Re: [Lynx-dev] mozilla's take on arc4random

[David Woolley]

Thorsten Glaser wrote:
> Michael S Gilbert dixit:
>
> > the mozilla developers are working this predictable PRNG issue, and
> > they have indicated that arc4random would not be sufficently secure

Note that, for the traditional, mathematical, uses of random numbers 
generators, predictability is desirable, as it allows detailed results 
to be reproduced.

> No, it’s simply not needed. For Mozilla, they have their NSS stuff,
> which is used by the browser internally, and the javascript random
> functions have different requirements. Such functions should not
> deliver a constant random stream, but instead use a random seed –
> whether this comes from NSS or arc4random is irrelevant – for a

As I understand the requirement that started this thread, what is 
required is that it not be possible to deduce the internal state of the 
mechanism that generates MIME delimiters.  That can be achieved either 
by having no internal state, or by using cryptographically strong 
methods (true random numbers, or a truly random seed and an algorithm 
for which it is computationally infeasible to deduce later numbers in 
the sequence from earlier ones).

The MIME requirements can, I believe, be met by having no internal 
state, and simply searching for a string that does not appear in any of 
the parts.

> function like an LFSR. (They could expose the NSS (P)RNG, but there
> is no language standard for it.)
>
> This is irrelevant for Lynx, as it does not do Javascript.

The original requirement is not irrelevant to Lynx, as Lynx can generate 
MIME multipart form submissions.
>

--
David Woolley
Emails are not formal business letters, whatever businesses may want.
RFC1855 says there should be an address here, but, in a world of spam,
that is no longer good advice, as archive address hiding may not work.