PGP support

PGP support

[Simon Josefsson]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Thanks to Sascha Lüdecke ROT13(<fnfpun@zrgn-k.qr>) we now have PGP (as
in RFC 1991, not 2015/3156 which we already have) support in CVS.

As you can see, the PGP armors are QPified.  I suspect this isn't
perfect, but I can't find a good document that says how this should
work.  RFC 2440 section 7 talks a little about it, but I'm not sure if
that is what is used out there?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE78EOZ8U/viFyYAJcRAm7lAKDFx/Mg3DDaEX51Ua5vJxc+UScOiwCgjdum
Y6fpjEXoEa8nqCAwb5qgsGk=
=JPtk
-----END PGP SIGNATURE-----

Re: PGP support

[Matt Armstrong]

Simon Josefsson <jas@extundo.com> writes:

> Thanks to Sascha Lüdecke ROT13(<fnfpun@zrgn-k.qr>) we now have PGP (as
> in RFC 1991, not 2015/3156 which we already have) support in CVS.

For the RFC ignorant, what is the difference?

I think we had PGP/MIME.  What is RFC 1991?


> As you can see, the PGP armors are QPified.

For the lingo ignorant like myself, what does QPified mean?  :-)


-- 
matt

Re: PGP support

[Josh Huber]

"Matt Armstrong" <matt+dated+1008201839.c20230@lickey.com> writes:

> I think we had PGP/MIME.  What is RFC 1991?

Apperently that's the original RFC for clearsigning messages.
Although I've never seen QP encoded non-pgp-mime messages before...

> For the lingo ignorant like myself, what does QPified mean?  :-)

Quoted printable encoding, converts those characters that MTAs like to
munge, like "^From" at the start of a line when delivering into an
mbox folder.

ttyl,

-- 
Josh Huber

Re: PGP support

[Simon Josefsson]

On Mon, 12 Nov 2001, Matt Armstrong wrote:

> Simon Josefsson <jas@extundo.com> writes:
> 
> > Thanks to Sascha Lüdecke ROT13(<fnfpun@zrgn-k.qr>) we now have PGP (as
> > in RFC 1991, not 2015/3156 which we already have) support in CVS.
> 
> For the RFC ignorant, what is the difference?
> 
> I think we had PGP/MIME.  What is RFC 1991?

PGP (RFC 1991/2440) means plain (Open)PGP blobs inserted into mail without
any MIME awareness at all.  It is what Mailcrypt natively supports.  
PGP/MIME (RFC 2015/3156) is PGP with MIME awareness.  Oort has supported
PGP/MIME for some time, using gpg.el or Mailcrypt as the backend.

With a few more menu items and commands added to Gnus, such as "Insert
Public Key", I don't have to use the Mailcrypt menu hooks at all.

Also, it would be nice if we could interop with those Outlook PGP plugins.  
It seems as if they support a MIME-aware PGP mode that isn't 2015/3156, 
but rather having all MIME headers etc look as if it was a unencrypted 
mail, and before sending, wrapping each MIME body in a PGP blob.  Everyone 
I communicate with seem to use the stuff, so I'll probably add it.

Maybe we should add a FSF-owned PGP-backend as well.  The FLIM/SEMI people
seem to have one, but it requires a few files that look old and
non-FSF-owned.  Hm..

> > As you can see, the PGP armors are QPified.
> 
> For the lingo ignorant like myself, what does QPified mean?  :-)

QP is a horrible escape language for non-ASCII stuff in mail.  "To QPify" 
means to add QP encoding in a piece of data.  That's about all you want to 
know about QP, I think. :-)

Re: PGP support

[Fabien Penso]

[-- Attachment #1: Type: text/plain, Size: 1169 bytes --]


Simon on Tue, 13 Nov 2001 10:53:53 +0100 (CET) wrote:

 >> > Thanks to Sascha Lüdecke ROT13(<fnfpun@zrgn-k.qr>) we now have PGP (as
 >> > in RFC 1991, not 2015/3156 which we already have) support in CVS.
 >> 
 >> For the RFC ignorant, what is the difference?
 >> 
 >> I think we had PGP/MIME.  What is RFC 1991?

 > PGP (RFC 1991/2440) means plain (Open)PGP blobs inserted into mail without
 > any MIME awareness at all.  It is what Mailcrypt natively supports.  
 > PGP/MIME (RFC 2015/3156) is PGP with MIME awareness.  Oort has supported
 > PGP/MIME for some time, using gpg.el or Mailcrypt as the backend.

 > With a few more menu items and commands added to Gnus, such as "Insert
 > Public Key", I don't have to use the Mailcrypt menu hooks at all.

Good ! Mailcrypt doesn't support sig with international characters
(which sux). If Gnus now does, I'll be able to use plain text sign for
my usenet posts.

 > Also, it would be nice if we could interop with those Outlook PGP plugins.  

Yeap. I agree.

-- 
Fabien Penso <penso@linuxfr.org> | LinuxFr a toujours besoin de :
http://perso.LinuxFR.org/penso/  | http://linuxfr.org/dons/


[-- Attachment #2: Type: application/pgp-signature, Size: 239 bytes --]

Re: PGP support

[Simon Josefsson]

On Tue, 13 Nov 2001, Fabien Penso wrote:

>  > PGP (RFC 1991/2440) means plain (Open)PGP blobs inserted into mail without
>  > any MIME awareness at all.  It is what Mailcrypt natively supports.  
>  > PGP/MIME (RFC 2015/3156) is PGP with MIME awareness.  Oort has supported
>  > PGP/MIME for some time, using gpg.el or Mailcrypt as the backend.
> 
>  > With a few more menu items and commands added to Gnus, such as "Insert
>  > Public Key", I don't have to use the Mailcrypt menu hooks at all.
> 
> Good ! Mailcrypt doesn't support sig with international characters
> (which sux). If Gnus now does, I'll be able to use plain text sign for
> my usenet posts.

The PGP/MIME standard allows non-ASCII, so this should already work.  If
people like PGP and/or MIME on the UseNet is a different question though,
and the answer seems to depend on what hierarchy you are in.

Re: PGP support

[Florian Weimer]

"Matt Armstrong" <matt+dated+1008201839.c20230@lickey.com> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> Thanks to Sascha Lüdecke ROT13(<fnfpun@zrgn-k.qr>) we now have PGP (as
>> in RFC 1991, not 2015/3156 which we already have) support in CVS.
>
> For the RFC ignorant, what is the difference?
>
> I think we had PGP/MIME.  What is RFC 1991?

Is the predecessor of RFC 2440 (OpenPGP).  It hasn't got much to do
with Internet mail or news, so it's a red herring.

Re: PGP support

[Simon Josefsson]

Florian Weimer <fw@deneb.enyo.de> writes:

> "Matt Armstrong" <matt+dated+1008201839.c20230@lickey.com> writes:
>
>> Simon Josefsson <jas@extundo.com> writes:
>>
>>> Thanks to Sascha Lüdecke ROT13(<fnfpun@zrgn-k.qr>) we now have PGP (as
>>> in RFC 1991, not 2015/3156 which we already have) support in CVS.
>>
>> For the RFC ignorant, what is the difference?
>>
>> I think we had PGP/MIME.  What is RFC 1991?
>
> Is the predecessor of RFC 2440 (OpenPGP).  It hasn't got much to do
> with Internet mail or news, so it's a red herring.

Unfortunately it is the only thing that Outlook PGP plugins seem to
grok...  (At least if you regard 1991 and 2440 as almost the same
thing, which, from Gnus' point of view, they are.)

Re: PGP support

[Per Abrahamsen]

Simon Josefsson <jas@extundo.com> writes:

> As you can see, the PGP armors are QPified.

Actually, I can't.  Probably because of 

          X-MIME-Autoconverted: from quoted-printable to 8bit by sheridan.dina.kvl.dk id WAA23581

Would such convertions affect PGP?

Re: PGP support

[Simon Josefsson]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> As you can see, the PGP armors are QPified.
>
> Actually, I can't.  Probably because of 
>
>           X-MIME-Autoconverted: from quoted-printable to 8bit by sheridan.dina.kvl.dk id WAA23581
>
> Would such convertions affect PGP?

It shouldn't, but it might.  Could you verify the PGP signature?

Re: PGP support

[Per Abrahamsen]

Simon Josefsson <jas@extundo.com> writes:

> It shouldn't, but it might.  Could you verify the PGP signature?

I have no idea.  I have never used PGP on Unix, and neither "W g" nor
"W s" does anything visible.

Re: PGP support

[Josh Huber]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> Actually, I can't.  Probably because of 
>
>           X-MIME-Autoconverted: from quoted-printable to 8bit by
>           sheridan.dina.kvl.dk id WAA23581
>
> Would such convertions affect PGP?

Well, did the message verify for you?

I suspect that the message is signed first, and then encoded QP, which
would explain the QP encoding of the PGP headers.

FWIW mutt does not do this.  Does Outhouse Express support PGP signed
messages that are QP encoded like this?

ttyl,
-- 
Josh Huber

Re: PGP support

[Matt Armstrong]

Josh Huber <huber@alum.wpi.edu> writes:

> Well, did the message verify for you?
>
> I suspect that the message is signed first, and then encoded QP, which
> would explain the QP encoding of the PGP headers.
>
> FWIW mutt does not do this.  Does Outhouse Express support PGP
> signed messages that are QP encoded like this?

I didn't even realize that Simon's original post was signed.  Is Gnus
supposed give some kind of visual cue that the message was signed?

I have this setup in .gnus, where my-save-require catches any require
errors and just returns nil -- but in this case 'gpg was successfully
loaded.

(cond
 ((my-safe-require 'gpg)
  (setq mml2015-use 'gpg)
  (setq gpg-temp-directory (expand-file-name "~/.gnupg/tmp"))
  (setq gpg-command-default-alist (quote ((gpg . "gpg")
					  (gpg-2comp . "gpg"))))))


-- 
matt

Re: PGP support

[Simon Josefsson]

"Matt Armstrong" <matt+dated+1008276238.4e3562@lickey.com> writes:

> Josh Huber <huber@alum.wpi.edu> writes:
>
>> Well, did the message verify for you?
>>
>> I suspect that the message is signed first, and then encoded QP, which
>> would explain the QP encoding of the PGP headers.
>>
>> FWIW mutt does not do this.  Does Outhouse Express support PGP
>> signed messages that are QP encoded like this?
>
> I didn't even realize that Simon's original post was signed.  Is Gnus
> supposed give some kind of visual cue that the message was signed?

There is a `p' in the Article buffer's modeline.  Using the fringe for
this is one idea.  You can also get big buttons if you like to clutter
up your article buffer:

(setq gnus-buttonized-mime-types
      '("multipart/encrypted" "multipart/signed"))

Re: PGP support

[Matt Armstrong]

Simon Josefsson <jas@extundo.com> writes:

> "Matt Armstrong" <matt+dated+1008276238.4e3562@lickey.com> writes:
>
>> I didn't even realize that Simon's original post was signed.  Is Gnus
>> supposed give some kind of visual cue that the message was signed?
>
> There is a `p' in the Article buffer's modeline.  Using the fringe for
> this is one idea.

Yes, a single 'p' is not very obvious.  ;-)

Also, without buttons, how are you supposed to verify the signatures?
I saw no obvious way -- W p didn't do much.


> You can also get big buttons if you like to clutter up your article
> buffer:
>
> (setq gnus-buttonized-mime-types
>       '("multipart/encrypted" "multipart/signed"))

Ahh, cool, there is the visual cue I was looking for, as well as the
UI necessary to see gpg's output, etc.  Perhaps this should be the
default?


-- 
matt

Re: PGP support

[Simon Josefsson]

"Matt Armstrong" <matt+dated+1008279199.7e6a5d@lickey.com> writes:

> Yes, a single 'p' is not very obvious.  ;-)
>
> Also, without buttons, how are you supposed to verify the signatures?
> I saw no obvious way -- W p didn't do much.

Try `W s'.  Oops, this wasn't documented.  Fixed.

>> You can also get big buttons if you like to clutter up your article
>> buffer:
>>
>> (setq gnus-buttonized-mime-types
>>       '("multipart/encrypted" "multipart/signed"))
>
> Ahh, cool, there is the visual cue I was looking for, as well as the
> UI necessary to see gpg's output, etc.  Perhaps this should be the
> default?

It was the default, but perhaps it interfered with reading.  If we can
find a non-obtrusive mechanism to let the user know the this
information, it would be good.

Also, the buttons create false security.  I posted a mail to this list
that showed PGP buttons that said the message verified OK, but I
created them using ^H which Gnus bolidifies.  Whatever the mechanism
will be, the article buffer is the wrong place for it.

Re: PGP support

[Per Abrahamsen]

Simon Josefsson <jas@extundo.com> writes:

> It was the default, but perhaps it interfered with reading.  

Yes, one requirement for signing to be popular is that it doesn't
annoy people who don't care about it.

Re: PGP support

[dme]

* jas@extundo.com [2001-11-13 21:52:16]
> Also, the buttons create false security.  I posted a mail to this
> list that showed PGP buttons that said the message verified OK, but
> I created them using ^H which Gnus bolidifies.  Whatever the
> mechanism will be, the article buffer is the wrong place for it.

How about a background pixmap which indicates the verification state
of text ?

dme.

Re: PGP support

[Simon Josefsson]

On Wed, 14 Nov 2001 dme@dme.org wrote:

> * jas@extundo.com [2001-11-13 21:52:16]
> > Also, the buttons create false security.  I posted a mail to this
> > list that showed PGP buttons that said the message verified OK, but
> > I created them using ^H which Gnus bolidifies.  Whatever the
> > mechanism will be, the article buffer is the wrong place for it.
> 
> How about a background pixmap which indicates the verification state
> of text ?

Then I'll just send a text/html mail with the pixmap as background.  The 
article buffer can't be used securely for this purpose.

The modeline is perhaps a good place, if we can get a nice icon that says
"Signature Good", "Signature Bad", "Signature OK but sender
unknown/untrusted".  Any takers?

Re: PGP support

[Andreas Fuchs]

[-- Attachment #1.1: Type: text/plain, Size: 608 bytes --]

Today, Simon Josefsson <jas@extundo.com> wrote:
> The modeline is perhaps a good place, if we can get a nice icon that
> says "Signature Good", "Signature Bad", "Signature OK but sender
> unknown/untrusted".  Any takers?

(playing devil's advocate:) What about more than one part? Also, an evil
one could just go ahead and sign an empty line of text (with an
untrusted key). Then, I presume, the user will see the "OK but
untrusted" message. I would just die to see the fringe be used for this
(or an overlay or whatever, for xemacs).

-- 
Andreas Fuchs, <asf@acm.org>, asf@jabber.at, antifuchs

[-- Attachment #2: Type: application/pgp-signature, Size: 231 bytes --]

Re: PGP support

[Simon Josefsson]

Andreas Fuchs <asf@void.at> writes:

> Today, Simon Josefsson <jas@extundo.com> wrote:
>> The modeline is perhaps a good place, if we can get a nice icon that
>> says "Signature Good", "Signature Bad", "Signature OK but sender
>> unknown/untrusted".  Any takers?
>
> (playing devil's advocate:) What about more than one part? Also, an evil
> one could just go ahead and sign an empty line of text (with an
> untrusted key). Then, I presume, the user will see the "OK but
> untrusted" message.

Right.  Not good.  I haven't seen any attempt at solving this problem
though.  Partially signed MIME bodies are too complex to comprehend
for users that aren't familiar with how things are implemented.

> I would just die to see the fringe be used for this (or an overlay
> or whatever, for xemacs).

Yes.  However, the more I think about it, it is not perfect either.  I
could send a (signed) text/html part that includes a cid: URL pointing
to another (unsigned) MIME part containing a gif picture of a Gnus
article buffer with some text in.  The fringe would then most likely
be highlighted for the lines where the image is (because they could
contain signed text as well), which would be incorrect.  It becomes
even more problematic if you consider charset-splitted MIME bodies.
How could you usefully highlight that some characters on a line are
digitally signed and the rest of the line wasn't?  We also have the
problem of giving a false impression of what is digitally signed or
encrypted -- the RFC 2822 headers never are.

It seems as if there is something fundamentally flawed here (perhaps
my reasoning :)), and we should give up.

Re: PGP support

[Nevin Kapur]

Simon Josefsson <jas@extundo.com> writes:

> It seems as if there is something fundamentally flawed here (perhaps
> my reasoning :)), and we should give up.

How about adding something that changes each time the article is
displayed?  What if the current time was added to the button?

*[[PGP Signed Part: XYZ <xyz@abc.xom> "Wed Nov 14 12:34:04 2001"]]*

-- 
Nevin

Re: PGP support

[Matt Armstrong]

Nevin Kapur <nevin@jhu.edu> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> It seems as if there is something fundamentally flawed here
>> (perhaps my reasoning :)), and we should give up.
>
> How about adding something that changes each time the article is
> displayed?  What if the current time was added to the button?
>
> *[[PGP Signed Part: XYZ <xyz@abc.xom> "Wed Nov 14 12:34:04 2001"]]*

I like that idea.  I think using buttons is the way to go.  I'm
thinking:

    - Inserting a "verified on TIME/DATE" into the button is a good
      idea, and greatly hinders people's ability to spoof it.

    - However, I'd want to be able to turn this off to save some
      screen real estate.  I am not personally worried about button
      spoofing.  In the cases where I truly care I'll verify the
      button is real by either moving the mouse over it (when under a
      GUI) or selecting the button to see the raw gpg output.

    - Buttons work both under a GUI and when emacs is running in a
      terminal.

-- 
matt

Re: PGP support

[Florian Weimer]

Nevin Kapur <nevin@jhu.edu> writes:

> How about adding something that changes each time the article is
> displayed?  What if the current time was added to the button?
>
> *[[PGP Signed Part: XYZ <xyz@abc.xom> "Wed Nov 14 12:34:04 2001"]]*

Ah, yes, that's the way to go!  I should have read the entire thread
first.

(mutt does the same thing, BTW, and it includes the complete GnuPG
output by default.)

Re: PGP support

[Paul Jarc]

Nevin Kapur <nevin@jhu.edu> wrote:
> How about adding something that changes each time the article is
> displayed?  What if the current time was added to the button?
>
> *[[PGP Signed Part: XYZ <xyz@abc.xom> "Wed Nov 14 12:34:04 2001"]]*

Another idea: when the signature is checked, modify the summary line -
add a mark or something.  That can't be faked even if articles can
dynamically include content from a URL embedded in the article, which
I imagine will probably happen at some point, if it's not already
possible.


paul

Re: PGP support

[Andreas Fuchs]

[-- Attachment #1.1: Type: text/plain, Size: 993 bytes --]

Today, Paul Jarc <prj@po.cwru.edu> wrote:
> Another idea: when the signature is checked, modify the summary line -
> add a mark or something.  That can't be faked even if articles can
> dynamically include content from a URL embedded in the article, which
> I imagine will probably happen at some point, if it's not already
> possible.

Another idea I just got would be some kind of MIME part tree, which
shows signature state. That way, you could even show the tree-like
structure of multipart-message attachments which themselves contain
multipart messages.

Something looking like this (inspired by speedbar):

[-] Part 1 [Message] (signed)
 [+] Part 1 [Message] (signed, failed)
 [ ] foobar.gif [Image] (signed)
[+] Part 2 [Message] (signed, untrusted)

I can imagine that this could only be compromised by a w3 page with
frames. Maybe I'll hack something together, but don't hold your breath
yet. (-:


-- 
Andreas Fuchs, <asf@acm.org>, asf@jabber.at, antifuchs

[-- Attachment #2: Type: application/pgp-signature, Size: 231 bytes --]

Re: PGP support

[Raymond Scholz]

Andreas Fuchs <asf@void.at> writes:

> Something looking like this (inspired by speedbar):
>
> [-] Part 1 [Message] (signed)
>  [+] Part 1 [Message] (signed, failed)
>  [ ] foobar.gif [Image] (signed)
> [+] Part 2 [Message] (signed, untrusted)
>
> I can imagine that this could only be compromised by a w3 page with
> frames. Maybe I'll hack something together, but don't hold your breath
> yet. (-:

Hint: looks very similar to the output you get if pressing `C-d' on a
MIME message.

Cheers, Ray
-- 
WYSIWYG is a step backwards. Human labor is used to do that which the
computer can do better.                         (Andrew S. Tanenbaum)

Re: PGP support

[Andreas Fuchs]

[-- Attachment #1: Type: text/plain, Size: 613 bytes --]

On 2001-11-18, Raymond Scholz <ray-2001@zonix.de> wrote:
>> I can imagine that this could only be compromised by a w3 page with
>> frames. Maybe I'll hack something together, but don't hold your
>> breath yet. (-:
> 
> Hint: looks very similar to the output you get if pressing `C-d' on a
> MIME message.

You're right. Been too long since I used that particular feature. Now,
if I could just display this buffer beside the message buffer (and still
have it convey the information I pointed out above), and we'd be half
done (-:

Thanks,
-- 
Andreas Fuchs, <asf@acm.org>, asf@jabber.at, antifuchs

[-- Attachment #2: Type: application/pgp-signature, Size: 231 bytes --]

Re: PGP support

[Florian Weimer]

Andreas Fuchs <asf@void.at> writes:

> (playing devil's advocate:) What about more than one part? Also, an evil
> one could just go ahead and sign an empty line of text (with an
> untrusted key). Then, I presume, the user will see the "OK but
> untrusted" message. I would just die to see the fringe be used for this
> (or an overlay or whatever, for xemacs).

Works only on X11, I think.  We have to include the necessary
information in the message body itself, with the current time.

Re: PGP support

[Florian Weimer]

Simon Josefsson <jas@extundo.com> writes:

> The modeline is perhaps a good place, if we can get a nice icon that says
> "Signature Good", "Signature Bad", "Signature OK but sender
> unknown/untrusted".  Any takers?

Some people run GNU Emacs inside GNU screen: I don't think they have
got icons in the modeline.

Re: PGP support

[Lars Magne Ingebrigtsen]

Simon Josefsson <jas@extundo.com> writes:

> The modeline is perhaps a good place, if we can get a nice icon that says
> "Signature Good", "Signature Bad", "Signature OK but sender
> unknown/untrusted".  Any takers?

For Emacs 21, the fringe could display a symbol to the left (or right)
of each part that gives you the status of the part.  For instance, use
the same symbol Mozilla does -- that lock thingie.  Broken when bad,
etc.

For fringe-less environments, other methods might be used.

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

Re: PGP support

[news]

Simon Josefsson <jas@extundo.com> writes:

> "Matt Armstrong" <matt+dated+1008279199.7e6a5d@lickey.com> writes:
>
>> Yes, a single 'p' is not very obvious.  ;-)
>
> It was the default, but perhaps it interfered with reading.  If we can
> find a non-obtrusive mechanism to let the user know the this
> information, it would be good.
>
> Also, the buttons create false security.  I posted a mail to this list
> that showed PGP buttons that said the message verified OK, but I
> created them using ^H which Gnus bolidifies.  Whatever the mechanism
> will be, the article buffer is the wrong place for it.

How about a header field:

  X-Gnus: pgp-verified signature with key 0x454545 valid untrusted

Chris

Re: PGP support

[Werner Koch]

On Fri, 16 Nov 2001 13:26:20 -0500, Chris Beggy said:

> How about a header field:

>   X-Gnus: pgp-verified signature with key 0x454545 valid untrusted

BTW, I might have missed it but an important feature would be to add a
line like:

 X-Gnus-Orig-Encrypted-To: 0x12345678, 0x34567890

So that one can see that the message was originally encrypted and even
more important to automagically suggest to encrypt any reply.

Ciao,

  Werner

Re: PGP support

[Simon Josefsson]

Werner Koch <wk@gnupg.org> writes:

> On Fri, 16 Nov 2001 13:26:20 -0500, Chris Beggy said:
>
>> How about a header field:
>
>>   X-Gnus: pgp-verified signature with key 0x454545 valid untrusted
>
> BTW, I might have missed it but an important feature would be to add a
> line like:
>
>  X-Gnus-Orig-Encrypted-To: 0x12345678, 0x34567890
>
> So that one can see that the message was originally encrypted and even
> more important to automagically suggest to encrypt any reply.

A header to say that replies should be encrypted seems useful, but it
isn't really Gnus specific, and I would also feel better if requests
for security are signed (i.e. in S/MIME SignerInfo, I don't know if
OpenPGP has something similar).  Hm.

I don't understand the value of having a header line that says the
message was originally encrypted, the client already knows this?  And
users should probably not trust such headers without the client saying
it is OK, and if the client can do that, the client could use some
other (better) way of conveying this information anyway.

Re: PGP support

[Josh Huber]

Simon Josefsson <jas@extundo.com> writes:

> I don't understand the value of having a header line that says the
> message was originally encrypted, the client already knows this?

Right...is there an easy way to access this information (if the
message was signed or encrypted) from a reply buffer?

I'm thinking about a way to implement the reply_sign, reply_encrypt,
and reply_signencrypt options of mutt...If you get a message which is
encrypted, and reply to it, Gnus would automatically encrypt the
reply.

-- 
Josh Huber

Re: PGP support

[Simon Josefsson]

Josh Huber <huber@alum.wpi.edu> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> I don't understand the value of having a header line that says the
>> message was originally encrypted, the client already knows this?
>
> Right...is there an easy way to access this information (if the
> message was signed or encrypted) from a reply buffer?

I don't think so... someone please contribute it. :-)

> I'm thinking about a way to implement the reply_sign, reply_encrypt,
> and reply_signencrypt options of mutt...If you get a message which is
> encrypted, and reply to it, Gnus would automatically encrypt the
> reply.

That would be cool.  Perhaps the idea could be integrated with BBDB as
well (or maybe it already is, I think I remember something similar).

Re: PGP support

[Jack Twilley]

>>>>> "Josh" == Josh Huber <huber@alum.wpi.edu> writes:

Josh> I'm thinking about a way to implement the reply_sign,
Josh> reply_encrypt, and reply_signencrypt options of mutt...If you
Josh> get a message which is encrypted, and reply to it, Gnus would
Josh> automatically encrypt the reply.

>>>>> "Simon" == Simon Josefsson <jas@extundo.com> writes:

Simon> That would be cool.  Perhaps the idea could be integrated with
Simon> BBDB as well (or maybe it already is, I think I remember
Simon> something similar).

bbdb-pgp is an addon.  When you C-c C-c, it checks the BBDB record of
the addressee.  If their record has the field 'pgp-mail' set to
'sign', it asks you if you want to sign the message.  If their record
has the field 'pgp-mail' set to 'encrypt', it asks you if you want to
encrypt the message.  As distributed, it's mailcrypt-specific.  The
changes to work with mml-secure are very very obvious one-liners.

Jack.
(I use it all the time)
-- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash

Re: PGP support

[Josh Huber]

Simon Josefsson <jas@extundo.com> writes:

> That would be cool.  Perhaps the idea could be integrated with BBDB
> as well (or maybe it already is, I think I remember something
> similar).

Yeah, I use a partially hacked version of bbdb-pgp.el to do this.

But, this just lets you say whether or not to /always/ sign or encrypt
to a certain person.  This works well, but if someone (say, not in
your bbdb) sends you encrypted mail, it's nice to return the favor :)

ttyl,

-- 
Josh Huber

Re: PGP support

[Jack Twilley]

>>>>> "Josh" == Josh Huber <huber@alum.wpi.edu> writes:

[...]

Josh> But, this just lets you say whether or not to /always/ sign or
Josh> encrypt to a certain person.  This works well, but if someone
Josh> (say, not in your bbdb) sends you encrypted mail, it's nice to
Josh> return the favor :)

What *I*'d like, personally, is to get that gpg-ring.el software
working, and then every week traverse my gpg key ring, and create bbdb
entries with pgp-mail settings for each key I've got.

Josh> ttyl,

Josh> -- Josh Huber

Jack.
(so whenever someone owns and fixes gpg-ring, I'll be ecstatic)
-- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash

Re: PGP support

[Fabien Penso]

[-- Attachment #1: Type: text/plain, Size: 800 bytes --]


Jack on Fri, 16 Nov 2001 13:38:54 -0800 wrote:

 >>>>>> "Josh" == Josh Huber <huber@alum.wpi.edu> writes:
 > [...]

 Josh> But, this just lets you say whether or not to /always/ sign or
 Josh> encrypt to a certain person.  This works well, but if someone
 Josh> (say, not in your bbdb) sends you encrypted mail, it's nice to
 Josh> return the favor :)

 > What *I*'d like, personally, is to get that gpg-ring.el software
 > working, and then every week traverse my gpg key ring, and create bbdb
 > entries with pgp-mail settings for each key I've got.

Is this the solution ? You would hen get lots of bbdb entries when
everyone will use GPG :p

-- 
Fabien Penso <penso@linuxfr.org> | LinuxFr a toujours besoin de :
http://perso.LinuxFR.org/penso/  | http://linuxfr.org/dons/


[-- Attachment #2: Type: application/pgp-signature, Size: 239 bytes --]

Re: PGP support

[Jack Twilley]

Jack> What *I*'d like, personally, is to get that gpg-ring.el software
Jack> working, and then every week traverse my gpg key ring, and
Jack> create bbdb entries with pgp-mail settings for each key I've
Jack> got.

>>>>> "Fabien" == Fabien Penso <penso@linuxfr.org> writes:

Fabien> Is this the solution ? You would hen get lots of bbdb entries
Fabien> when everyone will use GPG :p

I already have nearly six thousand entries.  It's all good. :-)

Jack.
(bbdb good, gpg good, gnus good, all three triple good)
-- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash

Re: PGP support

[Florian Weimer]

Jack Twilley <jmt+usenet@twilley.org> writes:

> What *I*'d like, personally, is to get that gpg-ring.el software
> working, and then every week traverse my gpg key ring, and create bbdb
> entries with pgp-mail settings for each key I've got.

I'm sorry that I don't finish gpg-ring.el, but there's the risk that
RMS decides to throw the code away and tell someone who has signed
papers to start from scratch.  That's why I'm not working on this
anymore.

(Yes, this has happened to me before.)

Re: PGP support

[Jack Twilley]

>>>>> "Florian" == Florian Weimer <fw@deneb.enyo.de> writes:

Jack> What *I*'d like, personally, is to get that gpg-ring.el software
Jack> working, and then every week traverse my gpg key ring, and
Jack> create bbdb entries with pgp-mail settings for each key I've
Jack> got.

Florian> I'm sorry that I don't finish gpg-ring.el, but there's the
Florian> risk that RMS decides to throw the code away and tell someone
Florian> who has signed papers to start from scratch.  That's why I'm
Florian> not working on this anymore.

RMS can't throw away stuff that's part of Gnus, can he?  I thought
that was up to the maintainers of Gnus -- the only thing RMS could do
AFAIK is just not allow Gnus to be part of the GNU Emacs distribution
until Lars bends to his whim.  Am I misunderstanding?

Besides, if you finish working on it, then dozens of people can
actually use it, copyleft notwithstanding.  Isn't that good?

Florian> (Yes, this has happened to me before.)

I'm sorry that happened.  Bullshit politics suck all around.

Jack.
-- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash

Re: PGP support

[Per Abrahamsen]

Jack Twilley <jmt+usenet@twilley.org> writes:

> RMS can't throw away stuff that's part of Gnus, can he?  

Nobody can "throw away" stuff that has been released as free software.

However, most Gnus users are using the version bundled with Emacs, and
at least some Gnus devlopers (like me) for that reason among several
really don't see the point of working on a Gnus version that cannot be
bundled with Emacs.

So code for which there is no signed papers live in "contrib", and
Gnus proper cannot depend on it, and can't be enabled by default.
Which mean it become a lot less useful than code with papers.  Which
again means that it will be a huge benefit for everyone if someone who
will and can sign papers creates a replacement.

Re: PGP support

[Florian Weimer]

Jack Twilley <jmt+usenet@twilley.org> writes:

> Besides, if you finish working on it, then dozens of people can
> actually use it, copyleft notwithstanding.  Isn't that good?

Working on gpg.el and gpg-ring.el means changing Gnus, too.  I've
signed a copyright assignment for Gnus (probably a mistake in
retrospect, removing my contributions wouldn't have been too hard, and
contrary to my expectations, there was plenty of time before the Emacs
21 release), but gpg.el and gpg-ring.el are copyrighted by my employer
(well, they have the entire usufruct, I only own the moral rights).
This conflict troubles me because I'm not sure whether or not my
employer can claim copyright on my Gnus work if it is related to
gpg.el or gpg-ring.el.

I might obtain the usufruct from my employer, but I think I would need
legal advice to examine the required contract.  (The probability that
the FSF sues me based on a copyright assignment is extremely low, but
with my employer, it is existent---fairness is not an issue for some
decision makers.)

Re: PGP support

[Kai Großjohann]

What's a usufruct?
kai
-- 
I like BOTH kinds of music.

Re: PGP support

[Norbert Koch]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:

> What's a usufruct?

1 definition found

>From Webster's Revised Unabridged Dictionary (1913) :

  Usufruct \U"su*fruct\ (?; 277), n. [L. usufructus, ususfructus,
     usus et fructus; usus use + fructus fruit.] (Law)
     The right of using and enjoying the profits of an estate or
     other thing belonging to another, without impairing the
     substance. --Burrill.

norbert.

Re: PGP support

[Florian Weimer]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:

> What's a usufruct?

It's called 'Nießbrauch' in German (but I'm sure this doesn't help
much ;-).

A usufruct is the right to use a thing for gaining profit of some
kind, but without diminishing the value of the thing (for example,
growing crops on a field in a responsible manner).

I was told that 'usufruct of a copyright' is the correct term, but
some people prefer 'exploitation rights', I think.

German copyright law is a complex field.  Copyright does not only
include exploitation rights, but also moral rights of the author.  In
particular, copyright can only be transferred among heirs, and not
during lifetime of the author.  For computer programs, there are
additional twists.  For example, you can run COTS software legally
without accepting the enclosed EULA in Germany, but you must accept
most parts of the GPL before distributing GPLed code.

In this area, even professionals make mistakes: Recently, the German
branch of IPFI suggested to assign back the exploitation rights for
online transmission to the original authors (invalidating all previous
contracts), and to legalize Napster in its first incarnation (with
songs stored on a central server).  All this was done by accident, I
suppose.

Re: PGP support

[Fabien Penso]

[-- Attachment #1: Type: text/plain, Size: 1324 bytes --]


Simon on Fri, 16 Nov 2001 21:40:54 +0100 wrote:

 > Josh Huber <huber@alum.wpi.edu> writes:
 >> Simon Josefsson <jas@extundo.com> writes:
 >> 
 >>> I don't understand the value of having a header line that says the
 >>> message was originally encrypted, the client already knows this?
 >> 
 >> Right...is there an easy way to access this information (if the
 >> message was signed or encrypted) from a reply buffer?

 > I don't think so... someone please contribute it. :-)

 >> I'm thinking about a way to implement the reply_sign, reply_encrypt,
 >> and reply_signencrypt options of mutt...If you get a message which is
 >> encrypted, and reply to it, Gnus would automatically encrypt the
 >> reply.

 > That would be cool.  Perhaps the idea could be integrated with BBDB as
 > well (or maybe it already is, I think I remember something similar).

There is a bbdb-pgp.el which can does something similar, you can tell in
a bbdb field if you want to crypt or sign the mail, but it doesn't do
what does Mutt as said previously.

I modified bbdb-pgp.el so you can use plain sig, MIME sig, plain crypt
or MIMe crypt. See http://perso.linuxfr.org/penso/lisp/


-- 
Fabien Penso <penso@linuxfr.org> | LinuxFr a toujours besoin de :
http://perso.LinuxFR.org/penso/  | http://linuxfr.org/dons/


[-- Attachment #2: Type: application/pgp-signature, Size: 239 bytes --]

Re: PGP support

[Fabien Penso]

[-- Attachment #1: Type: text/plain, Size: 868 bytes --]


Josh on Fri, 16 Nov 2001 14:37:14 -0500 wrote:

 >> I don't understand the value of having a header line that says the
 >> message was originally encrypted, the client already knows this?

 > Right...is there an easy way to access this information (if the
 > message was signed or encrypted) from a reply buffer?

 > I'm thinking about a way to implement the reply_sign, reply_encrypt,
 > and reply_signencrypt options of mutt...If you get a message which is
 > encrypted, and reply to it, Gnus would automatically encrypt the
 > reply.

This would be _really_ cool. What about setting a var to t when you
verify sign or cryptn and nil if not. Then when you send the reply, it
would look for it. Is this stupid ?

-- 
Fabien Penso <penso@linuxfr.org> | LinuxFr a toujours besoin de :
http://perso.LinuxFR.org/penso/  | http://linuxfr.org/dons/


[-- Attachment #2: Type: application/pgp-signature, Size: 239 bytes --]

Re: PGP support

[Florian Weimer]

Fabien Penso <penso@linuxfr.org> writes:

> This would be _really_ cool. What about setting a var to t when you
> verify sign or cryptn and nil if not. Then when you send the reply, it
> would look for it. Is this stupid ?

Text properties would be even better.  They survive cut and paste.
we're slowly approaching the B2 Orange Book level here. ;-)

Re: PGP support

[news]

Simon Josefsson <jas@extundo.com> writes:

> Werner Koch <wk@gnupg.org> writes:
>
>> On Fri, 16 Nov 2001 13:26:20 -0500, Chris Beggy said:
>>
>>> How about a header field:
>>
>>>   X-Gnus: pgp-verified signature with key 0x454545 valid untrusted
>>
>> BTW, I might have missed it but an important feature would be to add a
>> line like:
>>
>>  X-Gnus-Orig-Encrypted-To: 0x12345678, 0x34567890
>>
>> So that one can see that the message was originally encrypted and even
>> more important to automagically suggest to encrypt any reply.

> I don't understand the value of having a header line that says the
> message was originally encrypted, the client already knows this?  And
> users should probably not trust such headers without the client saying
> it is OK, and if the client can do that, the client could use some
> other (better) way of conveying this information anyway.

I thought you'd like this idea :-)  I know you don't like the
results placed in the message body, where they can be spoofed, as
you showed.

Previous posts in this thread have been discussing the
shortcomings of:

   1. poor visual cue from [hp e] in the modeline to convey
      signature/encryption status (signed,encrypted,valid,trusted?)
   2. poor security of placing encryption status in the message
      body (you demonstrated this...)
   3. introducing Orig-Encrypted-To info, presumably to confirm
      Cc and To fields, and to promote key exchange and
      web-of-trust scoring systems

Using header fields addresses these points because it is a
good place for visual cues when reading mail/news, it can be
turned off if the reader doesn't want to be bothered, and it is
a good place for ephemeral, timestamped information.

Chris

Re: PGP support

[Werner Koch]

On Fri, 16 Nov 2001 20:20:03 +0100, Simon Josefsson said:

> A header to say that replies should be encrypted seems useful, but it
> isn't really Gnus specific, and I would also feel better if requests

No, that is not the point.  Actually I send most mails in the clear
and usually reply in clear.  However some messages I receive are
encrypted and the whole encryption exercise would be rendered
meaningless if I would accidently forget to encrypt - actually it
happens from time to time and the only chance I have is to kill the
uucp job when my fingers were faster than my brain. 

So the default for an reply to an encrypted message must be
'encrypt'.

> I don't understand the value of having a header line that says the
> message was originally encrypted, the client already knows this?  And
> users should probably not trust such headers without the client saying

Header lines have the advantage that you can store the mail
unencrypted for easier grepping but Gnus is still aware that this
message is kind of confidential and will suggest to encrypt a reply.

Ciao,

  Werner

Re: PGP support

[Matt Armstrong]

Simon Josefsson <jas@extundo.com> writes:

> Werner Koch <wk@gnupg.org> writes:
>
>> On Fri, 16 Nov 2001 13:26:20 -0500, Chris Beggy said:
>>
>>> How about a header field:
>>
>>>   X-Gnus: pgp-verified signature with key 0x454545 valid untrusted
>
> I don't understand the value of having a header line that says the
> message was originally encrypted, the client already knows this?
> And users should probably not trust such headers without the client
> saying it is OK, and if the client can do that, the client could use
> some other (better) way of conveying this information anyway.

It would be a display-time only field that Gnus generates -- it
wouldn't reside on disk.

-- 
matt

  Piki, you handsome cat, rest in peace  http://www.lickey.com/piki/

Re: PGP support

[Simon Josefsson]

"Matt Armstrong" <matt+dated+1008537652.ebf9f0@lickey.com> writes:

>>>> How about a header field:
>>>
>>>>   X-Gnus: pgp-verified signature with key 0x454545 valid untrusted
>>
>> I don't understand the value of having a header line that says the
>> message was originally encrypted, the client already knows this?
>> And users should probably not trust such headers without the client
>> saying it is OK, and if the client can do that, the client could use
>> some other (better) way of conveying this information anyway.
>
> It would be a display-time only field that Gnus generates -- it
> wouldn't reside on disk.

Ah. I get it now.  Yes, it would be useful (just as long as Gnus
doesn't care about the header in incoming articles).  I'm not sure it
should be displayed by default, but having more ways of conveying
information about security status would be nice.

Re: PGP support

[Andreas Fuchs]

[-- Attachment #1.1: Type: text/plain, Size: 795 bytes --]

On 2001-11-16, Chris Beggy <news@kippona.com> wrote:
>> Also, the buttons create false security.  I posted a mail to this
>> list that showed PGP buttons that said the message verified OK, but I
>> created them using ^H which Gnus bolidifies.  Whatever the mechanism
>> will be, the article buffer is the wrong place for it.
> 
> How about a header field:
> 
>   X-Gnus: pgp-verified signature with key 0x454545 valid untrusted

<advocate whose="devil">
Again, what about more than one signed part? And the case where only one
(maybe empty) line is signed even with an untrusted key and the rest
isn't?

There should be a simple solution (like banning all html messages and
not inlining images <-;)...
</advocate>

-- 
Andreas Fuchs, <asf@acm.org>, asf@jabber.at, antifuchs

[-- Attachment #2: Type: application/pgp-signature, Size: 231 bytes --]

Re: PGP support

[Florian Weimer]

Simon Josefsson <jas@extundo.com> writes:

> Also, the buttons create false security.  I posted a mail to this list
> that showed PGP buttons that said the message verified OK, but I
> created them using ^H which Gnus bolidifies.  Whatever the mechanism
> will be, the article buffer is the wrong place for it.

mutt includes the current time when presenting output from the OpenPGP
implementation.

Re: PGP support

[Simon Josefsson]

Florian Weimer <fw@deneb.enyo.de> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> Also, the buttons create false security.  I posted a mail to this list
>> that showed PGP buttons that said the message verified OK, but I
>> created them using ^H which Gnus bolidifies.  Whatever the mechanism
>> will be, the article buffer is the wrong place for it.
>
> mutt includes the current time when presenting output from the OpenPGP
> implementation.

I can write a JavaScript that extracts current time and places it in
the article buffer via W3. I can include a image/gif external MIME
part with the current time. Or simply guess when the receiver will see
the mail and add that time to the mail.

I don't think we'll find The One Secure Method of conveying this
information to the user, so multiple methods that each increases the
awareness of the user are nice:

        Indicators in fringe
        MIME buttons (including time?)
        Header line created by Gnus (including time?)
        Modeline icon
        Summary Buffer icons
        ... other ideas

Obtrusive things (MIME buttons, automatic display of the header line,
etc) should probably be turned off by default, Gnus is about reading
what people say not about reading digital signatures.

Re: PGP support

[Florian Weimer]

Simon Josefsson <jas@extundo.com> writes:

>> mutt includes the current time when presenting output from the OpenPGP
>> implementation.
>
> I can write a JavaScript that extracts current time and places it in
> the article buffer via W3.

So disable W3 in conjunction with crypto.  That's the best idea anyway
(W3 can compromise anonymity).

> I can include a image/gif external MIME part with the current
> time.

Same problem, same solution.

> Or simply guess when the receiver will see the mail and add that
> time to the mail.

'g' doesn't change the timestamp, then. ;-)

Well, I forgot that Gnus isn't just a simple mail or news reader.  For
mutt, doing these things is much easier.

Re: PGP support

[Per Abrahamsen]

Simon Josefsson <jas@extundo.com> writes:

> There is a `p' in the Article buffer's modeline. 

Apropos, we should make these have tool tips and be mouse sensitive.
The following hack does the first, but not the last.

Anyone who can see what is wrong?

PS: The code below creates the string and properties each time
`gnus-article-wash-status' is called, they should probably only be
created at start.

(defconst gnus-article-wash-status-strings
  '((cite "c" "Citation toggles" "No citation toggles" 
	  gnus-article-hide-citation)))

(defun gnus-gnus-article-wash-status-entry (key value)
  (let* ((entry (assoc key gnus-article-wash-status-strings))
	 (string (if value (nth 1 entry) (copy-seq " ")))
	 (map (make-sparse-keymap)))
    (define-key map [ mouse-2 ] (nth 4 entry))
    (put-text-property 0 1 'help-echo (if value (nth 2 entry) (nth 3 entry))
		       string)
    (put-text-property 0 1 'local-map map string)
    string))

(defun gnus-article-wash-status ()
  "Return a string which display status of article washing."
  (save-excursion
    (set-buffer gnus-article-buffer)
    (let ((cite (memq 'cite gnus-article-wash-types))
	  (headers (memq 'headers gnus-article-wash-types))
	  (boring (memq 'boring-headers gnus-article-wash-types))
	  (pgp (memq 'pgp gnus-article-wash-types))
	  (pem (memq 'pem gnus-article-wash-types))
	  (signed (memq 'signed gnus-article-wash-types))
	  (encrypted (memq 'encrypted gnus-article-wash-types))
	  (signature (memq 'signature gnus-article-wash-types))
	  (overstrike (memq 'overstrike gnus-article-wash-types))
	  (emphasis (memq 'emphasis gnus-article-wash-types)))
      (concat (gnus-gnus-article-wash-status-entry 'cite cite)
	      (format "%c%c%c%c%c"
		      (if (or headers boring) ?h ? )
		      (if (or pgp pem signed encrypted) ?p ? )
		      (if signature ?s ? )
		      (if overstrike ?o ? )
		      (if emphasis ?e ? ))))))

Re: PGP support

[Per Abrahamsen]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> Apropos, we should make these have tool tips and be mouse sensitive.
> The following hack does the first, but not the last.

I have commited the tool tips code for the washing status string, but
not the mouse sensivity code.

Re: PGP support

[Jack Twilley]

I'm reading your original message with CVS up-to-the-minute Oort, and
I don't see any 'p' in the Article buffer's modeline.

The buttons do kinda sorta work, though.

Jack.
-- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash

How to minimize buttons? (was Re: PGP support)

[Jack Twilley]

>>>>> "Simon" == Simon Josefsson <jas@extundo.com> writes:

[...]

Simon> There is a `p' in the Article buffer's modeline.  Using the
Simon> fringe for this is one idea.  You can also get big buttons if
Simon> you like to clutter up your article buffer:

Simon> (setq gnus-buttonized-mime-types '("multipart/encrypted"
Simon> "multipart/signed"))

One side question:  how do I minimize the button after clicking on it?
Clicking again doesn't do anything useful.  Suggestions?

Jack.
(who likes buttons, yes he does.)
-- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash

Re: How to minimize buttons? (was Re: PGP support)

[Andreas Fuchs]

[-- Attachment #1.1: Type: text/plain, Size: 514 bytes --]

Today, Jack Twilley <jmt+usenet@twilley.org> wrote:
> Simon> (setq gnus-buttonized-mime-types '("multipart/encrypted"
> Simon> "multipart/signed"))
> 
> One side question:  how do I minimize the button after clicking on it?
> Clicking again doesn't do anything useful.  Suggestions?

Hitting the "g" key (gnus-summary-show-article) works fine, although it
minimizes all mime content, and it takes some time to re-evaluate the
pgp content.


-- 
Andreas Fuchs, <asf@acm.org>, asf@jabber.at, antifuchs

[-- Attachment #2: Type: application/pgp-signature, Size: 231 bytes --]

Re: How to minimize buttons? (was Re: PGP support)

[Jack Twilley]

>>>>> "Andreas" == Andreas Fuchs <asf@void.at> writes:

Jack> One side question: how do I minimize the button after clicking
Jack> on it?  Clicking again doesn't do anything useful.  Suggestions?

Andreas> Hitting the "g" key (gnus-summary-show-article) works fine,
Andreas> although it minimizes all mime content, and it takes some
Andreas> time to re-evaluate the pgp content.

Okay, that's not a good answer.  How tough would it be to make the
button a toggle, such that clicking the button again would return it
to its previous state?

Andreas> -- Andreas Fuchs, <asf@acm.org>, asf@jabber.at, antifuchs

Jack.
(wishing he were just a tad more skilled at this elisp stuff.)
-- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash

Re: PGP support

[Sascha Lüdecke]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1



Josh Huber <huber@alum.wpi.edu> writes:

> Per Abrahamsen <abraham@dina.kvl.dk> writes:
>
>> Actually, I can't.  Probably because of 
>>
>>           X-MIME-Autoconverted: from quoted-printable to 8bit by
>>           sheridan.dina.kvl.dk id WAA23581
>>
>> Would such convertions affect PGP?
>
> Well, did the message verify for you?

This message is in the 'old'format.  You should be able to verify it.

>
> I suspect that the message is signed first, and then encoded QP,
> which would explain the QP encoding of the PGP headers.

Thats exacly what mml1991 does.  Since the message body can contain
non 7-bit characters, one has to protect them from certain MTAs.
mml2015, which I took the code from, does that by QP encoding the
message body and signing afterwards.  This works fine, since the body
is a separate MIME part.

As far as I can guess/see there are two options with old PGP style:

   - using ASCII armouring as protection, which renders the messages
     unreadable without PGP

   - protecting the 8bit characters after signing.


RFC 2045 (Multipurpose Internet Mail Extensions, Part one) states on
page four (ASCII version), talking about MIME headers:

    ...

    (3)   A Content-Transfer-Encoding header field, which can be
          used to specify both the encoding transformation that
          was applied to the body and the domain of the result.
          Encoding transformations other than the identity
          transformation are usually applied to data in order to
          allow it to pass through mail transport mechanisms
          which may have data or character set limitations.

    ...




RFC 2049 (Multipurpose Internet Mail Extensions, Part five) states:

   A mail user agent that is MIME-conformant MUST:

    (1)   Always generate a "MIME-Version: 1.0" header field in
          any message it creates.

    (2)   Recognize the Content-Transfer-Encoding header field
          and decode all received data encoded by either quoted-
          printable or base64 implementations.  The identity
          transformations 7bit, 8bit, and binary must also be
          recognized.

          ...

Thus I assumed the QP transfer encoding is safe.  Unless your
recipient doesn't have a MIME capable mail tool.

> FWIW mutt does not do this.  Does Outhouse Express support PGP
> signed messages that are QP encoded like this?

I haven't tested how it works with other mail tools beside Gnus yet.
Unfortunately the reason for hacking mml1991 together, an OutOfLuck
user who coudln't make use of PGP/MIME, refused to further test it --
installation of plugins seemed to be too compilcated.

Best regards,
Sascha

- -- 
Nothing is free in this world, so go get some nothing before they
charge for it!
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iEYEARECAAYFAjvzEdYACgkQqI/9z8xhHua0LQCeMSLMQ3di9Y55F5RXhCYR1cZr
PmkAn3Izr5mt9Mu/EEBbhTNR2/mAwjxt
=dHDI
-----END PGP SIGNATURE-----

Re: PGP support

[Florian Weimer]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

>           X-MIME-Autoconverted: from quoted-printable to 8bit by sheridan.dina.kvl.dk id WAA23581
>
> Would such convertions affect PGP?

Yes, they can invalidate the signture.  That's why you should use the
RFC 3156 method.