Quoting "Michael (michka) Kaplan" <michka@trigeminal.com>:
> From: "Andrew Cunningham" <andjc@ozemail.com.au>
>
> Well, I guess this is one of those huge "maybe" type questions, since
> there
> is no universal definition of what "supports Unicode x.xx" means. Here
> are
> some sample posers:
>
LOL
yep i understand and agree ... I suppose that working predominately with
community languages in Australia, I tend to get asked more often for those
scripts in unicode 3.0 that Microsoft don't support yet in any way shape or
form.
*shrugs*
'tis the weave. One of the inherent problems with working with multilingual
community information. Life would be easier if I was working on teh business
side reather than teh community side of the field.
>
> > and if only they did allow latin script support in uniscribe .... but
> i
> > guess support for african langaguageds is extremely low on their list
> of
> > priorities.
>
> I would not ever presume such a thing... what issues in latin scripts
> are
> you referring to? I am not sure Uniscribe is where such a fix would be
> (all
> the issues I know of would involve keyboards and potentially fonts).
>
Lets see ... one problem i'm having at the moment .. is how to support Dinka
(Southern Sudan) in Unicode on web pages displaying on windows
95/98/ME/NT4/2000.
four characters come to mind, each of teh four characters can be represented
ideally by a pair of code points ...
U+0254 U+0308
LATIN SMALL LETTER OPEN O + COMBINING DIAERESIS
U+0186 U+0308
LATIN CAPITAL LETTER OPEN O + COMBINING DIAERESIS
U+025B U+0308
LATIN SMALL LETTER OPEN E + COMBINING DIAERESIS
U+0190 U+0308
LATIN CAPITAL LETTER OPEN E + COMBINING DIAERESIS
also there is a convention for indicating tone that is not part of teh formal
orthography of teh langauge, but is useful in materials deisgned for students
learning teh language. A set of combining diacritics are used to indicate tone.
Riusing tone indicated by an acute, and falling tone indicated by a breve.
so U+0254, U+0186, U+025B, U+0190 would have to combine with an acute and a
grave.
All breathy vowels (indicated by a diaeresis) would also have to combine with a
grave or acute .. so you'd have a base vowel: a,e, open e, i, o, open o, and u
each with two combining diacritics, one a diaeresis and teh second an acute or a
grave.
theoretically I know what unicode characters would be in teh data stream, I
could use keyman for instance to input teh appropriate characters/vowels and teh
combining diacritics. The problem comes with display.
I can cheat ... and create glyphs in teh PUA for all teh necessary charcater ..
that would mena that instead of entering U+0254 U+0308, I'd have the input
software input a single code point in teh PUA ... a rather daft approach for
future compatability since an appropriate codepoint sequence already exists
(U+0254 U+0308).
In theory this could be handled using glyph substitution .. its possible to
create an open type font that uses glyph substitution to render the required
glyphs.
buut this is where the problems start, from my understanding adobe's indesign
supports some open type font features using teh latin script, but Microsoft's
uniscribe does not support Latin script.
since my knowledge of font renderimg technology is rather limited, are you aware
of another way i can render these characters in IE5+ on various windows
platforms
I suppoose if i restricted myself to fixed width fonts I could create combining
diacritics that would be correctly spaced ... but since I really need
proportional fonts ... I'm not sure how to proceed.
currently we're using custom character sets (8-bit) taht were explicitly made
for the Dinka language.
This problem isn't unique to Dinka, you'll find it exists in other african and
some australian aboriginal languages. So teh question is ... how should one
handle kllangauges that use combinations of latin letters and diacritics and
where a precomposed form does not exist?
Andj.
Andrew Cunningham
Multilingual Technical Project Officer
Vicnet, State Library of Victoria
andrewc@vicnet.net.au
This archive was generated by hypermail 2.1.2 : Fri Jul 06 2001 - 00:17:16 EDT