RE: Encryption and the processing model

>
> > I'm not sure what the right way to design this would be.
> You could encrypt
> > the entire header element.  Then when it's decrypted,
> whoever decrypted it
> > could see the actor, plus whatever else is in the header.
> Unless we allow
> > nested encryption, which makes my brain hurt even more.
> >
> > I see a rat's nest ahead as we tumble down the slippery slope.
>
> You and your slippery slopes. 8-)
>

I know, I'm on a slippery slope of seeing slippery slopes :-)  I think I
should go snowboarding or something with all the slopes.

> > I really believe we need more use cases on these various
> types.  Oh wait,
> > I'm supposed to write use cases.
>
> Sure, those would help.
>
> I'm thinking that we need to require that the root element of each
> header block be unencrypted.  In other words, the header "name"
> and qualifiers (actor, mustUnderstand) MUST be visible.  But the
> header "value" (block content) should be able to be encrypted.
>
> I think this is the minimum that should be done to ensure that
> encryption doesn't mess with the processing model.

I think I agree with this.  How/where should this be written up?

Cheers,
Dave

Received on Thursday, 10 January 2002 21:14:08 UTC