On Mon 2015-06-01 13:15:17 -0400, Werner Koch wrote:
On Mon, 1 Jun 2015 18:32, dkg(_at_)fifthhorseman(_dot_)net said:
I'm not convinced that an updated fingerprint requires a v5 packet
format. Can you explain that?
Without that we will have two fingerprints for the same key. That may
lead to confusion.
I think i'm still not convinced by this -- can you give an example of
what kind of confusion you're hoping to avoid?
In the past it was easy to decide what kind of key you have. A v5
packet format also makes it easier to change the implicit available
preferences (e.g. 3DES) because applications supporting a v5 would
also support the new preferences
Users can infer that keys using new algorithms will implicitly have the
new preferences; But we can also indicate the new preferences explicitly
within the old format.
I think the main difference you're offering with a v5 format is the
ability to explicitly disavow the 4880 MTI algorithms. Is that right?
Maybe there are other ways that implementations can explicitly disavow
the 4880 MTI algorithms, that can be done without introducing a v5
format? What's the advantage of introducing a mandatory
incompatibility?
My conclusion from the discussion at the OpenPGP summit was that there
is a need to add something to PGP/MIME. For example to split a long
message so that a (mobile) device does not need to download the entire
message for a summary view. But well, that would affect S/MIME as well
so might be better taken up by the MAIL WG (in case that WG exists).
Yes, i think this kind of thinking and work is good to do, but i don't
think we should focus on it in this WG until we're comfortable with a
4880bis. If folks want to discuss it elsewhere, and it turns out there
are pieces needed in 4880bis to make it work right, then we should
include those pieces in 4880bis. But i do think that touches on S/MIME
as well, so it'd be nice to keep that perspective in the loop.
--dkg
_______________________________________________
openpgp mailing list
openpgp(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/openpgp