Bob / Richard

Including the signature value and the ciphertext in the JSON would needlessly 
enlarge the resulting token as it was base64url encoded.

Having the header use a hierarchical structure seems to make it easier to 
support multiple keys, which is needed if you want to sign and encrypt a 
message -- which seems to be a common use case!

-- Dick

On Nov 9, 2012, at 8:29 AM, Bob Wyman <[email protected]> wrote:

> The "val" element in Barnes' proposal would need some parameters (typ, cty, 
> etc.) as well as the content in val:
> 
> {
>   "enc": { /* encryption parameters */ },
>   "sig": { /* signature parameters + value */ },
>   "val": {/* content parameters + plaintext */}
> }
> 
> 
> 
> On Fri, Nov 9, 2012 at 1:58 AM, Richard L. Barnes <[email protected]> wrote:
> Hey Dick,
> 
> To make sure I understand your use case correctly: You want to convey an 
> encrypted object, as well as a signature over the plaintext (without having 
> to encrypt the signature value).  Does that sound accurate?
> 
> It seems like there are three types of information you want in the object: 
> (1) cipher text, (2) encryption parameters, and (3) signature 
> value/parameters.  So why not encapsulate it like that?  To propose a JSON 
> syntax:
> 
> {
>   "enc": { /* encryption parameters */ },
>   "sig": { /* signature parameters + value */ },
>   "val": "/* ciphertext */"
> }
> 
> Obviously:
> -- JWE/JWS could be the special cases where only one of "enc" or "sig" is 
> present
> -- This cleanly supports multiple signatures via multiple "sig" values (e.g., 
> in an array)
> -- This cleanly supports multiple recipients via multiple "enc" values (e.g., 
> in an array)
> -- You could leave one of "enc" or "sig" parameters as flat lists (as in 
> JWS/JWE), but it seems cleaner to have them parallel
> -- For a compact serialization, you would want certain fields to be not 
> double-base64'ed.  We can figure that out later :)
> 
> This seems like kind of an appealing line of reasoning to me.  I would be 
> glad to do some work on figuring out the details.
> 
> --Richard
> 
> 
> 
> 
> On Nov 7, 2012, at 3:23 PM, Dick Hardt <[email protected]> wrote:
> 
> > To enable encrypting and then signing of the same token, we need to specify 
> > the encrypting and signing algorithms separately.
> >
> > Since we are using JSON, how about if we create an encryption object to 
> > contain all the parameters defined in JWE so that there is no overlap in 
> > the JWS namespace.
> >
> > "enc":
> >       {  "alg"
> >       ,  "enc"
> >       ,  "zip"
> >       }
> >
> > _______________________________________________
> > jose mailing list
> > [email protected]
> > https://www.ietf.org/mailman/listinfo/jose
> 
> _______________________________________________
> jose mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/jose
> 

_______________________________________________
jose mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/jose

Reply via email to