MORX and GSUB shaping in HTML

Question

Question: There is talk of supporting implementations of the intelligent composition model in HTML. Opera is talking of supporting some optional GSUB shaping tags and Safari supports the obligatory CMAP shaping tag so the audience will see the stylistics the author sees. OS X 10.5 Server, and Freeway as well, seem to be moving into this direction. Does anyone have comment or criticism of what at first sight seems to be the beginnings of a technical trend?
Issue: The intelligent composition model was iniated at WWDC 1989 and introduced at Seybold in September 1989 as TrueType. The idea is to separate processing of semantic character information from processing of stylistic glyph information in order to shield the public character set from insertion of incorrect public character information which is the consequence of one-to-one mapping models.
Hitherto, only page description models have supported the optional shaping tables in the SFNT Spline Font file format architecture, but the reality is that this does not work well. The Apple Portable Digital Document format supported the glyph stream but not the character stream and the Adobe Portable Document Format does not support the character stream when PS/EPS is intermediary.
Page markup models have had no support for preserving fixed appearance, not even the fixed appearance defined by the obligatory CMAP shaping. CMAP shaping is sufficient for informal Latin typography, but it is neither sufficient for formal Latin typography nor for informal and formal Arabic and Indic typography.
In other words, if page markup models support not only the character stream but the shaping that is necessary for formal Latin and informal non-Latin typography, the problems posed by page description models would be overcome. There are also other problems that would be overcome, since ISO-IEC 10646 is not problem-free in and of itself.
ISO-IEC 10646 supports private character information on Level 1, Level 2 and Level 3. ISO-IEC 10646 also supports private permutations of public characters (known in ISO 15076 as 'pre-evaluated transforms'). These private permutations have no public synonyms for searching and they appear to the audience as generative algorithmic graphemes.
Ideally, the point of coding text is to match the meaning and the point of composing type is to match the mark. ISO-IEC 10646 permits authors to mix these modes and attempt to hardcode appearance into the plain text. In this case the plain text is unsearchable, a problem which providing support for shaping in page markup models would also solve.
If page markup models evolve to support shaping, then there is no reason to try to hardcode shaping into plain text, and the risk of private information in publishing is reduced.
Best wishes,
Henrik Holmegaard
technical writer, mag.scient.soc.
References:
http://www.mactech.com/articles/mactech/Vol.05/05.10/Typography/index.html
http://unicode.org/mail-arch/unicode-ml/y2008-m02/0043.html
http://www.mail-archive.com/[email protected]/msg12126.html

Answer 1

OK, everybody publishes in ISO-IEC 8859-1:1987 for the Anglo-American world and Western Europe as it was before the fall of the Iron Curtain? No Greek, no Cyrillic, not even Slavonic in the Latin script? So, no problem with Private Use Area charactoids and Combining Diacritic charactoids in comprehensive content cataloguing since these two categories are not possible in ISO-IEC 8859-1:1987.
Well, well ... there is a wider world.
Best,
Henrik