From: Nobody@eniac.ac.siue.edu (Anonymous)
Date: Sun, 5 Feb 95 10:03:55 PST
To: cypherpunks@toad.com
Subject: BROKEN REMAILERS
Message-ID: <199502051800.NAA02153@bb.hks.net>
MIME-Version: 1.0
Content-Type: text/plain


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

Here's a bug report on some remailers that won't work in a nested chain
when it is encrypted.
First, a look at how a chain is supposed to work.  Three good ones, chained
like:
1) Tower->Alumni->Homer. OK
2) Homer->Tower->Alumni. OK
3) Homer->Alumni->Tower. OK

This was in a test of the new Tower remailer. In other words, 
it doesn't matter if Tower is at the beginning, middle or end of a chain.
The message will reach its intended destination. (In fact, quite fast).

But some other remailers fuck up when they become part of a nested, encrypted
chain.  One of those? The Syrinx remailer.  Here are the test results:
1) Syrinx->alumni->Homer. NO
2) Homer->syrinx->alumni. NO
3) Homer->alumni->syrinx. OK

The third one came back to be in under an hour. No reply from number 1 and 2.
I don't know why it doesn't work but it doesn't! Maybe one explanation is
that as Syrinx gets a messages it sees the :: Encrypted: PGP and then
decodes it using its key (as it should), but then in the decrypted message
it sees another :: Encrypted: PGP and attempts to decode it again using its
key (again), fails (of course) and aborts. This to me is the only explanation
as to this strange behaviour. It is a serious problem and I do think that
it should be brought to the attention of everybody. Now I haven't tested
all of the remailers so I don't really know which one also will fail the
test. In that test I choose Homer and Alumni because I know that these 2
are reliable and don't care in what order they are in a chain, I also always
send the same test message it reads:

This is a test
line1
line2
line3
line4
Sent <TIME and DATE>
to #1->#2->#3

So its easy to do. And when I get a reply I know which one (1, 2 or 3)
worked, since the path is in the message. Now I know that syrinx doesn't
work. So I add it to my list of those remailers that failed the test.

Also on this list are:
* Nately
and 
* Usura
as they both share the same properties as Syrinx: The chained message gets
home ONLY is Syrinx / Usura / Nately is the _last_ link of the chain.
(Nately *used* to be as good as Homer, Tower, Hal and Portal, but not
anymore. Around Christmas, it changed its performance. Maybe the operator
tinkered - don't what what he did, but he made it worse.)

Two other remailers failing our tests, but for other reasons, are:
* Extropia
and 
* Vox.
Neither forwards attached files. In other words, anything outside of the 
original encrypted block gets sent on.  This makes both of them useless
for reply blocks, for instance.


- --
anon
- ---
[This message has been signed by an auto-signing service.  A valid signature
means only that it has been received at the address corresponding to the
signature and forwarded.]

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2
Comment: Gratis auto-signing service

iQBFAwUBLzUSGCoZzwIn1bdtAQH3IgF9FIUMOTd6rsihkaXUCKn4w14qTtOEfjcD
sXFPB/62K6NQZCuHTUPBDa30jh2c8sJY
=6tBU
-----END PGP SIGNATURE-----