Re: tmm.el: non-ASCII characters with locale-coding-system

[Stefan Monnier]

> Could you please consider applying the following patch to handle
> non-ASCII characters with locale-coding-system for tmm.el?

> -             (setq gl-str elt)
> +             (setq gl-str (decode-coding-string elt locale-coding-system))
[...]
> +  (setcar elt (decode-coding-string (car elt) locale-coding-system))

In both cases, you assume that the elt (or its `car') is a unibyte
string that needs to be decoded.  But those should have been decoded
already.  Please try and figure out why they're still unibyte strings
rather than pre-decoded multibyte strings.

> The patch is created by Tetsuo Tsukamoto (czkmt at remus.dti.ne.jp)
> at http://mail.ring.gr.jp/skk/200803/msg00009.html to prevent a
> problem of displaying Japanese characters with tmm.el reported
> by Noritada Kobayashi at Debian BTS http://bugs.debian.org/470646.

> To reproduce the problem,

> * Install Daredevil SKK 13.1 (http://openlab.jp/skk/) and Emacs
>   23.0.60 or 22.2.

> * Run Emacs with `LC_ALL=ja_JP.UTF-8 emacs -q'.

> * Type `M-x skk-mode RET', `M-`' and `s' to select the menu `s==>SKK',
>   and then garbled characters (\351\240\230...) are displayed.

My guess is that if you try the same thing with a different locale, it
will fail, because the skk-mode for some reasons places utf-8 encoded
unibyte strings in the menu, so the if any decoding is needed it should
be with utf-8 rther than with locale-coding-system.


        Stefan