[XeTeX] On combining diacritics again

[Jonathan Kew]

On 23 Jan 2006, at 12:23 am, Will Robertson wrote:

> On 23/01/2006, at 9:04 , Adam Twardoch wrote:
>
>> I'll just add that in Mac OS 10.4.0-10.4.2, if a font included  
>> both AAT and OpenType Layout features, the OpenType Layout  
>> features were used. As of 10.4.3, the behavior was changed (and  
>> rightly so) and AAT features are given preference.
>
> Just getting a bit off-topic here -- I can understand that for  
> backwards compatibility this is totally the way to go here, but are  
> new fonts being implemented with AAT tables at all now?

Yes, at least in a few cases.

> And if so, is it just because Mac OS X doesn't support OpenType  
> well enough?

Partly...

> Or are there still relevant areas of AAT that are superior to  
> OpenType?

Each technology has some particular strengths and weaknesses; neither  
is superior in every respect.

Personally, I think AAT's greatest weakness in comparison to OT is  
the lack of a dynamic mark attachment mechanism; this makes it  
virtually impossible to support arbitrary sequences of diacritics on  
a base character (as opposed to supporting a predetermined, limited  
set of combinations).

On the other hand, AAT allows the font designer to create new layout  
features, provide them with human-readable (even localized) names,  
and expose them through the app's user interface (e.g., the  
Typography panel), with no code changes in any layout engine or app.  
In the OT model, this would require updates to the code of all client  
applications (or a shared library they use). So AAT gives the font  
designer much greater freedom--and control--in this area, which is a  
great strength.

JK