Re: [tlug] GTK2: Displaying Japanese font names in Romaji

[Matt Gushee <matt@example.com>]

Botond Botyanszki wrote:

> > maybe a year ago the J-font names would be shown in romaji. At some 
> > point, I guess when I upgraded GTK, all the GTK2 apps started trying to 
> > show Japanese font names in kanji, with the undesirable results described.
>
> Looks like this is a new feature in fontconfig:
> http://www.fontconfig.org/wiki/2_2e3_20release_20notes

Okay, but it says:

> Multi-lingual family/style/fullname values with parallel language 
tags > to enable locale-sensitive presentation of these names.

If I read that right, it means the font names *should* be shown in 
Romaji in an English environment, and in Kanji in a Japanese 
environment, which is exactly what I want.

Stephen J. Turnbull wrote:

> ------------------------------------------------------------------------
>
> ゴシック:familylang=xx,ja:style=Regular,

That's a bit different from mine. My fonts.cache-1 contains entries like

  familylang=en,ja,en:style=Regular:stylelang=en:slant=0:weight=80: ...

I don't know exactly what 'en,ja' means, but it looks about right, since 
I use those two languages, but primarily English.

> 標準:stylelang=xx,ja

Is that a literal value of 'xx', or your notation for 'some other 
language'? Either way, I don't have it. My font cache has only

  stylelang=en

> Since the romaji comes first, I would guess that your fonts.conf
> (singular?  bu-wha-ha-ha-ha! and of course the chance that the one
> that did this to you is documented is nil) contains a selector for
> "familylang=ja" somewhere.  Nuke that baby!

Unfortunately, that's not it. "familylang=.." doesn't appear anywhere in 
my config files. And I ran fc-cache with strace to make sure it wasn't 
accessing any other secret config files; it wasn't.

Here's something interesting, though. Many of the entries in 
fonts.cache-1 begin like this:

  "DFHsm3.ttc" 1 "ÇcÇeÇoïΩê¨ñæí©ëÃW3,ＤＦＰ平成明朝体W3,DFPHSMincho\\-W3:

In case that gets mangled in the mail, let me describe it: after the 
file name and the number (face index, I suppose?) there appear three 
versions of the font name separated by commas. The second is the name in 
Japanese characters, the third is in Romaji, and is the "official" 
PostScript name, I think.

The first is the weird one. It is clearly some form of mojibake, but the 
range of characters are different from anything I can recall seeing 
before. Characteristically, there are capital Cs with cedillas 
interspersed with lowercase Roman letters--mostly e's, c's, and o's--and 
here and there an Omega. The reason this odd string is of interest is 
that this is what is appearing in my font menus.

I suppose fontconfig is extracting the funky names from the font files, 
so maybe it's a bug in the fonts themselves. The Japanese fonts I mainly 
use are from a Dynalab collection I bought shortly before leaving Japan 
in '99; I guess the state of the art in I18n was a bit sketchy in those 
days.

Anyway, I guess the next step is to figure out exactly where the weird 
strings come from, and see if there's a way to tell fontconfig to ignore 
them.

Does anyone know of a good font analysis tool for Freetype 2? The 
ft-tools from Freetype 1 were great for this sort of thing, but for some 
reason they never got ported.

--
Matt Gushee
: Bantam - lightweight file manager : matt.gushee.net/software/bantam/ :
: RASCL's A Simple Configuration Language :     matt.gushee.net/rascl/ :