If you are going to update the hash code, you also should parallelize the calculation.
Ie: call the initialization function of hash loop: read a block of the file mix it into each hash in turn end loop: finalize each hash It is silly and, for large DIST files, slow to read(2) the file n times when it is only needed once. To be clear, I'm not suggesting doing each hash in its own thread. Just using their lower-level APIs avoid read(2)ing through the file n times. -JimC -- James Cloos <cl...@jhcloos.com> OpenPGP: 1024D/ED7DAEA6