Re: Verifying inline gpg signatures

[Jesper Harder <harder@myrealbox.com>]

Simon Josefsson <jas@extundo.com> writes:

> Jesper Harder <harder@ifa.au.dk> writes:
>
>> Yes, format=flowed is the problem -- I can reproduce that with f=f
>> messages produced by Gnus, too.  Like this one.
>
> The reason is that `gnus-display-mime' says
>
> ...
>       (let* ((handles (or ihandles
> 			  (mm-dissect-buffer nil gnus-article-loose-mime)
> 			  (and gnus-article-emulate-mime
> 			       (mm-uu-dissect))))
> ...
>
> which means that pre-MIME parsing (mm-uu-dissect) is only done when
> MIME isn't found.  Apparently, a MIME message with only a text/plain
> part makes mm-dissect-buffer return nil, so it is treated as a
> pre-MIME message.  So messages with CT parameters, such as
> format=flowed, or generally any non-trivial MIME message, is treated
> as a text part and mm-uu-dissect is never run.
>
> Perhaps mm-uu-dissect should be run on all text/plain parts?  Or all
> text/.* parts?  Opinions?

It's a step in the right direction ... but not enough.  The problem is
that text/plain parts with a format parameter haven't been QP decoded
at this stage, so uu-dissect will fail.

It happens because of this code in `article-decode-charset':

     (when (and (or (not ctl)
		     (equal (car ctl) "text/plain"))
		 (not format)) ;; article with format will decode later.
	(mm-decode-body
	 charset (and cte (intern (downcase
				   (gnus-strip-whitespace cte))))
	 (car ctl) prompt))

I didn't investigate what happens if the exception for text/plain parts
with a format parameter is removed -- but it was obviously put there
deliberately.