[emacs-bidi] Hebrew characters categories

[Yair Friedman (Jerusalem)]

Hi,

Things seems to be more complex than I estimated before. It took me
some time to understand that composition heavily relay on character
categories.

international/characters.el specifies various categories under the
"phonetic classifications" section.  However it seems that they mix
phonetic properties of characters (consonant/vowel/tone/semivowel) with
typographical properties of characters (independent vowel/upper
vowel/lower vowel).

Apparently this scheme doesn't fit Hebrew at all.

Here is how things look:

The accents (cantillation marks).  I think that the best phonetic
description would be tone mark. We might want to add the typographical
information we would need these categories: tone-top-left,
tone-top-center, tone-top-right, tone-bottom-center, and
tone-bottom-right.

The points can split into these phonetic categories: The Sheva-Qubuts
range are clearly vowel signs. This are vowel-bottom-center or
vowel-top-left.

The Shin/Sin Dots and the Rafe are consonant-modifying diacritical mark
(currently nonexistent category). These might be considered just as
"diacritical mark". The typographical categories for these would be
accent-top-right, accent-top-left and accent-top-center accordingly.

The Point Dagesh or Mapiq or Shuruq is the most problematic one.  As
Dagesh or Mapiq its a consonant-modifying diacritical mark, while as
Shuruq it's a vowel sign... In addition it's a center mark.

The letters can be classified as consonant, but some of them are
sometimes independent vowels. 

If Emacs will continue with the mixed approach, then we would need 13
categories for Hebrew, most of them new.

On the other hand, separating phonetic from typological information
might unnecessary complicate things for other languages.

Maybe a new approach is needed?

What is the approach taken in Arabic?