[rc5] Security, Java, and Source

Marc Sissom msissom at dnaent.com
Fri Oct 10 15:38:51 EDT 1997


Eric Gindrup wrote:
> 
>      redirected your module retrieval to be from their own machine.
> You>      got their code.  Looked like it was from Bovine, though...  And>      there's no "perfect" signature scheme to avoid this.  Consider
> the>      current effort as sufficient refutation.

Yes there is. Just build a public/private key verification
system into the communications components. Bovine pubs their
key and signs the client modules. The client gets the key
when they sign on, or has it embedded, and checks when it
loads a client core module. No match, no run. The same can
be done when you load an executable from the disk, network,
or whatever.

The same can be done for the work that the client does. You
register with the bovine server when you decide to run a
client. Then you just have the client sign the work with
your private key and let the servers verify it against the
registered info. BFD.

If you want to be really simple, just implement the SSL that
we all know and love for all client/server transactions -
including core module loading.

You can publish the interfaces, let others develop core
components, new/improved communications/UI components as
well as have security. Bovine can manage the projects,
check new components, and provide forums for the discussion
of ideas that are submitted by that vast pool of unused
coders out there, let us all freely read the source and
carry on coding our grubby little fingers off.

Yes, any security system can be broken. That's the point of
the exercise isn't it? The idea is to have sufficient
security without undue loss of freedom. What we have now is
very little security at a cost of considerable loss of freedom.

--
----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list