Re: Fribidi and joining

[Behdad Esfahbod <behdad at bamdad dot org>]

On Fri, 23 Aug 2002, Nadim Shaikli wrote:
> On Sat, 24 Aug 2002 02:32:18 +0430 (IRST),
>   "Behdad Esfahbod" <behdad bamdad org> wrote:
> > 
> > To be honest, the problem is that fribidi's pride is that it's
> > "light, simple & still conformant".  So nor me, niether anyone
> > else is going to sacrifice these to get arabic shaping in.
> > 
> > So until arabic joining/shaping (whatever you call prefer) gets
> > standard, fribidi's CVS will not let any arabic specific code in.
> 
> Uh oh :-)  In brief, if all the new code was added as optional code
> (ie. ifdef'ed out) with the user opting for its inclusion or not
> should be something that would fall right-in as far as fribidi
> being "light, simple & conformant" (esp. for those not choosing to
> include arabic shaping).  The problem, as was stated before, is if
> you think the inclusion of an Arabic-centric library is something
> that is easy to do you might be kidding yourself.  The likelihood
> of including bidi support (due to its benefits to not only Arabic,
> but Persian, hebrew, etc) is much higher than segmenting all those
> things into separate bits and pieces.  In other words, an application
> is much more likely to go through the headache (if any) to support
> 4-5 languages and not just one.  As noted, I've said all this before
> so I don't want to sound like a broken record - squeek squeek.

See my comments below.

> > The solution, as I can see is to have a real simple,
> > piece-of-code/library implementing the arabic part, and ask
> > people to use it (I've got the link to import it in AbiWord, for
> > example).  The code itself, Roozbeh and I have one implementation
> > which I sent to Moe a few days ago.  It needs some cleanup and
> > interface design before going into Arabeyes/FarsiWeb CVSs.
> 
> I might be missing something here -- you are suggesting having a
> piece of code that does shaping/joining only, right ?  Post-bidi
> or Pre-Bidi (warning: this is a loaded question :-) ?  And that
> would work without the noted problem ?  Or are you simply saying
> you don't want to see Arabic-centric issues involved in fribidi ?

No, I mean that: "A conformant implementation is far from 
possible right now, so lets have a something-that-works to use, 
but don't expect me to include it in fribidi :(.

> In any regard, I still highly suggest including any fragment within
> fribidi (given that fribidi's main functionality is unaffected) with
> it being sectioned off (ifdef 0 - if you will, simply to show intent
> and for this topic not to rear its ugly head again).

Ok, this is I wanted myself, but the fact is that when hacking 
pango a few nights ago, I found that having joining and bidi in 
one package needs a really complex interface to be usable by 
everyone.  To be exact:  I found that my dreams about the 
interface, does not satisfy the needs of pango.

> Behdad, what are your thoughts on how to go about getting this
> "arabic shaping get [more] standard" ?
> 
> In passing, I don't see how the fribidi API is really that affected.
> In other words, can't you go ahead with your current plans for the
> API (sans Arabic) and then add-in Arabic (and possible a new call or
> two to the API ?).  That would most certainly be backwards compatible
> and ought to not affect anyone using whatever API was already in use.

In designing the new API for fribidi 1.0, I consider any aspect 
that helps integrating joining code.

> PS: although we keep using and saying Arabic, all these issues affect
>     Persian, correct ?  Sorry for the bias :-)

Yeah :), so please don't get mad of me that: "He does not do that 
because its Arabic and he does not need it himself blah blah..."


> On Sat, 24 Aug 2002 02:36:39 +0430 (IRST),
>   "Roozbeh Pournader" <roozbeh sharif edu> wrote:
> > On Sat, 24 Aug 2002, Behdad Esfahbod wrote:
> > 
> > > To be honest, the problem is that fribidi's pride is that it's
> > > "light, simple & still conformant".  So nor me, niether anyone
> > > else is going to sacrifice these to get arabic shaping in.
> > 
> > Well, I was shocked when I saw this. Asked Behdad and he told me that here
> > the emphasis is on *conformant*, and not light. So we'll hopefully get
> > some shaping in FriBidi after we reach a clear consensus with Unicode bidi
> > experts.
> 
> heheheheh - this was pretty funny; but I do like what I hear about its
> ultimate inclusion :-)  You can thus disregard the various questions
> above except for the "how to go about getting a more standard shaping
> solution".
> 
> Salam !! (the way its meant in Persian :-)
> 
>  - Nadim

Find my current dreams here:

http://sourceforge.net/mailarchive/forum.php?thread_id=586443&forum_id=3189

(They have changes a little since the message is too old)

Salam
-- 
Behdad Esfahbod		2 Shahrivar 1381, 2002 Aug 24 
http://behdad.org/	[Finger for Geek Code]

#define is_persian_leap(y) ((((y)-474)%2820+2820)%2820*31%128<31)