Re: [ANN] Working SSL In Squeak

[Bill Schwab]

Hans-Martin, Ron,

======================================
> I did not do any comparison because I did not yet download the DES
> plugin - just wanted to make this thing work in the unit tests before
> going to bed :-) I just added a very thin layer (mostly converting
from
> ByteArrays to LargeIntegers and vice versa) to use my old code mostly
> unchanged. The performance can probably be improved a bit by changing
> that old code to directly work with ByteArrays.
>  
Ok, I did some measurements. Results are pretty disappointing.
Squeak on my AMD 3800+ processor encrypts 32 KBytes using 3DES CBC mode
in about 8 seconds - unusable for anything but a proof-of-concept.
Therefore I think that DES (and maybe some other small cryptographic
primitives) should be part of the base VM.
======================================

Just to clarify, is this a discussion regarding where a plugin or DLL
belongs, or regarding whether we need one at all?

Bill






Wilhelm K. Schwab, Ph.D.
University of Florida
Department of Anesthesiology
PO Box 100254
Gainesville, FL 32610-0254

Email: bills@...
Tel: [phone]
FAX: [phone]

_______________________________________________
Cryptography mailing list
Cryptography@...
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/cryptography

[Hans-Martin Mosner]

Bill Schwab schrieb:
>
> Just to clarify, is this a discussion regarding where a plugin or DLL
> belongs, or regarding whether we need one at all?
>  
I think both:
1. We need a primitive - the Smalltalk code is too slow to be usable.
2. It should be in the base VM because it should be easy to access,
without downloading, compiling, installing etc. That was the point in
making the Smalltalk implementation compatible to the plugin, so that we
could use it in cases where the plugin is not available. But if the
plugin is built-in, that point becomes moot.

Cheers,
Hans-Martin
_______________________________________________
Cryptography mailing list
Cryptography@...
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/cryptography

[Bill Schwab]

Hans-Martin,

==========================
> Just to clarify, is this a discussion regarding where a plugin or DLL
> belongs, or regarding whether we need one at all?
>  
I think both:
==========================

Good thing I asked :)

==========================
1. We need a primitive - the Smalltalk code is too slow to be usable.
==========================

That is as I would expect.  Should I be surprised by the results?


==========================
2. It should be in the base VM because it should be easy to access,
without downloading, compiling, installing etc. That was the point in
making the Smalltalk implementation compatible to the plugin, so that we
could use it in cases where the plugin is not available. But if the
plugin is built-in, that point becomes moot.
==========================

The down side is that it becomes (potentially) more difficult to upgrade
it.  However, one can misconfigure systems of almost any design.  It
might be a good idea to build a version number check into it so that
newer images will refuse to run against sufficiently dated VM code, or
something like that.  It would have to blow up in a way that: (1) won't
crash Squeak[*]; (2) does not silently send clear text.

Bill

[*] I mention it because my OpenSSL package for Dolphin crashes the
image quite badly when the DLL is not present.  I tolerate it, but would
not want to foist such junk on others :)






Wilhelm K. Schwab, Ph.D.
University of Florida
Department of Anesthesiology
PO Box 100254
Gainesville, FL 32610-0254

Email: bills@...
Tel: [phone]
FAX: [phone]

_______________________________________________
Cryptography mailing list
Cryptography@...
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/cryptography