Re: MIME and mailcrypt

Re: MIME and mailcrypt

[Lars Magne Ingebrigtsen]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> Is is feasable to do the actual insertion immediately and "hide" it so that
> PGP signing or encrypting works properly?

No, because charset encoding will also mess up PGP signing.

WHere is the signing/encryption done from?  `message-send-hook'?  If
so, I can just move the MML handling to before that hook.

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

Re: MIME and mailcrypt

[Kai.Grossjohann]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

  > It's being done from the message buffer, well before message-send-hook,
  > with mailcrypt being invoked manually (C-c / s).  Which is why I asked if
  > it were feasable to do all that stuff immediately.

Maybe Mailcrypt should be changed so that it signs or encrypts shortly
before the message goes out?  One could put a
mc-maybe-sign-or-encrypt-message function rather late in the hook, and
the signing and encryption commands just set a variable.

Or is that a no-no for some reason?

kai
-- 
This gubblick contains many nonsklarkish English flutzpahs,
but the overall pluggandisp can be glorked from context. -- David Moser

Re: MIME and mailcrypt

[Hrvoje Niksic]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> Least serious but most problematic is signing vs. signing and
> encrypting vs. doing nothing at all.  For example, most of the time
> I sign, sometimes I sign and encrypt, and occasionally I do neither.
> Mailcrypt cannot be totally automated.  Not to mention that this
> would require a radical change in how mailcrypt users use mailcrypt.

Can't you be asked about it?  I, for one, have no idea how mailcrypt
is used.

> Most serious is if if mc-sign or mc-encrypt are called as
> message-mode is attempting to hand off the message to the transport
> and a bad pass phrase is entered.  Either message will bomb out,
> leaving the message buffer in a not particularly happy state,

Message could simply remember the state of the original buffer, and
restore it if anything bad happens.  This would be useful for other
errors, and could be trivially implemented with a condition-case.

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
Oh lord won't you buy me a color TV...

Re: MIME and mailcrypt

[Jari Aalto+list.ding]

-----BEGIN PGP SIGNED MESSAGE-----

##
Subject: Re: MIME and mailcrypt

| 1998-12-02 Stainless Steel Rat <ratinox@peorth.gweep.net> list.ding
| "KG" = Kai Grossjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> writes:
|
| KG> Maybe Mailcrypt should be changed so that it signs or encrypts shortly
| KG> before the message goes out?

Could gnus recognize if buffer has been PGP-touched? (signed encrypted ...)
- --> If it contains PGP, then gnus steps away and does no buffer modifications.
This would preserve the PGP integrity.

jari

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3ia
Charset: noconv
Comment: Processed by Emacs TinyPgp 2.97

iQBVAwUBNmZP78C67dVHFB01AQH+eQIAjoeYv+ns4rCo7bKBxj5tkjNj9MsGyXb9
mHAlVI33TFV870Mjgrb1dAuP60dzkA++7upCR17sjDNa8DD0mVjpEA==
=8oA2
-----END PGP SIGNATURE-----

Re: MIME and mailcrypt

[Kai.Grossjohann]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

  > Least serious but most problematic is signing vs. signing and
  > encrypting vs. doing nothing at all.  For example, most of the
  > time I sign, sometimes I sign and encrypt, and occasionally I do
  > neither.  Mailcrypt cannot be totally automated.  Not to mention
  > that this would require a radical change in how mailcrypt users
  > use mailcrypt.

I didn't say that mailcrypt should decide all by itself.  I suggested
the following: Users type C-c / s to sign and/or C-c / e to encrypt,
but mailcrypt doesn't do it immediately, it only remembers to do it
later, from message-send-hook.

kai
-- 
This gubblick contains many nonsklarkish English flutzpahs,
but the overall pluggandisp can be glorked from context. -- David Moser

Re: MIME and mailcrypt

[Lars Magne Ingebrigtsen]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> It's being done from the message buffer, well before message-send-hook,
> with mailcrypt being invoked manually (C-c / s).  Which is why I asked if
> it were feasable to do all that stuff immediately.

Yes, it can be done.  `message-encode-message-body' is the function
responsible for doing the MML thing on the body, so perhaps that
function should be made into a command and bound to a key or
something... 

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

Re: MIME and mailcrypt

[Lars Magne Ingebrigtsen]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> New versions of mailcrypt should get the raw buffer just on general
> principle, anyway.  The change for that is trivial: in mc-toplev.el,
> gnus-summary-select-article should be given a non-nil argument.  This
> appended patch is against mailcrypt 3.5b7 but should work for earlier
> versions.  Anyone know if this will break any version dependencies with
> Gnus?

Couldn't mc do its work in `gnus-article-original-buffer', which is
guaranteed not to have any Mimetilations mangling done to it?

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

Re: MIME and mailcrypt

[Lars Magne Ingebrigtsen]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> I suppose so... if gnus-article-original-buffer were defined somewhere.
> `gnus-article-original-buffer' does not exist anywhere I can grep in pgnus
> 0.62.

gnus-original-article-buffer.  :-)

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

Re: MIME and mailcrypt

[Dave Love]

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

 Lars> Couldn't mc do its work in `gnus-article-original-buffer',
 Lars> which is guaranteed not to have any Mimetilations mangling done
 Lars> to it?

The current version should be doing that already; I sent a patch which
was in at some stage.  (It's relevant other than for MIMEcry,
specifically for washing functions that mung the buffer.)

Re: MIME and mailcrypt

[Stephen Zander]

>>>>> "Rat" == Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

    Rat> This is a MIME multipart message.  If you are reading this,
    Rat> you shouldn't.  -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

    Rat> This is a MIME multipart message.  If you are reading this,
    Rat> you shouldn't.

    Rat> - --==-=-=

Was this supposed to be a well-formed MIME message or did mailcrypt
mangle it?

-- 
Stephen
---
It should be illegal to yell "Y2K" in a crowded economy.  :-) -- Larry Wall

Re: MIME and mailcrypt

[Karl Kleinpaste]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:
> PGP armored it.

If there is a general "interpret *Article* as MIME" function, it could 
be manually invoked after `W W p' (gnus-article-hide-pgp).

It ain't pretty, but it would at least begin to work.

Re: MIME and mailcrypt

[Hrvoje Niksic]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> "SZ" == Stephen Zander <gibreel@pobox.com> writes:
> 
> Rat> - --==-=-=
> 
> SZ> Was this supposed to be a well-formed MIME message or did mailcrypt
> SZ> mangle it?
> 
> PGP armored it.
> 
> Do you begin to see some of the problems with mixing PGP and MIME?

Wouldn't this work right if Message knew about multipart/signed,
described in rfc2015?  Then Gnus could, at least in theory, call the
correct mailcrypt hooks to check the signature, and then display the
second part of the multipart as usual MIME stuff.

Or, am I missing something?

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
A radioactive cat has eighteen half-lives.

Re: MIME and mailcrypt

[Hrvoje Niksic]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> "Hrvoje" == Hrvoje Niksic <hniksic@srce.hr> writes:
> 
> Hrvoje> Or, am I missing something?
> 
> I think what you are missing is that mailcrypt (through PGP)
> encloses the entire message buffer, from the '--text follows this
> line--' line to (point-max).  In other words, mailcrypt has no idea
> what a multipart message is.

This sounds like a design deficiency in mailcrypt.  But it sounds
fixable; a PGP-enhanced message.el could create the MIME wonder, and
then call mailcrypt on the result, and then trivially insert the
multipart/signed junk around it.  This could be done around the time
when `message-send-hook' is run, or even after that.

The only thing that should be "fixed" in mailcrypt is the dependency
on "--text follows this line--"-style message; ideally, a mailcrypt
function should be called to handle the encryption/signing of a
textual region, and it should return the result in Lisp.

If this process bombs out, the user should be presented with the
original message buffer, MML and all.

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
Don't get even, get odd!

Re: MIME and mailcrypt

[David S. Goldberg]

> [...] ; ideally, a mailcrypt function should be called to handle the
> encryption/signing of a textual region [...] 

That is essentially how TM generates pgp-mime; it calls internal
mailcrypt functions around the region representing the part being
signed.

-- 
Dave Goldberg
Post: The Mitre Corporation\MS B305\202 Burlington Rd.\Bedford, MA 01730
Phone: 781-271-3887
Email: dsg@mitre.org

Re: MIME and mailcrypt

[Hrvoje Niksic]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> Hrvoje> The only thing that should be "fixed" in mailcrypt is the
> Hrvoje> dependency on "--text follows this line--"-style message;
> Hrvoje> ideally, a mailcrypt function should be called to handle the
> Hrvoje> encryption/signing of a textual region, and it should return
> Hrvoje> the result in Lisp.
> 
> I think this is reaching the point where message needs to use parts
> of mailcrypt as a library.  mc-pgp-sign-region and
> mc-pgp-encrypt-region would be the functions to ultimately be
> invoked.

Yes; that's how TM is reported to work.  Adding this to Gnus now
remains a small matter of programming.  Would such an addition really
be illegal to export from the US?

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
When a girl can read the handwriting on the wall, she may be in the
wrong rest room.

Re: MIME and mailcrypt

[Colin Rafferty]

Stainless Steel Rat writes:
> "Hrvoje" == Hrvoje Niksic <hniksic@srce.hr> writes:

Hrvoje> Yes; that's how TM is reported to work.  Adding this to Gnus now
Hrvoje> remains a small matter of programming.  Would such an addition
Hrvoje> really be illegal to export from the US?

> Yes, it is.  The way the Department of Justice got ITAR (International
> Trade in Arms Regulations) worded, it is functionality that is restricted.
> Although the DoJ seems to turn a blind eye to hooks, so long as the strong
> crypto is either not there or disabled.  To wit, Netscape Navigator export
> versions, which can be trivially patched to use 128-bit keys instead of
> 40-bit keys.

Then doesn't this mean that Gnus w/mailcrypt is still legal to export, 
since it actually isn't doing encryption, but calling a program that
does it for you.

Really, mailcrypt (and Gnus calling it) is only a hook.  We could tack 
on a weak encryption program, or a strong one -- it wouldn't be able
to tell the difference.

> Then again, there is the fact that if the code is written outside the US,
> the DoJ can't bitch about it (which is what we did for PGP 2.3 and 2.3a).

This doesn't matter, since we cannot re-export it from xemacs.org (in
the US).  It would really suck to have a part of XEmacs not under the
same cvs tree as the rest.

-- 
Colin

Re: MIME and mailcrypt

[Lars Magne Ingebrigtsen]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> I would not worry too much about it.  If the DoJ wanted to go after the
> XEmacs people, they would have done it over crypt++ a long time ago.  If
> the DoJ tried to step in now, the EFF would be all over them.

Probably.  But I really don't want to add any code to Gnus that could
possibly get American redistributors landed in jail.  It's not my neck 
that's on the line; it's others, and I like to be careful about other
people's necks.  

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

Re: MIME and mailcrypt

[Dave Love]

FWIW, the current Emacs TODO says:

* Implement a variant of uncompress.el or jka-compr.el that
  works with GNU Privacy Guard for encryption.

I don't know how the interface to GPG compares with PGP, but
presumably it's similar enough that the mechanism should support both.

Re: MIME and mailcrypt

[Dave Love]

>>>>> "Rat" == Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

 Rat> "DL" == Dave Love <d.love@dl.ac.uk> writes:

 DL> * Implement a variant of uncompress.el or jka-compr.el that
 DL> works with GNU Privacy Guard for encryption.

 Rat> mailcrypt 3.5.1 supports GPG, at least to a limited degree.  But
 Rat> we are again getting into the same area as PGP/MIME support.

Sure.  I was thinking that this implied that there should be canned
functionality for it in Emacs (the `format-alist' stuff seems
appropriate) and that rms wanted it included, presumably cognizant of
the legalities, since Lars was expressing doubts.

Re: MIME and mailcrypt

[Hrvoje Niksic]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> Then again, there is the fact that if the code is written outside
> the US, the DoJ can't bitch about it (which is what we did for PGP
> 2.3 and 2.3a).

Lars is in Norway.  Does the above means that, if he writes the code,
it would be re-exportable from the US?

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
"Silence!" cries Freydag. "I did not call thee in for a consultation!"
"They are my innards! I will not have them misread by a poseur!"

Re: MIME and mailcrypt

[Colin Rafferty]

Hrvoje Niksic writes:

> Stainless Steel Rat <ratinox@peorth.gweep.net> writes:
>> Then again, there is the fact that if the code is written outside
>> the US, the DoJ can't bitch about it (which is what we did for PGP
>> 2.3 and 2.3a).

> Lars is in Norway.  Does the above means that, if he writes the code,
> it would be re-exportable from the US?

The answer is no; the US is the black hole of cryptography.

There is no distinction between exporting something home-grown and
exporting something that has just been imported.

-- 
Colin

Re: MIME and mailcrypt

[Stephen Zander]

>>>>> "Rat" == Stainless Steel Rat <ratinox@peorth.gweep.net> writes:
    Rat> Not any more.  32 other countries just signed an agreement
    Rat> reducing them to the same level as the US.

Yeah :( Somebody want to send me the immigration application for
Norway? :/

-- 
Stephen
---
It should be illegal to yell "Y2K" in a crowded economy.  :-) -- Larry Wall