On Thu, 2007-08-30 at 17:12 -0400, Bob Johnson wrote: > > But an audio CD will run out of data before you fill a 1GB file. Region > encoding tries to get in the way with a DVD. > > For repeated use, I would be inclined to create a DVD full of data from > urandom, then use dd to pull data off of it with various offsets to create > test files. Or you could copy a few music CDs to one big file that you burn > to your DVD to create a collection of semi-random data for test files. I would shake my head ruefully and how you guys are overthinking this, and just use /dev/urandom, because that's what it's there for and it's not like /dev/random stores infinite entropy forever if you just leave it alone. The default size of /dev/random's entropy pool is 512 bytes. It doesn't take long to renew that. Unless you're doing constant ultra-secure cryptography like streaming intelligence intercepts or something, you really shouldn't concern yourself with the effects of pulling /dev/urandom to your heart's content. I guarantee you, there are few purposes short of a one-time-pad generation where a gig of /dev/urandom won't be statistically indistinguishable from random.