Hi, I just finished implementing an Md5 hash function, and compared mine with two other implementations : RFC 1321 (Ron Rivest's implementation, the reference implementation) and GNU's md5sum. The all give the same results on small files, OK. However, I checksumed a tgz-ed gimp distribution (~40 Mb, I believe), and it appears that md5sum gives a different value than the reference implementation and my own. I don't really know how much md5sum is used, but given that I consider GNU tools as some of the best-quality, I thought it would be important to signal that bug, to keep quality where it belongs : GNU. Regards, Carl Seleborg -- ___________________________________________________ PGP public key : <http://perso.wanadoo.fr/carlov/perso/pgp/carl.asc>