Re: [openpgp] a new draft overlapping the WG draft

[Tom <[e-mail]>]

Given how this conversation above has turned out, I think Neal's comment is
on point.

Cheers,
Tom

On Fri, Oct 7, 2022 at 9:41 AM Neal H. Walfield <[e-mail]> wrote:

> Hi,
>
> On Mon, 26 Sep 2022 22:25:13 +0200,
> Stephen Farrell wrote:
> > The top level issue for the WG at this point is whether we
> > (a) change course to work more in line with [1], or,
> > (b) continue with the close-to-done WG draft [2], or,
> > (c) try to tread a line in between.
> >
> > So, can WG participants please take a look at the diff [3]
> > between [1] and [2], then express your opinions as to which
> > of the actions a, b or c you think the WG should take. (In
> > the case of option (c), please also provide further detail.)
>
> I'm strongly in favor of continuing with the crypto refresh.  First, I
> think it is the technically better standard.  Second, I think
> draft-koch-openpgp-2015-rfc4880bis does not address a number of
> important technical issues that need to be addressed (no argon, the
> kopenpgp attacks, etc.).
>
> My reading of Werner's recent message:
>
>   Let us stop the whole mess now.  Please consider my current draft as a
>   description of what GnuPG has implemented and what has been widely
>   deployed to the masses (Gpg4win) and for critical infrastructures.
>
>
> https://mailarchive.ietf.org/arch/msg/openpgp/WqM7VPXsqDJ9mLnvQQ4mBf5rYj0
>
> is that he does not want to collaborate.  Based on this:
>
>   - I think it makes sense to review any additions in
>     draft-koch-openpgp-2015-rfc4880bis, and adopt those whose added
>     value justify any added complexity.
>
>   - At the same I'd like the DT to review concessions that were made
>     to Werner.  In particular, my understanding is that keeping AEAD's
>     flexible chunk size parameter was a compromise with Werner.  I'd
>     like to repropose removing the chunk size parameter, and fixing it
>     to a single value.
>
>
> https://datatracker.ietf.org/doc/html/draft-ietf-openpgp-crypto-refresh-06#section-5.13.2
>
>   - To decrease the burden on distinguishing GnuPG messages from
>     OpenPGP messages, we should probably bump the version of any
>     conflicting versions or identifiers (packet versions, s2k
>     identifier assignment, etc.).
>
>     I realize that there will be quite a bit of work to regenerate the
>     test vectors, but I think that that work will be better over all.
>     Implementations can more easily produce better error messages
>     (e.g., by returning "unsupported message format" instead of
>     "malformed message").  It makes it easier for implementations to
>     support both the OpenPGP message format and the GnuPG message
>     format.  And, it potentially avoids or at least complicates cross
>     grade attacks in which an attacker changes a message from one
>     format to the other to exploit a bug, as was done in the efail
>     attack, which downgraded SEIP packets to SE packets.
>
>
> Finally, I'd like to thank the design team for their hard work and
> their attempts to be inclusive.  I was repeatedly impressed at how
> constructively Paul, Stephen, and DKG, in particular, have engaged
> with the community.
>
> Thanks,
>
> Neal
>
> _______________________________________________
> openpgp mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/openpgp
>


-- 
Tom at FlowCrypt