Re: error in arabic

[Eric M. <ma...@jt...>]

At 09:26 PM 5/29/2002, Yongliang Luo wrote:
>  In ArabicShaping.c the array ArabicShaping::shapeTypes hold the shape 
> types of all Arabic letter. Where 0x649 should have type _r_, but in this 
> file it is _n_.
>
>Best Regards

Here's the relevant line from ArabicShaping.cpp:

    _d_, _d_, _d_, _d_, _d_, _d_, _d_, _r_, _d_, _d_, _t_, _t_, _t_, _t_, 
_t_, _t_,   // 0x641 - 0x650


The entry for 0x0641 is "_d_", which means that it joins on both sides.

Here is the entry for that character in ArabicShaping.txt from the Unicode 
site (http://www.unicode.org/Public/UNIDATA/ArabicShaping.txt)

0649; ALEF MAKSURA; D; YEH

Here is the meaning of the entries, also from ArabicShaping.txt:

# Each line contains four fields, separated by a semicolon.
#
# The first field gives the code point, in 4-digit hexadecimal
#   form, of an Arabic or Syriac character.
# The second field gives a short schematic name for that character,
#   abbreviated from the normative Unicode character name.
# The third field defines the joining type: R right-joining,
#   D dual-joining, U non-joining
# The fourth field defines the joining group.

So, it looks like the entry for 0x0649 (ARABIC LETTER ALEF MAKSURA) is 
correct. Looking at the CVS history of this file, I see that I changed some 
of the shapeTypes in version 1.2 in ICU 1.8; before that the value was "_r_"

I don't remember the details, but I think that in traditional calligraphy, 
such as in the Koran, this letter can join on both sides.

Regards,
Eric Mader
IBM GCoC San Jose
5600 Cottle Rd.  M/S 50-2/B11
San Jose, CA 95193