Re: [AVTCORE] [Cryptex] PR for recommending Cryptex over RFC 6904

How about this:

If one of the peers has advertised both the ability to receive cryptex and
the ability to receive header extensions encrypted as per {{RFC6904}} in
the SDP exchange, it is RECOMMENDED for the other peer to use Cryptex over
{{RFC6904}} when sending RTP packets so all the header extensions and CSRCS
are encrypted unless there is a compelling reason to use {{RFC6904}} (e.g.
need for some header extensions to be sent in the clear so that so they are
processable by RTP middleboxes) in which case, it SHOULD use {{RFC6904}}
instead.

I don't think we should add the note about checking that the header
extensions can be actually sent in clear, as we have put that reason as an
example.

What do you think?
Sergio

On Tue, Aug 2, 2022 at 6:30 PM Bernard Aboba <bernard.aboba@gmail.com>
wrote:

> The wording is a bit confusing, because both the Offerer and Answerer can
> be "sender and receiver" in this context.
>
> You are trying to give guidance on what a peer should send where the other
> peer has advertised both the ability to receive cryptex and the ability to
> receive header extensions encrypted as per RFC 6904.
>
> The guidance is to send cryptex unless there is a compelling reason to
> send RFC 6904 (e.g. need for some header extensions in the clear).
> However, this only makes sense if the header extensions to be sent in the
> clear can actually be sent (e.g. if they are sendonly or sendrecv on the
> sending peer and recvonly or sendrecv on the receiving peer).
>
> On Tue, Aug 2, 2022 at 2:44 AM Sergio Garcia Murillo <
> sergio.garcia.murillo@gmail.com> wrote:
>
>> Hi all,
>>
>> As discussed during the last AVTCORE meeting, I have prepared a PR for
>> adding the recommendation of using Cryptex over RFC6904 adding an exception
>> in the case of some of the header extensions should be sent in clear for
>> RTP middleboxes processing:
>>
>> If both Cryptex and the Encryption of Header Extensions mechanism defined
>> in {{RFC6904}} are supported by both the sender and receiver, it is
>> RECOMMENDED to use Cryptex over {{RFC6904}} so all the header extensions
>> and CSRCS are encrypted, except when some of the header extensions should
>> be sent in clear so they are processable by RTP middleboxes, in which case,
>> it SHOULD use {{RFC6904}} instead.
>>
>> https://github.com/juberti/cryptex/pull/111
>>
>> Would be great if someone could review the wording as I think it could be
>> simplified.
>>
>> I will be on vacation from next week, so I will merge the PR and submit a
>> final draft by the end of this week if I don't receive any further feedback
>> on the draft.
>>
>> Best regards
>> Sergio
>> _______________________________________________
>> Audio/Video Transport Core Maintenance
>> avt@ietf.org
>> https://www.ietf.org/mailman/listinfo/avt
>>
>