Date:        Thu, 11 Sep 2003 22:54:06 +0900 (JST)
    From:        Noriyuki Soda <soda@sra.co.jp>
    Message-ID:  <200309111354.h8BDs5I18131@srapc342.sra.co.jp>

  | No, there is a problem.
  | It's bad thing that libc functions use rand() (or random())
  | internally, because 3rd party applications (especially scientific
  | simulation software) want that their call to rand() returns same
  | random sequence.

Yes, agreed, that's something to be concerned about.

It wasn't the specific functions that I mean though, just anything which
has the kind of "randomness" that rand() produces (which isn't much, but
is plenty good enough for the communications between a stub resolver and
its cache (aka back end resolver)) in almost any situation - and in the
cases where it isn't, something like TSIG is required, not a better random
number generator.

kre