bug#19393: 25.0.50; Emacs cannot determine coding system of ISO-8859 encoded files

[Eli Zaretskii]

> From: Tassilo Horn <tsdh@gnu.org>
> Date: Tue, 16 Dec 2014 16:21:10 +0100
> 
>   ftp://ftp.fu-berlin.de/pub/misc/movies/database/movies.list.gz
> 
> which contains all movies known to the international movie database
> (IMDb.com).  When I open that file using "emacs -Q movies.list.gz" (or
> unzip it first) and then do M-x describe-coding-system I can see that it
> is "t -- raw-text-unix".  As a result of this, the last movie in that
> file is displayed as "\374\347 (2012) 2012".
> 
> However, according to the `file' command, the file is plain ISO-8859.

Looks like some kind of bug, although with such a large file, it's not
easy to be sure.

> I also can't force Emacs to use ISO-8859 for that or the original file.
> `C-x RET f iso-8859-15 RET' results in a query that certain characters
> cannot be encoded using latin-9, e.g., \374 and \347, and I'm expected
> to choose another encoding.

That's not how you force Emacs to use a specific encoding when
visiting a file.  You should do this instead:

  C-x RET c iso-8859-15 RET C-x C-f movies.list RET

IOW, revisit the file, forcing Emacs to decode it as ISO-8859-15.
(The same works with the original compressed file.)