By the way, I've never been able to get a clear explanation of *why* pycrypto's random is better than say sysrandom in this case? IIRC sysrandom will default init PRNG using systime and wouldn't exhibit this problem for threads/forks.
By the way, I've never been able to get a clear explanation of *why* pycrypto's random is better than say sysrandom in this case? IIRC sysrandom will default init PRNG using systime and wouldn't exhibit this problem for threads/forks.