Boost Users :
From: Peter Dimov (pdimov_at_[hidden])
Date: 2006-05-19 11:39:43
Greg Link wrote:
> [...] Re-seeding naively can be
> dangerous, as it's not just entropy you want in a seed - it's things
> like relative primality, etc. As noted, seeding with a clock isn't
> nearly as effective as most think. Process ID's tend to be 'low' (not
> 11+ bits long, at least), as do ThreadID's, and other such things.
> Not nearly as much entropy as many think.
Well, that's why one would typically combine as many entropy sources as
possible using a hash function (MD5, SHA, even boost::hash if 32 bits are
For more demanding uses there's always CryptGenRandom, or RtlGenRandom on
XP+. On VC 8.0 there's also rand_s.
Boost-users list run by williamkempf at hotmail.com, kalb at libertysoft.com, bjorn.karlsson at readsoft.com, gregod at cs.rpi.edu, wekempf at cox.net