Re: [Fribidi-discuss] fribidi & arabic shaping

[Chahine M. H. <mc...@ch...>]

Nadim Shaikli a *crit :

> >
> > The FriBidi library should be restricted to BiDi text ordering.  Another
> > library may be defined for text ordering and shaping in Arabic (and other
> > languages, which need this), and it will define an environment data
> > structure, which will include FriBidiEnv as a member.
>
> I disagree wholeheartedly - the inclusion of Arabic is a logical next step.

Yes it is, but not in Fribidi. Bidi ordering and Glyph shaping can be viewed as
two different layers of treatment. Besides, adding shaping to fribidi is
reinventing the wheel as Pango which relies (relied?) on Fribidi already does it
(and yes, Pango is toolkit independant, even though the fact gnome/gtk relies on
it could mistake people into thinking otherwise). Pango should be considered the
standard (for non QT based apps) unless you have good reasons to think it's
unsuitable and the work should be redone.

> The other argument that could be put forth is that fribidi's inclusion is
> difficult enough in most applications (try convincing a application author
> of its inclusion and he/she will go into a rant about it only catering to a
> very small minority of users and not worth the headache).  To have those
> authors also consider a second library would create more problems than it
> would solve when it comes to getting it adopted.

Inclusion of pango in such cases would include in the same step fribidi. In any
case, inclusion of two libraries amounts pretty much to inclusion of one, so if
any author is open to inclusion of an additional library it wouldn't be a
problem whether you are in fact talking about fribidi, pango or both. The
linking process only requires indicating which libraries you're using, and
there's no mention of libraries at the code level.

Salaam,
Chahine