Re: lynx-dev lynx2.8.4dev.17

[Doug Kaufman]

On Sun, 11 Feb 2001, David Woolley wrote:

  (quoting me)
> > rand48 family of functions available. I have used this in the SSL
> 
> As pointed out before, these should not be used for cryptographic
> applications as they are invertable.

I am not sure that I was able to follow that discussion. From the
man pages of DJGPP and FreeBSD, it sounds as if random() is better
than rand48(), but it wasn't clear if rand() was actually better or
worse than rand48(). DJGPP has all three functions, but Cygwin is
missing rand48(). I had thought the question of how they were seeded
was more important than the actual algorithm involved. DJGPP doesn't
do multitasking, so you can't run an entropy gathering daemon. I
believe seeding there is usually done by calling srandom() using
some type of time function as argument. I am outside my depths here
and would appreciate some basic guidance about this (or pointers
to references). The rand48() and random() functions are said to be
non-POSIX, non-ANSI. Are you saying that a system that doesn't have
them does not have the ability to do encryption properly?
 
> > the libc code of the Debian distribution. They say that their code is
> 
> Linux libcs are Gnu libc's.  They are strongly GPL compatible (they are
> either GPL or LGPL - I think they are full GPL)!  (You could, always
> use the versions from www.gnu.org.)

I tried that first, but all I could find was glibc. I was unable to
get the rand48() functions from glibc to compile under Cygwin. I had
assumed that the libc code was GPL, but I searched for a while in the
code itself and on the Debian site without being able to find a clear
licensing statement.
                              Doug
__ 
Doug Kaufman
Internet: address@hidden


; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden