[Cryptography Team] DES status?

[Ron Teitelbaum]

Wonderful Hans-Martin.  Thank you for working on organizing the class
structure.

I have the same issue with SHA256.  There appears to be a primitive written
for SHA1 for the expand and hash function, but since I have no experience in
creating primitives, I didn't go that route.

I would think that we need two things to answer this question.  First we
need to quantify the benefits of the primitive.  What kind of speed
improvements do we actually get vs. the hiding of the code and making it
harder to distribute, and making it more difficult to support.  There would
have to be major (orders of magnitude) speed improvements to make it worth
it.  Does anyone have primitive writing experience that would like to take
this issue on?

In general I would say that every method that has a primitive call, (unless
it would be totally impossible because of time limitations) should have a
fall back method written.  So yes I would say you should write the DES code.

Ron

-----Original Message-----
From: cryptography-bounces at lists.squeakfoundation.org
[mailto:cryptography-bounces at lists.squeakfoundation.org] On Behalf Of
Hans-Martin Mosner
Sent: Sunday, October 23, 2005 8:11 AM
To: Cryptography Team Development List
Subject: [Cryptography Team] DES status?

Hi folks,
I noticed that Cryptography-DES has no non-primitive version. Since the 
primitive is not in my current VM, I can't test it.
Another issue is that interpreterProxy shows up as Undeclared when I 
load the DES stuff, which probably means that the primitive won't 
compile cleanly - but I have not looked further into this.
Would you mind if I modify the DES code to fall back to the 
non-primitive implementation which I wrote for OpenPGP?
I would have to adapt to the slightly different API anyway (my code used 
8-byte Integers as keys and data blocks, while the Cryptography code 
uses 8-byte ByteArrays which make more sense).

BTW, I've made a first cut creating separate packages 
(Crpytography-Core, Cryptography-ASN1, ...) which seems to work 
reasonably well. Once I've successfully loaded all of them into a base 
image, I'll publish them to our repository. I don't know how MC manages 
prerequisites yet, so anyone loading these would have to take care of 
the load order by himself.
 
Cheers,
Hans-Martin
_______________________________________________
Cryptography mailing list
Cryptography at lists.squeakfoundation.org
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/cryptography