Justification (was Re: Unicode Font Maker)

[Gregg Reynolds <gar at arabink dot com>]

Thomas Milo wrote:
> Meor Ridzuan Meor Yahaya wrote:
>
>
> > Do you have ideas on arabic justification that you don't mind sharing
> > with us?
>
>
> Yes. Considering keshide or madd a justification aid does not do justice to
> the art of calligraphy or typesetting. Keshide is an esthric device, not a
> trick. It serves to change the general appearance of text, in other words,
> you use them or you don't. If used, in calligraphy and well-executed
> typesetting, keshide is bound to many contextual constraints, which, just
> like the ligature system, are not generally know among computer
> enthousiasts, with devastating consequences for the Islamic art of text
> composition.
>
> To sum it up: justification can very well be achieved by subtle variation in
> intra-word and inner-word spacing. Only as a last resort keshide should be
> used for justification.

Here's my solution:  define <TATWEEL> to mean "this place is a candidate 
for extension of the line, regardless of coloration", such that it takes 
on color based on context.  E.g. <k><tatweel><t><b> means the positive 
(inked) space between <k> and <t> (i.e. the tiestroke) may be extended. 
 <d><tatweel><w><r> means the negative space (whitespace) between <d> 
and <w> may be extended.  The relative extent of such stretching would 
be set by typesetting policy in the software.  Then there are four 
groups of extension candidates: ordinary inter-word negative space, 
explicitly specified stretch candidates (using tatweel), and ordinary 
intra-word positive (black) and negative (white) space.  It's up to the 
composition system to set policy regarding how extra space is 
distributed among these categories.  This is pretty much how the TeX 
justification algorithm works.

-gregg