Re: per-character "stories" in a database

From: Andrew C. West (
Date: Tue Mar 18 2003 - 08:04:06 EST

  • Next message: Kent Karlsson: "RE: U+00D0, U+01b7 -- variants or distinct chars?"

    On Tue, 18 Mar 2003 04:13:44 -0800 (PST), Rick McGowan wrote:

    > Not really. Perfectly nice font designs for Tibetan don't need 800 odd
    > glyphs, they can work very well with a small number of pieces and combine
    > them at run-time. What you need is pretty much a set of full-height things
    > and a set of half-height things... Take a look at just about any non-PRC
    > Tibetan font.

    Well, I'll admit that I'm not a fontographer, and may not know what I'm talking
    about when it comes to designing a font. Nevertheless, ...

    I only know of two freely available Unicode Tibetan fonts that support
    consonant-vowel stacks, and they both implement these by having a set of
    full-height consonants and a set of combining half-height consonants and
    combining vowel signs. This works OK for very simple Tibetan text, but the
    various combined elements interfer with each other when a stack has more than
    one subjoined consonant, or where (for example) a stack has a subjoined
    consonant and a subjoined vowel, with the result that complex stacks are
    unreadable with these fonts, and even simple stacks can be quite ugly. It may be
    that a more sophisticated implementation could position the combining elements
    correctly for any given stack to ensure that there is no interference (as well
    as select the appropriate ligatures for subjoined YA, vowel sign U, etc.), but I
    imagine that it would be just as much work to calculate the correct positioning
    for each element in every conceivable stack for combination at run-time as it
    would be to pre-compose the individual glyph elements within the font.

    I apologise if my analysis is naive, but I know that there are members of this
    list who have experience in creating Tibetan fonts, and perhaps they could
    better explain the mechanics of dealing with complex stacks.


    This archive was generated by hypermail 2.1.5 : Tue Mar 18 2003 - 08:36:28 EST