Re: [Openpgp-dt] first set of PRs created from step-by-step branch

[Daniel Kahn Gillmor <[e-mail]>]

On Sat 2021-07-24 08:41:04 -0400, Paul Wouters wrote:
> On Jul 23, 2021, at 23:11, Daniel Kahn Gillmor <[e-mail]> wrote:
>> If you'd like, i can try to propose an alternate, short series that just
>> includes the current AEAD specification plus the AEAD-specific set of
>> test vectors.  If we can get consensus on that, then moving through the
>> v5 SKESK and #253 S2K should be simpler, bite-size pieces by comparison.
>
> That would be good.

I've done that over here:

https://gitlab.com/openpgp-wg/rfc4880bis/-/merge_requests/59

Hopefully commentary on that MR should be able to be more
tightly-focused than dragging in all the SKESK, preferences, etc stuff.

>> If it would be useful, i can also propose a variant of AEAD with many
>> fewer knobs as a comparison.  Each knob can cause negotiation failures
>> and deprecation difficulties, and i'm worried that the current amount of
>> flexibility could saddle future implementations with a lot of technical
>> debt.
>
> How would that interfere with current deployment ?

I think we could win this simplicity and avoid interfering with any
pre-existing deployment by just picking a different "version number" and
documenting that, leaving documentation of "version 1" for some future
document.

> I would like this to be a separate issue on the tracker.

I agree this would be an entirely separate issue, and one mutually
exclusive with the AEAD stuff we're trying to pull in from the old
draft-10.  If anyone else on the design team wants to propose text, i
would be happy to review it.  or, i could take a stab at writing an
example text if y'all want.

If the DT deems this idea too crazy and off-the-wall and disruptive to
current deployments to consider, i'm also happy to drop it.  i'm
proposing it because having a less-flexible option might be a useful
comparison.

      --dkg