pgnus-0.77 bug in forwarding mime messages

pgnus-0.77 bug in forwarding mime messages

[lantz moore]

i just upgraded to 0.77 and thought i'd give the mml stuff a try.  so i
composed a pretty mime message to myself.  worked like a charm.

then i forwarded said message to myself.  the resulting message didn't
turn out quite like i expected.  neither gnus nor metamail could decipher
the mime components of the original message.  all the '=' in the original
message were encoded to =3D (which seems resonable, i guess).  this seemed
somewhat suspect, so i edited the raw nnml file and changed all the =3D
back to '=', at that point gnus and metamail could decode the content
inside the message/rfc822 part.

-- 
lantz moore, contigo software                              lmoore@contigo.com

Re: pgnus-0.77 bug in forwarding mime messages

[Lars Magne Ingebrigtsen]

lantz moore <lantzm@contigo.com> writes:

> then i forwarded said message to myself.  the resulting message didn't
> turn out quite like i expected.  neither gnus nor metamail could decipher
> the mime components of the original message.  all the '=' in the original
> message were encoded to =3D (which seems resonable, i guess). 

Hm.  Sounds like something qp-encododed the message.  Did the headers
say something about Content-Transfer-Encoding or the like?

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: pgnus-0.77 bug in forwarding mime messages

[lantz moore]

>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

Lars> lantz moore <lantzm@contigo.com> writes:
>> then i forwarded said message to myself.  the resulting message didn't
>> turn out quite like i expected.  neither gnus nor metamail could
>> decipher the mime components of the original message.  all the '=' in
>> the original message were encoded to =3D (which seems resonable, i
>> guess).

Lars> Hm.  Sounds like something qp-encododed the message.  Did the
Lars> headers say something about Content-Transfer-Encoding or the like?

the headers on the part are:
Content-Type: message/rfc822
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

mm-encode.el is definitely qp-encododing the message/rfc822 part on
purpose.

from mm-encode.el
(defvar mm-content-transfer-encoding-defaults
  '(("text/.*" quoted-printable)
    ("message/rfc822" quoted-printable)
    ("application/emacs-lisp" 8bit)
    ("application/x-patch" 8bit)
    (".*" base64))
  "Alist of regexps that match MIME types and their encodings.")

i changed the message/rfc822 from quoted-printable to 8bit in the alist
and forwarded the message to myself again.  using 8bit, things worked like
you would think that they would, ie, both gnus and metamail were able to
render the message.  but i still don't see what problem the
quoted-printable content-transfer-encoding is causing.

you should be able to reproduce this yourself by sending yourself a gif
and then forwarding that message back to yourself...  (pgnus-0.77).

on a somewhat related point, why do we need to qp-encode text/.* and
message/rfc822?  shouldn't we really only have to quote an occurance of
the boundary in the message text (which is quite unlikey anyway)?  just
curious.

-l

Re: pgnus-0.77 bug in forwarding mime messages

[Lars Magne Ingebrigtsen]

lantz moore <lmoore@contigo.com> writes:

> the headers on the part are:
> Content-Type: message/rfc822
> Content-Disposition: inline
> Content-Transfer-Encoding: quoted-printable
> 
> mm-encode.el is definitely qp-encododing the message/rfc822 part on
> purpose.

And RFC2046 forbids that.  Fix in Pterodactyl Gnus v0.80.

> on a somewhat related point, why do we need to qp-encode text/.* and
> message/rfc822?  shouldn't we really only have to quote an occurance of
> the boundary in the message text (which is quite unlikey anyway)?  just
> curious.

mml selects the MIME boundaries carefully -- it looks into each part,
and if it finds the boundary it was going to use in there, it selects
a new boundary.  So boundary selection is not an encoding issue.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen