Re: Using PUA for font mapping (was: Re: Transliteration of

Date: Tue May 16 2000 - 11:01:51 EDT

Doug Ewell wrote:
>Right, but we are also talking about using a PUA code point to refer to
that glyph in the font.

>The user would specify, or the file or other data stream would contain,
U+005A (or U+007A) plus U+0327. Since all or most of the font experts here
seem to agree that having the rendering engine create this glyph "on the
fly" would yield unsatisfactory results, there needs to be a precomposed
glyph already in the font. The glyph needs an internal code point, and the
PUA is being proposed as the place for it.

No, these assumptions are wrong. First of all, a precomposed glyph can be
used, but satisfactory results can also be obtained by using zero-width
diacritic glyphs. The main issue is one of getting appropriate positioning,
but technologies are appearing (in the case of AAT, have been around for
several years) that do this: OpenType, AAT, Graphite (see my previous

Secondly, even if a precomposed glyph is used, a PUA code point is *not*
needed to get that glyph. A PUA code point is only needed as an interim
solution in order to publish a document in the absense of the "smart font"
technologies referred to above. Neither OpenType, AAT or Graphite require
presentation-form glyphs to have character codes.

Peter Constable

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:02 EDT