[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Unicode and math symbols
> Does this dependence of OSs on the Unicode standard mean that whenever
> Unicode changes, everyone needs a new version of the OS or just new
> font-files, or what?
> The support for all this right now is still rather poor, actually.
> (1) For example, the Microsoft NT font installer has a fixed table
> of glyph name to UNICODE mappings (complete with typos like
> aroowleft, hungerumlaunt, traglf).
Microsoft wrongly uses Unicode as a kind of glyph repertoire,
and can get away with it as long as they don't deal with
Seems like they were forced into this by (1) inertia, (2) ignorance
maybe, and the fact that (3) there is nothing better - all a few years
ago when they started with this stuff. Today we are wiser, but still
not wise enough apparently...
The correct solution is to not have the system do the conversion
between character codepoints and glyph indices, but to leave that
to the font (which should know best). Quickdraw GX (from Apple) and
TrueType open are approaches to this, although I would prefer
it if the knowledge of the font about the mapping were available
as methods (in the OO sense) and not just as tables.
I don't agree with this. GX hides under hood much of what a
typesetting application needs to know about (such as ligatures).
The main idea behind GX is to add some apparent typesetting capabilities
to dumb appications. Useless for TeX.
> (3) The coverage of this table is limited, mostly WGL4 (which covers
> about 662 glyphs, including all the Latin alphabets, Greek, Cyrillic)
> (And in the current implementation it actually covers less).
Well, for the bulk of Unicode, CJK, names are a silly idea anyway.
Fine so they should accept unicode numbers written in hex as glyph
names. This is what the officially released version of Lucida Sans
Unicode in Type 1 format does. And which is why it does not work
ironically either in autoconverted TrueType form or in Windows NT ATM!