[Mailman-Developers] dkim-signature headers

[Barry Warsaw]

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

I'm not sure what the right answer is just yet, but I'll offer some  
of my thoughts FWIW.

I think the fundamental question is whether the mailing list is the  
originator of the messages its members receive or whether the  
original author is.  This question has come up in other contexts  
before, and I don't think it's ever been answered satisfactorily.   A  
quick search through DKIM archives seems to indicate that this  
question has come up there too, and I think answering it will help us  
understand what we should be doing here.

On Feb 1, 2007, at 3:00 PM, Mark Sapiro wrote:

> Consider that while Mailman doesn't do all of these things to every
> message, it can do any of the following:
[munge the original message]

 From the DKIM FAQ:

     - What is the purpose of DKIM?

     DKIM lets an organization take responsibility for a message.   
The organization taking responsibility is a handler of the
     message, either as its originator or as an intermediary. Their  
reputation is the basis for evaluating whether to trust the
     message for delivery.

I think you can make a legitimate case that Mailman is the originator  
of messages its members receive.  The message is certainly different  
than the one the original poster sent to the list, and Mailman is  
clearly an intermediary.  Perhaps the message has only been munged in  
very trivial ways, but it's also possible to munge it in ways that  
could potentially be viewed as spammy.  For example, what if a site  
decides to put some advertisements in the footer?

If you take this view then it seems reasonable to say that it is the  
mailing list's system that "take[s] responsibility for a message."   
Sure, the mailing list system could verify the DKIM headers on the  
message it receives, but ultimate, it is up to the mailing list  
system to decide whether that message (or some derivative of that  
message) gets transmitted to its recipients.

Or looked at another way, if I send a message through a mailing list,  
I wouldn't want to vouch for whatever comes out the other end because  
I don't know what they're going to do to my original content.  Maybe  
then, it's correct for the DKIM signature on the copied message to be  
broken because what recipients got was /not/ the message I sent, and  
I don't know how it was munged.  But that view implies that I am the  
originator of the recipient's message.  I am, sort of, but also sort  
of not.

I'm not convinced that DKIM is really designed to handle the mailing  
list use case.  It seems to me that it was designed to handle point- 
to-point messages, not messages that flow through an intermediary,  
because it's not an enveloping system.  Contrast that with S/MIME or  
OpenPGP.  I can sign the message I send from my mailer and that could  
be preserved through the transformations that Mailman performs, with  
Mailman wrapping my original in its own signature if it wanted to.

Practically speaking, if we can't come up with a consensus on the  
interpretation of which "organization [should] take responsibility  
for" the actual message that recipients receive, then what would be  
the right thing to do?  (Note that this answer is different depending  
on whether we're talking about Mailman 2.1 or some future version.)

When this came up before I statement my preference not to make a  
"strip DKIM headers" selectable by the list owner.  I still prefer  
this for Mailman 2.1 because doing so would clearly be a new  
feature.  Maybe a future version could treat the DKIM header the way  
it treats the RFC 2369 headers, with a separate selector for List- 
Post.  Ideally, we'd have a more general way to decide which headers  
get cleansed and which new ones get added.  But that's for the future.

One elaboration you /might/ be able to get away with in Mailman 2.1  
occurs to me as I look at Mark's list:

>  - Add text to the beginning of the message body (msg_header)
>  - Add text to the end of the message body (msg_footer)
>  - Remove text from the beginning of the message body (Approved: line)
>  - Add additional MIME parts to a multipart message (msg_header,
> msg_footer)
>  - Convert a single part message to multipart in order to add
> msg_header/msg_footer
>  - Remove parts from a multipart message (content filtering)
>  - Convert an HTML part to plain text (content filtering)
>  - Decode a base64 or quoted-printable encoded part and perhaps
> re-encode it with a different encoding.
>  - Change or delete various headers including Subject:, To:, From:
>  - Replace some MIME parts with URLs of where they were stored and
> flatten the entire message into a single plain text message  
> (scrubber).
>  - Probably other things I'm overlooking.

If you could identify the message transformations that break the  
signature, then you could remove the signature.  If the signature of  
the outgoing message were still valid because Mailman didn't touch  
any part of the message affecting the signature, then you could keep  
it.  The implementation of this would be fairly simple; the hard part  
is writing the code to verify a DKIM signature and parse the  
selectors (IIUC the specification) to figure out which of the above  
transformations would break the signature.  That might be enough to  
not do it in Mailman 2.1.

I'm not sure how much I like that anyway, so comments are definitely  
welcome.  After mulling over this post for an hour ;) I'm starting to  
believe that it's the mailing list system that needs to vouch for the  
messages its recipients receive.  Of course, it could be Mailman  
doing the DKIM signing, or it could be Mailman's outgoing MTA, etc.   
But, ISTM Mailman is ultimately deciding what goes into the list  
copy, so it is responsible for it.
- -Barry

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (Darwin)

iEYEARECAAYFAkXCvIgACgkQ2YZpQepbvXFnVgCfeUkfQ0+h/bBAKiwznDTdrHJ6
7V0An2O9TcUYBJYlFhYFpLtYUUGalabq
=yCgT
-----END PGP SIGNATURE-----