RE: New Public Review Issue

From: Peter Constable (petercon@microsoft.com)
Date: Wed Feb 25 2004 - 00:36:32 EST

  • Next message: Martin Duerst: "Character Model: Two new documents and Last Call"

    > From: unicode-bounce@unicode.org [mailto:unicode-bounce@unicode.org]
    On
    > Behalf Of Kenneth Whistler

    > > However, that's got to be about the second least attractive option
    > > imaginable. (Leaving the slot for truly least attractive option
    > > open here for some as-yet-undiscovered monstrosity ;-)
    >
    > BENGALI COMBINING KHANDA MODIFIER
    >
    > A combining mark, which only applies to a TA baseform, and
    > which has the effect of reshaping the TA into a khanda-ta
    > form. How's that for an alternative?

    Gackk!

    > Or, if you don't like that, we could have khanda-ta represented
    > by the sequence of Latin letters, <k,h,a,n,d,a,-,t,a> and
    > have the rendering engines and fonts remap that sequence
    > to the appropriate glyph.

    Naw, that's pretty lame, being rather similar to what is done in some
    systems representing characters as named entities. For instance, it
    wouldn't be hard to imagine &khandata; in an XML stream. The only thing
    missing is that would be a layer of representation one level removed
    from Unicode.

    What about creating a new control character? Some possible names:

    ZERO WIDTH CONJUNCTIVE NON-JOINER
    ZERO WIDTH NON-JOINING JOINER
    ZERO WIDTH HALF FORM NON-JOINER
    ZERO WIDTH SEMI-JOINER

    Hey, I think I like that last one. ;-)

    This could be used in the kinds of contexts in which ZWJ and ZWNJ have
    been used, but would provide a third alternative for situations like
    this in which there is a binary distinction but one of the two things to
    be represented doesn't exactly fit into the mold of ZWJ. Now, in most
    situations, where things *do* fit the mold of ZWJ, ZWSJ would behave
    exactly like ZWJ. But in a situation like this, it would be the
    opposite: ZWSJ would be used for the khanda ta; and as for how the
    corresponding sequence with ZWJ should be displayed, ZWJ would behave
    just like ZWSJ.
     
    Of course, we would be free to start inventing new renderings that can
    be given to things like Arabic letters preceded or followed by ZWSJ, or
    < c, ZWSJ, t >.

    (I'll bet this idea still doesn't reach the pinnacle of monstrosity.)

    Peter



    This archive was generated by hypermail 2.1.5 : Wed Feb 25 2004 - 01:32:54 EST