RE: Ligatured characters

From: Marco.Cimarosti@icl.com
Date: Fri Sep 15 2000 - 09:46:42 EDT


Roozbeh Pournader wrote:
> This sequence, ZWJ ZWNJ ZWJ, really worries me. In the Arabic
> script, my
> interest, this is always the case. The ZWNJ is not enough in any case,
> since it disconnects the letters.
>
> And this also means some change in many simple rendering
> programs that use
> other algorithms for cursive joining whose output was the
> same before the
> introduction of these. If only ZWL and ZWNL survived...

I don't see you point; it is rather the other way round. I understood that
ZWL and ZWNL were unified with ZWJ and ZWNJ, rather than encoded separately,
for the very purpose of being as transparent as possible for old
applications...

I agree that ZWJ ZWNJ ZWJ is indeed disgusting when you see it, but it
works.

An Arabic rendering engine that implements ZWJ and ZWNJ as defined in
Unicode 2.0 (or even 1.0) should have absolutely no problems in rendering
ZWJ ZWNJ ZWJ correctly: the two ZWJ's cause the proper connected forms to be
selected, while the central ZWNJ causes nothing, as it has no neighboring
letters to influence. So the whole sequence works as a single ZWJ, by the
point of view of an old implementation.

On the other hand, if ZWL and ZWNL were introduced as separate code points,
they would have caused old applications to break. An unaware Arabic engine
would have handled a them as any other non-Arabic character, with the result
of rendering the unconnected forms for the two boundary letters, separated
by a black box for the unrecognized character.

_ Marco



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:13 EDT