From: Yukihiro Matsumoto <matz@...>
Date: 2006-11-13T18:27:51+09:00
Subject: Re: External entropy pool for random number generator

Hi,

In message "Re: External entropy pool for random number generator"
    on Mon, 13 Nov 2006 18:09:37 +0900, "Kirill Shutemov" <k.shutemov@gmail.com> writes:
|> It uses /dev/urandom already.
|It uses only for initial random seed and when you call srand(0);

I am not sure I understand.  By your patch, if you specify something
that responds to "read", rand() reads 4 byte integer and call srand()
for each time, right?  In that case, what the difference from reading
from /dev/random directly, if we assume quality of randomness from
/dev/random?  I'm sure it is worth allowing the following

  srand(open("/dev/random"))
  loop do
    p rand(0x1000)
  end

instead of

  open("/dev/random") do |f|
    loop do
      srand(read(4).unpack("L")) rescue nil
      p rand(0x1000)
    end
  end

the former is shorter indeed.  But it seems more difficult to read the
author's intention.  Is this useful in cryptography field?

							matz.