Re: [avr-libc-dev] rand(3) in avr-libc

[E. Weddington]

On 10 Sep 2002 at 9:45, Theodore A. Roth wrote:

> 
> 
> On Tue, 10 Sep 2002, E. Weddington wrote:
> 
> :) On 10 Sep 2002 at 17:57, Joerg Wunsch wrote:
> :)
> :) > > 2. add a (small?) crypto
> :) > > package to take advantage of the new PRNG.
> :) >
> :) > Hmm, that's probably nothing i could do.  There's some crypto
> code one :) > could get, but that's all written with 32-bit integers
> (or more) in :) > mind, and probably way too huge to port to an AVR
> (like openssl or :) > so). :) :) Nothing I could do either. I didn't
> mean to imply "rolling our own" :) crypto package (ref Ted's msg). I
> was thinking more of porting one :) already in use. And perhaps it
> wouldn't have to be "top-of-the-line". :) However.... If there is
> nobody capable / willing to do that and the :) only real use for the
> larger PRNG code *would* be for a crypto :) package, then I don't see
> a particular advantage for adding the :) larger PRNG code. ....Though
> I would argue that the mega128 can hold :) an awful lot of code... :-)
> 
> I think in this case, porting is nearly equivalent to rolling your
> own. The problem as I see it is in making sure that the ported code is
> correct. I'm not trying to completely veto this idea, just playing
> devils advocate. People using crypto want it to work and it's just
> soooo easy for it to "look" like it works and be flawed in some very
> subtle way.
> 
> Ted Roth

Agreed wholeheartedly! Hence, I think the hestitance in any of us to 
tackle it! :-) 

And, so you know, I'm not exactly a strong proponent either of adding 
crypto... Personally, I don't have an application need for it, and I 
doubt I will for the near future. Maybe somebody else will, I don't 
know. I was just trying to add to the debate of whether to expand the 
PRNG module. Besides crypto, is there any other use for the heavier 
PRNG? If there isn't, then it would probably be more of a hassle than 
it's worth. I mean, unless you (Joerg) *really* wanna do it... :-)

Eric