testing quality of a /dev/random

[Sam Roberts]

I've ported the Linux random device (random.c) to QNX4. I'll
post the driver announcement after I've used it a bit more than
a day ;-).

I'm wondering at how to evaluate the quality of randomness
it generates. The Linux code itself I had to make only a few
trivial changes to (mostly ifdefing out the o/s interaction code
that's not used, and the format of inline assembler to read the
Pentium time-stamp counter), so I assume all the math is still
solid, but is there any way to analyze the output to determine
it's quality?

I can test it for function easily, quality I don't know how, so I'm
not sure whether or not to call it "Alpha" or "Beta", or how to
*prove* it's really working.

Sam

p.s. How should it be distributed? If I just stick it on my web
page no one will find it. Could it be part of a contrib directory,
or a link, or ...?

--
Sam Roberts, sam at cogent dot ca, www.cogent.ca