Re: Transliteration of Arabic characters into English

From: Antoine Leca (Antoine.Leca@renault.fr)
Date: Mon May 15 2000 - 05:01:10 EDT


Edward Cherlin wrote:
>
> >To generate the CAPITAL Z WITH CEDILLA, I must use LATIN CAPITAL LETTER Z
> >(U+005A) + COMBINING CEDILLA (U+0327). That fine for file storage, but I
> >will need to generate a glyph for display purposes showing the combination
> >character. [...]
>
> The glyph does not need a Unicode code point, and should not be given one.

I can't comment that point.

> Current font formats allow for tables of glyph properties,
> including which glyphs to use for rendering sequences of combining
> characters.

But I have problems here.

Yes, current font formats *allow* that. But OTOH, no widely deployed
font engine use them by default (Well, I did not check Apple's ATSUI).
You are requested to specifically enable additional "features" on
the renderer's side in order to see a correct result (in this case,
assuming that you are not using a monospaced font, to have the cedilla
optically centered with regard of the main stems of "Z" *and* "z";
problem is worse for the top diacritics, of course).

Plus there are plenty of fonts that do *not* implement the various
font formats that are available (last widely available example:
Arial Unicode MS; which is a great font, by the way).

Unicode is now available for nearly a decade, and this mechanism is here
from the beginning (although the problem was made milder with 1.0.1 and
the alignment with 10646). But we still have the rendering problem with
the composed characters. When will come some universally accepted
standard(s) for fonts to allow them to render correctly say 97% of
the encountered uses of combining characters (I understand that
achieving 100% is too tricky, because this is so easy to find an
ad hoc example that will immediately kill any implementable proposal).

Just trying to make a forward leap...

Antoine



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:02 EDT