From: das@razor.engr.sgi.com (Anil Das)
Date: Fri, 28 Mar 1997 18:24:56 -0800 (PST)
To: aba@dcs.ex.ac.uk>
Subject: Re: [ANNOUNCE] hash cash postage implementation
In-Reply-To: <199703281652.QAA03299@server.test.net>
Message-ID: <9703281824.ZM5275@razor.engr.sgi.com>
MIME-Version: 1.0
Content-Type: text/plain


Adam,

	How does Dr. Bernstein's announcement of finding
a 56 bit collision in md5 using a few hours on a Pentium
affect this scheme? It was not clear from his post whether
he was looking for a collision with a known hash, or just
two different strings with a collision of the given length.

On Mar 28,  4:52pm, Adam Back wrote:
>
> (Also I have not tested my SHA1 implementation on a big endian machine, it
> auto-detects byte endian-ness, theoretically).

	Works fine here. Big endian Mips R10K.

% ./sha1test
test 1

SHA1("abc") =
        a9993e364706816aba3e25717850c26c9cd0d89d test ok

test 2

SHA1("abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq") =
        84983e441c3bd26ebaae4aa1f95129e5e54670f1 test ok

test 3

SHA1("a" x 1,000,000) =
        34aa973cd4c4daa4f61eeb2bdbad27316534016f test ok

% ./hashcash -t -22
speed: 70921 hashes per sec
find: 22 bit partial sha1 collision
estimate: 30 seconds

--
Anil Das