[Fribidi-discuss] fribidi & arabic shaping

[Tomas Frydrych]

Just to clarify what we mean by shaping -- do you mean choosing 
initial/medial/final/isolate forms of letters based on the context?

If so, I did not know that FriBidi can do this, and I am not entirely 
convinced it should do it either. The reason for this is that while 
shaping can be done for Arabic by replacing the base Unicode 
value with one of the values from Arabic Presentation Forms A/B, 
this is merely the consequence of the Arabic Unicode set being 
created as a legacy one. There are other code sets that require 
shaping (e.g. Syriac) that do not have a legacy code set, and 
whose Unicode definition is strictly semantic (e.g., only one 
codepoint for 'nun' which can be represented by 4 different 
glyphshapes). The glyphshaping in such a case can only be 
implemented by a font-rendering system, using fonts that support it 
(OpenType), and I believe that it is the view of the Unicode 
consortium that this is where glyphshaping belongs.

The bottom line is, that you cannot create a Unicode-compliant 
glyshaping engine without meddling with fonts, and am not sure 
that FriBidi wants to do this (there are other projects working on 
this, e.g. freetype.org). As far as Arabic glyphshaping is 
concerned, I think it would be better to have a separate library for 
this, and leave FriBidi as an implementation of the Unicode BIDI 
algorithm only.

Tomas