note 96475 deleted from function.md5 by danbrown

Note Submitter: Paul Firth ---- Please *stop* using MD5, it has been decertified by NIST for over a decade. The algorithm is broken: Instead of the expected 2^64 hash computations to produce a collision, a collision can now be computed in roughly 2^20 hash computations - which can be performed in *a second* on a desktop PC. A hash collision is bad. Using this method an attacker can forge a perfectly valid MD5 hash of data that you did not expect as the author, potentially injecting this data into your database and costing you a lot of money. The current recommendation is to use one of the SHA-2 functions, as SHA-1 has also had flaws detected it in, which result in less than the predicted number of hash computations before a collision is produced. http://eprint.iacr.org/2008/391 http://csrc.nist.gov/groups/ST/toolkit/secure_hashing.html