Re: Pango arabic engine

[Karl Koehler <koehler or uni-bonn de>]

On Wed, 8 Nov 2000, Pablo Saratxaga wrote:

> Kaixo!
> 
> On Tue, Nov 07, 2000 at 07:51:10PM -0800, Chookij Vanatham wrote:
> 
> > For the case of more than 2 arabic vowels in the consecutive sequences,
> > what would the second arabic vowel be supposed to display ?
> > 
> > Ex:
> > 
> > 	U+0639 <--- arabic consonance
> > 	U+064B <--- arabic vowel
> > 
> >   logical input:  U+0639    + U+0639    + U+064B     + U+064B      + U+0639
> > 		   (cons)     (cons)      (vowel)      (vowel)       (cons)
> > 		   
> >   visual output: InitialForm  MiddleForm  combined     ??????        FinalForm
> 
> The first vowel should display on top of the previous consonant; and the
> second (and other) should display in top of a tatweel (that horizontal
> stroje used to fill blanks).
> 
> > I know that the second vowel, U+064B, is causing the invalid/in-correct
> > spelling of arabic words
> 
> But Arabic script is not used only to write Arabic language.
But the script calls for
a) A long vowel instead of two or more vowel signes,
   that means: Alif, Ya, or  Waw.
   So two of Fatha, Damma, Kasra is illegal in every case.
b) the representation of diphtongs ( au, ai ) be represented
   as Fatha  on the _preceding_ consonant,
   with an Alif, Ya, or Waw following.

> I agree that two same vowels in a row is odd, but I have no evidence that
> it is not possible in any language; two following different vowels may
> be very well used somewhere.
There rest only the combinations i+u, u+a, u+i which could make any sense.
In persian, they don't.

> So, I don't think it should be forbidden to type it.
I'll check out urdu and kurdish.

> > Here is my visual display for this sample.
> > 
> >   logical input:  U+0639    + U+0639    + U+064B     + U+064B      + U+0639
> > 		   (cons)     (cons)      (vowel)      (vowel)       (cons)
> > 
> >   visual output: InitialForm  FinalForm  combined    SpacingForm    IsolateForm
> >                               ^^^^^^^^^              ^^^^^^^^^^^    ^^^^^^^^^^^
> 
> I don't think it is good. A diacritic should not create a break in
> a word; only a space, a ZWNJ, or a letter that doesn't combine, could
> do it.
> Remember that a word in Arabic script should look the same with or without
> the diacritics; that is U+0639 + U+0639 + U+064B + U+064B + U+0639 should
> give the same shaping for letters than U+0639 + U+0639 + U+0639 (well,
> it could be more or less larger)
> 
> > The idea is that, for those invalid sequences,
> 
> I don't think Arabic has that notion of invalid sequences.
Sure ! Several vowels for examle is illegal in arabic. One would
only type such a thing by mistake. It's not good that it 
does not show with the current shaper. 
I think I'll implement them as printed on space/tatweel, depending on
the preceding letter; this would not break the ( hypothetical ) Damma+X
combinations and show typos well enough.


Karl