On Thu, Jun 13, 2024 at 08:46:51AM +0100, Neil Madden wrote:
> Hi all,
> 
> We appear to have yet another long WG discussion going on about how to
> try to squeeze the ground meat of HPKE into the intestinal lining of
> JOSE. I know that I at least don’t have the time to follow the
> minutiae of these threads. At some point should we ask if this is all
> worth it? My takeaway is that HPKE is at best an awkward fit for
> JOSE. 

I agree it is at best an awkward fit: Indirect HPKE is quite simple
(still having some pitfalls), but direct HPKE is definitely not,
requiring extending very core parts of JWE.

Especially so if headers need to be supported.


> And if we do finally manage to make the HPKE-JOSE sausage, what have
> we really gained? As far as I can tell the only real advantage is that
> we might eventually get a single ML-KEM/hybrid post-quantum encryption
> scheme. 

As of currently, HPKE has no significant advantages over what presently
exists in JOSE (KEM48 can not be used for spec stability reasons).

In the future, HPKE might gain post-quantum or hybrid KEMs that can be
incorporated to HPKE-JOSE.

However, direct KEM support, capable of using both PQ and hybrid KEMs,
would be a simple thing to add to JOSE.


> But with encapsulated keys that are >= 1KB in size and so totally
> unsuitable for most scenarios that JOSE is used for today, where size
> is extremely important. 

There are sites that use >1kB headers. I had to patch a reverse proxy
to increase the header size limit from 1kB to handle some callbacks.




-Ilari

_______________________________________________
jose mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to