Marco.Cimarosti@icl.com wrote:
>
> Of course Microsoft products use Unicode, so they can *represent* text for
> all the languages of India, at the binary level.
Also, proper tagging is requested, which means that Word 97, even if it was
able to store Unicode characters, was not able to tag the text properly.
> But I think that what they mean for _supporting_ a language is to have
> proper fonts, keyboard drivers, localized user interfaces, spellcheckers, etc.
... and a proper rendering tool, which is a must in the case of Indic
scripts ;-).
OK, so we go for a detailled look:
- the rendering engine is called Uniscribe (a.k.a. USP10.DLL), and
it is shipped with IE5 as well as Windows 2000 and Office 2000
(with differences). The engine does support all Indian scripts
(and Sinhalese will be probably supported in a next release).
- the fonts, as I said, are only delivered in Windows 2000; I am not
aware of such fonts available outside Microsoft, even if the specs
are published for some time now; but certainly this will evolve.
- keyboard drivers, localized UI, etc., is only available with
Windows 2000, this is very clear in Microsoft strategy and have been
said a number of times.
> The last time this was asked on the Unicode List (about Explorer), on March
> 2, 2000, F. Avery Bishop (Microsoft) wrote:
>
> > Microsoft Internet Explorer 5.0 is enabled for only Devanagari and
> > Tamil
Well, on first attempt I was believing this was untrue, because I am
able to see Gurmukhi characters in IE5 with a font I am developping...
But after careful verification (in fact, doing the test on a bare bone
box instead on my development machine!), it appears that Mr Bishop is
quite right, at least about Gurmukhi: the version of USP10.DLL that ships
with 98SE (and with IE 5.00), i.e. version 1.0175.0000.1, does not
understand Gurukhi. OTOH, the one that ships with Windows 2000,
version 1.0325.2180.1, _even when run on my Win95 box_, displays
perfectly Gurmukhi characters...
> But, actually, I have minor doubts also about MS's claim of supporting
> Devanagari.
It works quite well, assuming you have the correct fonts.
This is correctly pointed out in Mr. Bishop's articles on MSDN.
> I have Internet Explorer 5.0 (5.00.2314.1003) on Windows NT and MS Arial
> Unicode, but when I try Devanagari text in a HTML page, I see that the
> <virama> sign is visibly displayed, rather than causing conjunct consonant
> glyphs to be formed.
Yes, because Arial Unicode is *not* Open Type enabled, so Uniscribe does not
actually process the conjuncts. If you switch to a "better" font like
Mangal (and disconnects Arial Unicode for Latin, because it takes precedence
over Mangal which is a "hidden" font), then conjuncts appear as correctly
formed, for example with the TITUS test page for Devanagari:
<URL:http://titus.fkidg1.uni-frankfurt.de/unicode/alphabet/devatest.htm#trouble>
(if you disregard the spurious U+E900 characters that have been inserted
in the phrase: this is probably an artefact of the translation).
Antoine
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:01 EDT