bug#2317: emacs silently ignores buffer file encoding?

[Sven Joachim <svenjoac@gmx.de>]

On 2009-03-15 17:12 +0100, Chong Yidong wrote:

>> (1) Start emacs -Q
>> (2) Visit "/tmp/fff"
>> (3) Insert the following chars: "θϑΘ"
>> (4) Set the buffer file encoding to latin1 with: C-x C-m f latin-1 RET
>> (5) Save the buffer: C-x C-s
>>
>> For me, "/tmp/fff" is saved sucessfully, with no error or prompt, using
>> an encoding of utf-8 -- despite my having explicitly requested latin-1.
>
> I think this bug may have been fixed.  ISTR seeing the same result when
> I tested with your recipe a few weeks ago.  With latest CVS, however, I
> get a "Select coding system" prompt.  Can you confirm?

This has been fixed on March 3 by Handa-san, see [1].  However, this fix
has side effects that I find very annoying: if I visit a file containing
only ASCII characters and insert non-ASCII characters into the buffer,
the "Select coding system" prompt pops up as well, although I never
asked about a specific coding system and would be completely satisfied
with Emacs silently choosing UTF-8[2] in that case.


1. http://lists.gnu.org/archive/html/emacs-devel/2009-03/msg00011.html
2. Or whatever the preferred coding system for new files happens to be.