Change in rmail-start-mail

Change in rmail-start-mail

[Eli Zaretskii]

This change:

  revno: 100059.1.49
  committer: Kenichi Handa <handa@m17n.org>
  branch nick: 23-build
  timestamp: Tue 2011-02-22 00:19:08 +0000
  message:
    Decode "encoded-words" of header components on replying.

breaks rmail-reply for me: the outgoing message has a garbled header
for each header that was RFC 2047 encoded in the original message to
which I reply.

Why was this change made?  What was the problem it was supposed to
solve?

It is quite clear that the assumption was that some other code will
RFC 2047 encode these headers before the message is sent, but what
code is supposed to do that?  I use sendmail-user-agent and
smtpmail-send-it, and these do not encode those headers.  Should the
decoding introduced with the above revision be conditioned on some
specific MUA?

Re: Change in rmail-start-mail

[Stefan Monnier]

> Why was this change made?

So that the text displayed while you edit the buffer is correct rather
than being some "weird encoding".

> What was the problem it was supposed to solve?

I guess the above.

> It is quite clear that the assumption was that some other code will
> RFC 2047 encode these headers before the message is sent, but what
> code is supposed to do that?

The message sending function needs to do that in any case since
non-ascii chars may have been added manually anyway.

> I use sendmail-user-agent and smtpmail-send-it, and these do not
> encode those headers.

Looks like they have a problem.

> Should the decoding introduced with the above revision be conditioned
> on some specific MUA?

Maybe until the above functions are either fixed or declared obsolete,
we could maybe try and add a hack to leave the encoded text in the
edition buffer, but that sounds like pretty sub-standard behavior.


        Stefan

Re: Change in rmail-start-mail

[Eli Zaretskii]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Date: Sat, 12 Mar 2011 15:57:30 -0500
> Cc: emacs-devel@gnu.org, Kenichi Handa <handa@m17n.org>
> 
> > Why was this change made?
> 
> So that the text displayed while you edit the buffer is correct rather
> than being some "weird encoding".

No, I think Handa-san wrote something about double encoding.

Anyway, rmail-yank does decode the headers _in_the_body_ of the
message, so the text is legible where it matters.

> > It is quite clear that the assumption was that some other code will
> > RFC 2047 encode these headers before the message is sent, but what
> > code is supposed to do that?
> 
> The message sending function needs to do that in any case since
> non-ascii chars may have been added manually anyway.
> 
> > I use sendmail-user-agent and smtpmail-send-it, and these do not
> > encode those headers.
> 
> Looks like they have a problem.

Well, I asked which ones don't, and where's the code those others use
to encode the headers.  (I mean where they call rfc2047-encode-*.)
Hopefully, someone will show me that code.  Then perhaps the others
can be fixed to DTRT as well.  (They should have been fixed as part of
the said change, IMO.)