Redundancy eliminating file systems, breaking MD5, donating money to OSDL

(re: md5 weakness)

The only document I've seen with a rigorous demonstration of the possibility of an md5 collision created it by adding 0 (zero) bytes to an input (so the colliding inputs were not the same size in bytes).

Good luck finding a collision with blocks that are all the same size.

Anyway, hash matching algorithms for variable sized inputs (hashed extents, etc) can probably get an additional several orders of magnitude of safety by using two hashes (md5 and sha1, for example).

What are the chances that the same two different inputs that hash to the same value using one of them collides in the other, too? ("Left as an exercise for the ...")

Regards,

Clayton Weaver
<mailto: [email protected]>

___________________________________________________________
Sign-up for Ads Free at Mail.com
http://promo.mail.com/adsfreejump.htm