GNU bug report logs - #1399
23.0.60; Some Unicode glyphs incorrectly mapped to CJK

Previous Next

Full log

Message #15 received at 1399 <at> emacsbugs.donarmstrong.com (full text, mbox):

From: Kenichi Handa <handa <at> m17n.org>
To: 1399 <at> debbugs.gnu.org
Cc: ian <at> digg.com
Subject: 23.0.60; Some Unicode glyphs incorrectly mapped to CJK
Date: Tue, 17 Mar 2009 11:00:30 +0900

Sorry for the late response.

> It seems that some Unicode glyphs are incorrectly categorized.
> 
> For example, U+201C, U+201D, U+2018, U+2019 (LEFT/RIGHT SINGLE/DOUBLE  
> QUOTATION MARK) are all mapped into the CHK category. This results in  
> the use of the STHeiti font for those characters, which are a  
> different width than the normal font I've chosen.

Category doesn't affect the font selection.

As all of those characters are `symbol' script, Emacs at
first lists fonts that have at least one of #x201C, #x2200,
#x2500 (see script-representative-chars), and select one
that matches best with your default font's family, foundry,
etc.

In your case, perhaps all your listed fonts have different
family, foundry, etc than the default font, and thus Emacs
selects arbitrary one from the listed fonts.

Currently, Emacs can't know which kind of font is more
suitable for those charaters; a font that has double-width
glyphs for them, or a font that has single-width glyphs.

So, if you prefer a specific font for symbol characters, you
must modify the defualt fontset (or whatever fontset you are
using) for symbol characters, for example, as this:

(set-fontset-font
  "fontset-default"
  'symbol
  '("FAMILYNAME" . "iso10646-1"))

> I think it's incorrect for them to be categorized as CJK, since they  
> are widely used in latin scripts.

Character category is not exclusive.  Even if a character
has CJK category, it doesn't mean that the character is not
Latin.

---
Kenichi Handa
handa <at> m17n.org

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.