Re: Added hashes.
Available news archives: comp.lang.tcl - comp.lang.python - comp.security.firewalls - sci.crypt - comp.lang.php - comp.lang.javascript
Google
 
Web news.hping.org


sci.crypt archive

Re: Added hashes.

From: Kristian Gjøsteen <kristiag+news@item.ntnu.no>
Date: Mon Dec 12 2005 - 13:30:28 CET

Tino Schwarze <tino.schwarze.0401@tisc.de> wrote:
>I'd not xor the result but keep both results (e.g. append hash 1 to hash
>2). So if somebody is able to produce a plaintext for hash 1 (say MD5),
>chances should be very low that it also matches hash 2 (say SHA1). But I
>would only consider this for long-time archiving.
>
>This should[1] also significantly lower the probability of a hash
>collision.
>
>[1] read "sounds logical to me but I'm not a cryptologist."

This piece of folklore was proven false by "Multicollisions in
Iterated Hash Functions. Application to Cascaded Constructions" by
Antoine Joux.

The paper is very easy to understand and a nice read.

-- 
Kristian Gjøsteen
Received on Fri Dec 23 20:09:38 2005