bug#7797: Configure the use of from header for epg signing.

[Luca Capello]

Hi there!

On Fri, 07 Jan 2011 14:18:41 +0100, Daiki Ueno wrote:
> Luca Capello <luca@pca.it> writes:
>> Please, can you post it here, so I do not have to download the bzr
>> repository?  The Savannah bzr repository is disabled
>> <http://bzr.savannah.gnu.org/lh/emacs> and the Launhpad one
>> <http://bazaar.launchpad.net/~vcs-imports/emacs/trunk/files> does not
>> have it yet.
>
> Looks git mirror is up to date:
>
> http://git.sv.gnu.org/cgit/emacs.git/commit/?id=3d51f602

It was not at the time of my writing and looking around on the Internet
I was not sure if the Git repository is automatically updated (and when)
or not.

>> I am sorry, but I still think this is not the correct solution, thus I
>> reopened the bug (control server bcc:ed).  I think the cases should be
>> the following, which I have (sort of) explained in my first post and it
>> is similar to Ashish's one at:
>>
>>   <http://lists.gnu.org/archive/html/emacs-devel/2010-12/msg00612.html>
>
> I think there are 2 separate issues:
>
> (1) Gnus now picks signing keys from "From:" header by default. Which is
>     an incompatible behavior change from the previous versions.
>
> (2) mml2015-epg-find-usable-key, a function used to determine whether a
>     key is usable for signing, has a bug.
>
> I fixed (2) as r102771 and you are talking about (1).  

Exactly, but I do not understand the need for (2), since as I wrote in
my previous reply (point 1) IMHO by default EasyPG should do nothing WRT
GnuPG and which keys should be used, especially given that the error
message is useless without a non-nil `epg-debug':

  Message-ID: <87r5cpow4q.fsf@gismo.pca.it>
  URL: <http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7797#16>

Nevertheless, now that things went back to normality (read below), I do
not care such a lot about that.

>> I would have wanted to reply only after having implemented my solution,
>> but since I am quite busy right now (and I will be for the whole next
>> week as well), it is better to put here my ideas *now*.
>
> Well, probably you could first try the patch for (2) and then propose
> improvement ideas for (1) I think :-)

FWIW, your patch for (2) works perfectly (adapted to the
lisp/gnus/mml2015.el local copy I have), tested on both my working
environment and the fresh setup I used to report the bug.

On Fri, 07 Jan 2011 18:29:01 +0100, Daniel Dehennin wrote:
> Here is a patch to make the use of from header conditional, this solve
> issue 1 in [1].
>
> It can be fetched from my Gnus git repository [2].
>
> I do not update the Changelog to avoid conflict, I think changelog can
> be automatically handled by hook when Lars pull in its branch ;-)
>
> This patch is tested by me.

And tested by me as well, *without* Daiki's patch applied, for both
signing...
=====
/usr/bin/gpg --no-tty --status-fd 1 --yes --command-fd 0 --armor --textmode \
--output /tmp/epg-output3395dRp --detach-sign -u 8D3867D89DDB992B
gpg: skipped "8D3867D89DDB992B": unusable secret key
gpg: signing failed: unusable secret key

[patch applied]

/usr/bin/gpg --no-tty --status-fd 1 --yes --command-fd 0 --armor --textmode \
--output /tmp/epg-output3395qbv --detach-sign
[GNUPG:] USERID_HINT D91D57A03BE9F36D Luca Capello <luca@pca.it>
=====

...and encryption to my Debian address:
=====
/usr/bin/gpg --no-tty --status-fd 1 --yes --command-fd 0 --armor --textmode \
 --output /tmp/epg-output355543E --always-trust --encrypt --sign -u 
8D3867D89DDB992B -r 06EAA066E397832F
gpg: skipped "8D3867D89DDB992B": unusable secret key
gpg: [stdin]: sign+encrypt failed: unusable secret key

[patch applied]

/usr/bin/gpg --no-tty --status-fd 1 --yes --command-fd 0 --armor --textmode \
 --output /tmp/epg-output3555FCL --always-trust --encrypt --sign -r 
06EAA066E397832F
[GNUPG:] USERID_HINT D91D57A03BE9F36D Luca Capello <luca@pca.it>
=====

> @@ -1035,7 +1042,8 @@ If no one is selected, default secret key is used.  "
>    (let ((inhibit-redisplay t)
>       (context (epg-make-context))
>       (config (epg-configuration))
> -     (sender (message-options-get 'message-sender))
> +      (sender (if mml2015-use-from-address
> +                  (message-options-get 'message-sender)))

There are extra spaces before the parenthesis here ;-)

Thx, bye,
Gismo / Luca

pgpVHD5P0rWre.pgp
Description: PGP signature