[tsvwg] Re: The TSVWG WG has placed draft-duke-tsvwg-udp-ecn in state "Candidate for WG Adoption"

[Martin Duke <[e-mail]>]

Hi Koen!

On Wed, Nov 13, 2024 at 4:16 AM Koen De Schepper (Nokia) <
[e-mail]> wrote:

> Hi all,
>
> I strongly support this adoption. Thanks Martin, for collecting this so
> far. We experienced a similar search for the correct calls on multiple
> platforms during the UDP-Prague implementation.
>
> Some questions/comments/suggestions (didn't read the doc in full detail
> yet):
>
> Exact code fragments would be very useful (or references to it).
>

I did provide links to the Chromium source. I'm reluctant to provide line
numbers as those will move, but searching for the relevant cmsg types and
sockopts should work very well. If there is other publicly available code
that people would like to share as an illustration, I will happily accept
PRs that provide additional links as references.

I will even accept code snippets in the text, within reason, assuming
the IPR implications are tractable. I would need guidance on the IPR
matters.


>
> Would be good to extend even to higher layer "platforms" than OSs, like
> Android, Unity, WebRTC, ... or whatever makes sense.


I am not sure what to do with this comment. Feel free to file an issue on
the GitHub with something more specific if you like.


>
> How long will this doc stay up to date, and can/will/should we maintain it
> (updates, new revisions?) or is a "one-shot getting started" OK?
>

IIUC, OSes tend not to blow up their APIs on a regular basis. I think it is
reasonable to have an objective of capturing this moment in time.

If any OS maintainer wants to communicate that their API is going to
change, I am happy to document that. If they are all converging on a single
API, that's even better. If that actually ships before this document does,
I'm happy to throw out the document and write down the latest snapshot.

Publishing an update to this document in a later RFC is up to the Working
Group. I am not going to commit to that effort at this time.


>
> I liked the common API (and behavior) idea. Can it be the basis for
> defining/agreeing/recommending such an API, or are there other venues to
> come to a more standardized API?
>

I am not authoritative for any of these OSes. I am happy to document
planned changes. If the maintainers choose to use this document as a
guideline, I am happy to facilitate that.


>
> Another hurdle to tackle is pacing and its APIs. I saw there is a draft
> available (Michael and Wes). I didn't read that in full detail either, but
> I have the impression that this draft is more on the justification, less on
> actual implementations and their (best) APIs on different platforms. Maybe
> the format/structure/contents could be aligned?


I agree with the apparent consensus in Dublin that the scope of this
document should remain narrowly focused on ECN, but the group welcomes
other submissions related to other sockopts.

To be clear, I am not filing any github issues based on this review, but
would welcome an issue with something more actionable if you feel one is
needed.

Thanks,
Martin


>
>
> Regards,
> Koen.
>
> -----Original Message-----
> From: IETF Secretariat <[e-mail]>
> Sent: Wednesday, November 13, 2024 8:57 AM
> To: [e-mail]; [e-mail];
> [e-mail]
> Subject: [tsvwg] The TSVWG WG has placed draft-duke-tsvwg-udp-ecn in state
> "Candidate for WG Adoption"
>
>
> CAUTION: This is an external email. Please be very careful when clicking
> links or opening attachments. See the URL nok.it/ext for additional
> information.
>
>
>
> The TSVWG WG has placed draft-duke-tsvwg-udp-ecn in state Candidate for WG
> Adoption (entered by Gorry Fairhurst)
>
> The document is available at
> https://datatracker.ietf.org/doc/draft-duke-tsvwg-udp-ecn/
>
> Comment:
> Hello TSV WG,
>
> This is a formal adoption call that will run 2 weeks, ending on 29th
> November 2024. Please respond to this email thread with any comments
> indicating support or objection to adopting this draft as a basis for work
> on ECN support for UDP.
>
> During IETF-121, the WG discussed "Configuring UDP Sockets for ECN for
> Common Platforms" (draft-duke-tsvwg-udp-ecn)  and the sense in the room
> indicated some support for adoption, this email provides an opportunity to
> contribute comments on this planned work.
>
> Best wishes,
> Gorry and Marten
> (tsvwg co-chairs)
>
>