one-half glyph is kept...

    I am not discussing the introduction of stylistic variants, but a possible
    variant needed only because the ambiguity of semantic of the glyph to
    encode. If one blindly replaces the glyph of this bogous ZERO digit, may be
    this will break those texts that actually meant the one-half glyph which
    probably means zero.

    Adding this variant could be done independantly before a more complete
    update will include a new set of fraction numbers for Malayalam (and other

    Also, I think that the new reference glyph for zero should be near from the
    existing letter t.ha (i.e. the circle boldened on its right side, and very
    thin or slighly open on the left side), possibly a bit flattened
    horizontally to make it still recognizable from the letter, and to better
    fit with the dimensions of other figures. From the various documents that
    discussed this digit, it seems that this digit is flattened this way.

    This archive was generated by hypermail 2.1.5 : Sat Apr 09 2005 - 08:07:43 CST