Re: OpenType update for Unicode 5.2/6.0?

From: Vinod Kumar (
Date: Thu Oct 14 2010 - 21:04:45 CDT

  • Next message: Andrew Cunningham: "Re: OpenType update for Unicode 5.2/6.0?"

    On Fri, Oct 15, 2010 at 3:44 AM, Andrew Cunningham

    > I've been monitoring solutions for mobile devices esp smart phones in
    > south east asia, and there seems to be a growing trend to implement
    > legacy encodings for pseudo-Unicode solutions that aren't dependent on
    > complex rendering that's missing on mobile devices.
    > Unicode compliance does not mean that complex text support must be on Open
    Font format or its Feature tag based shaping technique. There is no doubt
    that the mobile phone community has accepted Unicode as the standard for
    text representation, and for what the text should look like when
    displayed. How the text should transform to the shapes is not under
    the purview of Unicode. If people find simpler or elegant ways of
    transforming Unicode text to shapes, these are not pseudo-Unicode solutions.

    For example, India had a font standard called INSFOC for Devanagari. A
    shaping engine that will convert Devanagari text in Unicode to INSFOC glyph
    code sequence would be completely Unicode compliant with respect to
    rendering of Devanagari. Some of my contacts have already extended this
    approach for Gujarati and plan to bring all the nine Indian scripts too
    under the same Unicode text to font glyph code standard.

    Open Font (OpenType) is mistakenly considered to be the unique and only
    Unicode compliant solution for complex scripts. The OT Feature tag based
    complex text shaping, introduced by Microsoft Typography is not the only
    method for using Open Fonts. This method is based upon the idea of Digital
    fonts- fonts that can change size, slant or boldness computationally
    depending on Tags embedded into the text manually or through a page
    layout style. The transformations possible with Digital fonts have been
    extended to complex shape formation. Thus Devanagari text <Ka Halant Pa> is
    tagged essentially as ( <half> Ka Halant </half> Pa ) by a Devanagari text
    analysis program (not manually). Lower layers of the shaping pipeline
    convert this to <halfKaGlyph PaGlyph>. (With INSFOC standard, the Devanagari
    text would be converted to (<halfKaINSFOCGlyphCode
    halfPaINSFOCGlyphCode verticalRightStemINSFOCGlyphCode> ). The INSFOC
    pipeline and font is much more simple than the OpenType Feature based
    shaping. It is not a surprise that mobile platforms would go for it.

    Moreover Feature based technique is not the only way Open Fonts can be
    employed for complex text shaping. In our IndiX-II project, we have treated
    Open Fonts as concrete Intelligent fonts (and not as Digital fonts) espoused
    in the recommendation ISO/IEC 15285. ISO/IEC 15285 deals with transforming
    sequence of characters to shapes. We have demonstrated that an ordered
    sequence of context sensitive glyph substitutions as implemented by the GSUB
    tables in Open Fonts are necessary and sufficient for shaping all the
    nine Indian scripts. In IndiX-II, the layer handling the Open Font will not
    tag the text with any tags like <half>. The substitution table in the
    font will transform (<KaGlyph HalantGlyph> <PaGlyph>) to (<HalfKaGlyph>
    <PaGlyph>). <Ka Halant> followed by <PA> should shape as <HalfKa><Pa>
    without question or tags. Some developers (mobile platforms) have seen
    that the text and font handling layers as well as the font itself have
    become much more simpler and robust. Shaping complex text treating Open
    Fonts as an Intelligent font cannot be considered as pseudo-Unicode. Both
    methods of using Open Fonts are Unicode compliant for rendering the text,
    but one is better.

    and in some cases there is retrograde movement
    > Finding alternate and better implementations under the framework of a
    standard is one of the stated advantages of any standard, not only Unicode.
    There is scope for diversity and innovation in the implementations.
    Standards (especially Unicode) strive to separate the implementation issues
    from the core representational and interface issues. We should not mistake
    particular implementation choices as a stipulation of the standard.
    Specifically, Feature tag based Open Fonts, or even Open Fonts themselves,
    cannot be propagated as the essence of the Unicode standard for complex
    text shaping.

    > Andrew
    > --
    > Andrew Cunningham
    > Senior Project Manager, Research and Development
    > Vicnet
    > State Library of Victoria
    > Australia
    > <>

    Vinod Kumar
    Project IndiX

    पृथिवी सस्यशालिनी
    the earth be green

    This archive was generated by hypermail 2.1.5 : Thu Oct 14 2010 - 21:08:31 CDT