Re: pgg-gpg: BEGIN_SIGNING

[Simon Josefsson <jas@extundo.com>]

Daiki Ueno <ueno@unixuser.org> writes:

>>>>>> In <87slp587cw.fsf@latte.josefsson.org> 
>>>>>>	Simon Josefsson <jas@extundo.com> wrote:
>> Daiki Ueno <ueno@unixuser.org> writes:
>
>> >>>>>> In <87irq1evdk.fsf@catnip.gol.com> 
>> >>>>>>	Miles Bader <miles@gnu.org> wrote:
>> >> Simon Josefsson <jas@extundo.com> writes:
>> >> > What do you think?  It might be ugly to depend on the 'gpg: writing'
>> >> > output instead of the [GNUPG:] stuff, but I see no option here.
>> >
>> >> Of course if such a kludge is used, an important thing is to report the
>> >> problem with the "official" interface as a bug...
>> >
>> > Yes.  I think the root of the problem is that gpg does not emit any
>> > trigger of signature creation.  I'll propose a patch to GnuPG so as to
>> > introduce BEGIN_SIGNING and END_SIGNING status.
>
>> That sounds like a good idea.
>
> BEGIN_SIGNING is now available in GnuPG 1.4.3.  Should we install the
> attached patch right now or wait for the new GnuPG packages available in
> most distributions?

I don't think we can assume that people use new GnuPG releases.

> -(defcustom pgg-gpg-use-agent nil
> -  "Whether to use gnupg agent for key caching."
> -  :group 'pgg-gpg
> -  :type 'boolean)
> -

So I think this variable has to stay.

> -    (if (and sign (not pgg-gpg-use-agent))
> -	(pgg-gpg-wait-for-status process '("GOOD_PASSPHRASE")))
> +    (if sign
> +	(pgg-gpg-wait-for-status process '("BEGIN_SIGNING")))
>      (process-send-region process start end)
>      (pgg-gpg-wait-for-completion process)
>      (save-excursion
> @@ -302,8 +296,7 @@
>  			"--local-user" pgg-gpg-user-id)
>  		  (if pgg-text-mode '("--textmode"))))
>  	 (process (pgg-gpg-start-process args)))
> -    (unless pgg-gpg-use-agent
> -      (pgg-gpg-wait-for-status process '("GOOD_PASSPHRASE")))
> +    (pgg-gpg-wait-for-status process '("BEGIN_SIGNING"))
>      (process-send-region process start end)
>      (pgg-gpg-wait-for-completion process)
>      (save-excursion

Can't we change it, so that if pgg-gpg-use-agent is nil, it will wait
for either GOOD_PASSPHRASE or BEGIN_SIGNING?  Then it will work fine
for everyone (both new and old GnuPG) except those with old GnuPG and
smart cards, I think.

I think we can get it so that pgg-gpg-use-agent is only ever needed if
you use an old GnuPG with smart cards.

We _could_ fix that too, and avoid pgg-gpg-use-agent completely, if we
also searched for the 'gpg: writing to' string.  But maybe that make
the code too ugly to maintain.