Re: strange UTF8 encoding problem (relevant to decoding-system-gone-awry?)

[Kenichi Handa]

In article <address@hidden>, Nic Ferrier <address@hidden> writes:

> I've noted the current discussion on Emacs coding.
> I am experiencing a strange problem with Emacs encoding which I
> thought I might share.

> I'm reading the tcpd package's hosts_acccess man page with Emacs man
> from this version of Emacs:

>   GNU Emacs 21.3.50.22 (i686-pc-linux-gnu, GTK+ Version 2.4.10) of
>   2004-12-14


> In the man page viewed on a terminal there are nice little bullet
> characters. Hexdump shows these characters as B7 so obviously the
> terminal is not UTF-8.

> The UTF-8 sequence for B7 is 0301 0267.

> When I view the man page in Emacs with utf-8 encoding on by default I
> get a \267. Encoding the page as unix produces:  \302\267 which
> *does* look like a valid UTF-8 byte sequence.

> When I do (what-cursor-position) on the character I get 302 which is
> the first byte in the sequence.

> I'm not sure what Emacs is doing here. It looks like valid UTF-8 and
> yet (what-cursor-position) obviously does not believe there is a UTF-8
> character.

> Anybody got any idea why the correct character doesn't display?

I can't reproduce it.  What I did is:
% LANG=de_DE.UTF-8 emacs -Q
and M-x man RET man RET

It surely decodes utf-8 output of man command correctly.

What is the value of enable-multibyte-characters?
Can you reproduce the bug with -Q arg?

---
Ken'ichi HANDA
address@hidden