1995-02-10 - Re: MIME based remailing commands

Header Data

From: lcottrell@popmail.ucsd.edu (Lance Cottrell)
To: Hal <cypherpunks@toad.com
Message Hash: e035782a13b3c07db229cc92974dd491e6e5ae4f94d3e704954be124f784aee6
Message ID: <ab60b81304021004b962@[137.110.24.250]>
Reply To: N/A
UTC Datetime: 1995-02-10 06:57:19 UTC
Raw Date: Thu, 9 Feb 95 22:57:19 PST

Raw message

From: lcottrell@popmail.ucsd.edu (Lance Cottrell)
Date: Thu, 9 Feb 95 22:57:19 PST
To: Hal <cypherpunks@toad.com
Subject: Re: MIME based remailing commands
Message-ID: <ab60b81304021004b962@[137.110.24.250]>
MIME-Version: 1.0
Content-Type: text/plain


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

<SNIP my comments>
>Ah, I see how you are doing it.  Having re-read your docs, I gather
>that when un-armored the file is in an encrypted binary format, and
>when decrypted at least the non-header portion of the file is still
>binary?  I think this is a good way to do it; it addresses the point
>Eric made recently about size expansion when an armored file is
>encrypted at each step.
>
That is about right. Inside the armor is a bunch of binary data. 20 header
blocks, and the message block. The first message block is a PGP encrypted
to the remailer, and contains the next destination. The other 19 headers,
and the message block are IDEA encrypted with a key and IV in the first
header. When decrypted, the second header is seen to be PGP encrypted to
the next remailer, and the rest is encrypted with the key in that header.
The remailer Shifts up all the headers, sticks junk in the 20th header, and
rearmors the whole thing.

This is rather more active manipulation than the current remailers.


>The one thing I would mention is that "::" was not originally intended
>as an indication that the message was to be remailed.  Rather, this was
>simply a "header pasting token" which could be used to move a few lines
>from the body up into the header for those people who can't set header
>fields on outgoing mail.  Then the presence of "Anon-To:" or whatever
>in the header is what actually causes the action.  So you don't need to
>use "::", you can just set your headers directly and get the same
>effect.  (This is not to say you need to do it like this, just that
>that is how the original design that Eric created worked.)
>

I realize that. It was easy to implement, since I already had to check for
:: to be compatible with type 1 messages.

>If you did want to follow this model, you could think about using a
>MIME header to indicate the type of the message contents rather than
>the "::".  Another alternative would be to use a different special
>field in the mail header, like perhaps your "Remailer-Type: 2.0", but
>I'm not sure that a new top-level header field is the right place for
>this.  It looks to me like most of the standard headers deal more with
>moving the message around rather than with telling what would be done
>with it on receipt.  It's kind of a fine line but it looks to me like
>more of a job for a MIME content type since that is really what it is
>for.  You could use something like:
>
>MIME-Version: 1.0
>Content-Type: application/remail; version="2.0"
>
>or
>
>MIME-Version: 1.0
>Content-Type: application/remail-mark-2
>
>Then the rest of the message could look just as you have it.  Or, to use
>a little more of the existing standard, you could add:
>
>Content-Transfer-Encoding: base64
>
>and take out your BEGIN and END lines since it looks like you are using
>base64, although the augmented kind that PGP uses with the CRC at the
>end; you'd have to lose the CRC in that case.  (I wonder if PGP will do
>that in the MIME-PGP integration draft that is supposedly being worked
>on.)
>
>One question is, how do you actually send your messages in the
>mixmaster client and servers?  Do you go directly to sendmail, or do
>you use a user agent like /bin/mail?  If the former then it doesn't
>seem like it would be too hard to add these header fields.  On the
>receiving end then hopefully also it would not be much harder to match
>the Content-Type: string than the one you are using.
>
>The advantage, again, is that to a considerable extent this kind of
>application is exactly what MIME was planning for with the "application"
>content-type.  This lets you mark the contents of the message in a
>standard way.  And you are already using something very close to the
>base64 encoding that MIME specifies.  So this does seem like a good
>opportunity to go with the internet mainstream by following this
>standard.  If this seems like something you want to do I'm sure our MIME
>experts here can tell how to define a new content type.
>
>Hal
>

I agree that changing mixmaster to search for a MIME line, rather than the
:: would be no effort at all.

I use sendmail for the client and the remailer (actually they are the same
code), so adding new headers is easy. Unfortunately I am having some
problems with sendmail. Sendmail is marking the messages as "Apperently
To:" rather than "To:". Any idea what might cause that?

I am not sure I want to give up the CRC on the armor (esp. since I am using
the PGP tools routines and not my own), but I would be interested in
defining a new MIME content type. So, experts, what all is involved in
that, and is there any problem with starting to use a MIME type before it
is official?

-----BEGIN PGP SIGNATURE-----
Version: 2.6

iQCVAwUBLzsIElVkk3dax7hlAQFlmgP9FNxSrabWU7CyiRu1Kv28Lz2htuukq1Ul
ANG4y/arJ9gseBvOxDKVDLVKWN1c2XyOKrEbUTVXxHDhk/WwhfpvJE/UrVWQuFEP
eaFZB4G1xBvyjkx+DoQxsLVTEVcF1V54Mo5tfARBGCqf7aHqQGoLiRu8kR6i04fj
nPgOrMeE5Bk=
=vg7R
-----END PGP SIGNATURE-----

--------------------------------------------------
Lance Cottrell  who does not speak for CASS/UCSD
loki@nately.ucsd.edu
PGP 2.6 key available by finger or server. Encrypted mail welcome.
Home page http://nately.ucsd.edu/~loki/
Check out my essay on the next generation remailer Mixmaster on the WWW page.
For anon remailer info, mail remailer@nately.ucsd.edu Subject: remailer-help

"Love is a snowmobile racing across the tundra.  Suddenly
it flips over, pinning you underneath.  At night the ice
weasels come."
                        --Nietzsche







Thread