[Groff] Re: experimental multibyte support

[Fumitoshi UKAI]

Hi,

Previous patch 1.18-6.u1 has a bug regarding `uXXXX'.
It failed to handle some groff glyph name, such as `ua' in devhtml.
This is fixed in version 1.18-6.u2. This version also works
groff -Thtml for Japanese.

At Sun, 22 Sep 2002 10:02:49 +0200 (CEST),
Werner LEMBERG wrote:

> > I'm developing multibyte support based on Japanese patch now.
> 
> Very good!  Hopefully, you can assist during the transition to
> Unicode.

Yes, I'll try.
 
> >  - \[u<code>] support, where code is unicode.
> >    So, you can write \[u3042] for `HIRAGANA LETTER A'
> 
> Uh, oh, please be careful.  \[...] are glyph entities, not input
> character codes.  Please read the article
> 
>   Re: [Groff] Some thoughts on glyphs
> 
> from Aug. 26th in the mailing list archive how I will support Unicode
> on the glyph side.

Oh, I missed to read the article.
Ah, my patch just support glyph name \[uXXXX] for glyph for Unicode
character U+XXXX which is not a composite character.
To avoid conflict with groff glyph names, it uses negative index for
glyph for unicode characters, where positive index for groff glyph names.
I'm not sure this is good solution, maybe there are more good one.
Anyway, it seems to work for me (at least Japanese).
 
> >  - some changes in font/* files
> >    + DESC has fontset directive
> >    + font file can have code range after charset
> >    + font/devps/<font> can have encodingname.
> >     (for example,see font/devps/M.proto, it make it possible to
> >      produce right Japanese postscript even if ja_JP.UTF-8 locale)
> 
> What about kinsoku shori and shibuaki?  This shouldn't be hardcoded.
> We need some requests for it -- the font files are the wrong place
> since both features are font independent.

Yes, I agree.
For kinsoku syori, I use cflags request with flag value 256 (pre kinsoku 
- don't break before) and 512 (post kinsoku - don't break after).
see tmac/euc-jp.tmac.

I didn't finish shibuaki not to be hardcoded. It's still hardcoded
in src/roff/troff/env.cc environment::add_char() as jgroff patch did.
It should be rewritten.

Regards,
Fumitoshi UKAI