Re: Twofish - details, details...

[tzeruch@xxxxxxxxxx]

On Tue, 12 Jan 1999, Jon Callas wrote:

> At 07:28 PM 1/12/99 -0500, uri wrote:
> 
>    So, Twofish is a logical successor for Blowfish - and let's treat it
>    like that. If people have Blowfish-encrypted stuff already (which I
>    doubt, but you never know) - they can do a one-time move. Better
>    than to carry unnecessary baggage.
> 
> I'm uncomfortable with re-declaring an identifier. It's just not good
> engineering practice to my mind. I think it is much better use another
> identifier. This would only put us up to 10, and that includes four
> identifiers that are presently only reserved. If 2440 were not out, I'd
> have no problem. But it is out there, and it just bugs me to redeclare an
> identifier if it inconveniences even a single user.
> 
> By the bye, an inconvenient page break in 2440 made me say something stupid
> in my last message. The identifier should be 10, not 8. 

10 would be better.  I already have blowfish in my implmentation even if
no one uses it.

Also, as someone else pointed out, we need to have established key lengths
AND I assume that it will be in the ECB mode with the PGP cfb with reset
at 10 bytes like everything else?

Does anyone have an implementation, or at least what calls they are using
if they are using the counterpane reference source?