bug#35507: Gnus mojibakifies UTF-8 text/x-patch attachments from Thunderbird

[Andy Moreton <andrewjmoreton@gmail.com>]

On Wed 01 May 2019, Noam Postavsky wrote:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>>> From: Andy Moreton <andrewjmoreton@gmail.com>
>>> Date: Wed, 01 May 2019 17:42:18 +0100
>>> 
>>> +		     (mm-decode-string text 'utf-8))))
>>
>> As I said, I'm not sure we should do this, let alone unconditionally
>> force UTF-8 here, but if we must, why not use decode-coding-string?
>> Do we really need the mm-* stuff?
>
> As far as I can tell, the mm-* version is useful for handling stuff lke
> "UTF-8" as the charset argument (which might be useful if we extract it
> from the "Content-Type: text/plain; charset=UTF-8" header).  If passing
> 'utf-8, then it's just the same as calling decode-coding-string.

OK, in that case we could indeed just call decode-coding-string.

> For a default if we don't find a charset header, I guess `undecided'
> would make more sense, right?  After all, Emacs already has the coding
> detection machinery, may as well use it.

Please re-read the original bug report: the problem is with malformed
messages that do not contain a charset field in the Content-Type header.

The one-liner patch changes the default for inline display in the
Gnus article buffer to assume UTF-8 when nothing is specified, rather
than just inserting the text without decoding it.

That should result in text that actually is UTF-8 being displayed
correctly, and no change to plain ASCII. For anything else, the user can
use the `gnus-mime-view-part-as-charset' command to override the
default.

    AndyM