Re: MIME and PGP (was: Gnus and Crypto stuff)

[Jack Vinson <jvinson@chevax.ecs.umass.edu>]

>>>>> "KK" == Karl Kleinpaste <karl@justresearch.com> writes:

KK> Jack Vinson <jvinson@chevax.ecs.umass.edu> writes:
>> It looked to me like there should have been about three parts, but
>> that the PGP signature broke the MIME parsing somehow.

KK> Assuming that one wants to PGP-sign an entire message, one needs the
KK> ability for Gnus to MIME-encode the whole thing and then let MailCrypt
KK> do its PGP thing to the result.  Similarly, on receipt, we need the
KK> ability to make gnus-article-hide-pgp happen rather early on, followed
KK> by MIME decoding.  The fundamental problem we're up against is that,
KK> at least for the recipient, PGP armor is removed rather late in the
KK> game, long after Gnus has reached its conclusions about the MIME
KK> components in the message.  PGP needs to be the last thing added and
KK> the first thing removed.

KK> I'm not really sure what semantic would be in order, if one wanted to
KK> PGP-sign individual components of a message, rather than the whole
KK> thing.

At least for message-send, the current set of actions is something like
	- fixup the message
	- encode the message body
	- run message-send-hook
	- send the message
		- message-header-hook, message-send-mail-hook or
		  message-send-news-hook 
	- deal with fcc
	- run message-sent-hook
	- housekeeping

It looks like message-send-mail/news-hook are run with the final version of
the message (i.e. just before it goes out to the world).  I am guess this
is where the pgp stuff *should* be called.  But that means adding pgp to
two hooks instead of just one.

As I mentioned earlier, I would like a hook that is called before the
message body is encoded so that things like ispell-message don't have to
deal with the MIME stuff.

-- 
Jack Vinson <jvinson@chevax.ecs.umass.edu>    http://www.cis.upenn.edu/~vinson/
Zippy: I'm shaving!!  I'M SHAVING!!