From: Timothy Newsham <newsham@wiliki.eng.hawaii.edu>
Date: Wed, 22 Sep 93 13:50:21 PDT
To: derek@cs.wisc.edu (Derek Zahn)
Subject: Re: Why RSA?
In-Reply-To: <9309221814.AA23907@balder.cs.wisc.edu>
Message-ID: <9309222048.AA20361@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


> Regarding the recent proposals for the construction of a toolkit,
> I'm all in favor and would personally welcome the opportunity to
> contribute to such an effort as a hands-on supplement to my
> crypto education.  I have extensive experience with C and C++,
> and am VERY familiar with TCL (pronounced 'tickle', for those
> not in the know).  A good start would be a clear statement of
> purpose.

purpose: to make routines implemented and implemented well within
the PGP program available to programmers.

future purpose: to make a general purpose library of routines
helpful in implementing various crypto systems and protocols.

plan:  Take the PGP source and rip it apart into tiny pieces
and put it back together in an organized way.  Change the
Makefile structure to build various libraries and isolate
the main user interface routines into a seperate group of
files.  Basically a restructuring of PGP that will provide
various intermediate libraries that may be used by other programers
for linking within their program.

*IF* this becomes part of the standard PGP distribution all
non-portable code will be rewritten for various platforms and
you will be able to find a lib for just about any platform
you are coding or porting to.

I havent read through the PGP code myself, but I suggest
at least libraries for the random number routines (including
the system-specific keyboard routines for getting random
seeds), a library for the RSA and IDEA routines, and a library
of the lower-than-RSA math routines.

After this is done then various projects such as a tcl shell
can be written on top of the libraries.

If the library is successful then various additions can be made
to it to make it a true crypto library rather than just a
PGP library.  This could also benefit PGP if it is changed
to allow various crypto systems.

The benefits of this approach are many:  When implementing
PGP front ends you usually dont want the normal PGP front
end at all, and would rather just re-write your own on
top of the PGP code.  When you are coding simple crypto
libraries (like 'link' and 'Circ') it is nice to have a
drop-in cryptosystem library.  The advantages to import-
export are nice as well, you can write code that has no
crypto code in it at all and let the users grab the crypto
library to compile.

> 
> derek
>