[tsvwg] Re: draft-ietf-tsvwg-udp-ecn-01 ready for WGLC?

["Koen De Schepper (Nokia)" <[e-mail]>]

Hi Martin,

I didn't have the time to review the doc in full detail and cross check with our implementation in UDP-Prague, but maybe following independent implementation reference to a simple wrapper on a UDP socket that supports ECN setting and getting on all the platforms you mentioned might be useful to add in this draft: udp_prague/udpsocket.h at main * L4STeam/udp_prague * GitHub<https://github.com/L4STeam/udp_prague/blob/main/udpsocket.h>. I had a look at the extensive Chromium-Windows reference, which probably has much more functionality, but I think is harder to find the "to the point" ECN implementation. Also seeing the implementation diffs on all platforms side-by-side and having a common wrapper API might lower the bar for developers using it.

Note that I didn't cross check if our implementation is in line with the draft, and we are open to corrections/improvements/extensions (eg, not sure we covered all IPV6 cases) and maybe we have used it in another way not yet described. Anyway, it has been tested and works on all mentioned platforms (but not tested in combination with IPV6).

Regards,
Koen.

From: Martin Duke <[e-mail]>
Sent: Monday, July 7, 2025 11:24 PM
To: [e-mail]; tsvwg <[e-mail]>
Cc: Gorry Fairhurst <[e-mail]>
Subject: [tsvwg] draft-ietf-tsvwg-udp-ecn-01 ready for WGLC?


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.


As an individual (author):

TL;DR WGLC please.

I've resolved all but one outstanding issue on the new version of the draft:

https://datatracker.ietf.org/doc/draft-ietf-tsvwg-udp-ecn/

Resolved:

1. Support for other languages: given that this a potentially large scope increase, and the energy surrounding this draft is declining, I have opted to instead add text clarifying that this document describes C APIs.

2. Although I am not totally convinced that the problem that spawned the issue is real, after a little code inspection I've described the lengths of the cmsg values that various OSes report.

If there are any objections to these changes and issue resolutions, don't hesitate to speak up.

There is still no action on a final read-through from the Apple folks, though I am confident that the text is correct based on the fact that it works in Chrome on MacOS.

To my co-chair, I ask for this draft to be placed in the queue for WGLC. I believe this has received enough review for an informational RFC and there is little to be gained by additional author effort. To the extent that it could use another read-through, WGLC is a good forcing function for that.

Martin