Re: [tsvwg] draft minutes from interim

[Pete Heist <[e-mail]>]

> On Mar 7, 2020, at 5:03 PM, Bob Briscoe <[e-mail]> wrote:
> 
> Joanathan,
> 
> For all L4S code, as said many times, pls always see links from the L4S landing page.
> https://riteproject.eu/dctth/#code
> 
> Also, normally, if someone is really interested in using some code and cannot find it, they just ask.

Any additional tests we provide will continue to use the ‘testing’ (default) branch at:

https://github.com/L4STeam/linux

This repo helps us identify what is ready for testing, and provides a single kernel where all the required pieces have been integrated together.

Pete

> See inline for specifics...
> 
> On 07/03/2020 08:50, Jonathan Morton wrote:
>>> On 5 Mar, 2020, at 5:04 am, Wesley Eddy <[e-mail]> wrote:
>>> 
>>> https://datatracker.ietf.org/meeting/interim-2020-tsvwg-01/materials/minutes-interim-2020-tsvwg-01-202002200900
>> Following up on this point from the minutes:
>> 
>>> - Bob says all code has been available for some time.
>> We had checked on the morning of the call to find that no commits had apparently been made to the previously publicised L4S codebase since November.  Nevertheless, the L4S team's presentations and commentary appeared to rely on new work which clearly post-dated that.  It was subsequently established that the paced-chirping and classic-detection code in particular had *not* yet been published anywhere, and we have also determined that the queue-protection code is not implemented in the public Linux version of the DualQ qdisc.
> ==Paced Chirping==
> Paced Chirping has been published code since Mar 2019 at least (when we explained it at netdev). The link was in the netdev presentation, video, paper, etc. as well as off the L4S landing page, as well as IRTF iccrg presentations, mailing lists, etc:
> https://github.com/JoakimMisund/PacedChirping
> 
> The reason the Paced Chirping repo is separate from TCP Prague is that it's intended to be useful for any congestion control (it solely uses delay), altho you'll probably have to do some work to integrate it into each CC.
> 
> Under there, you'll find that, in Feb Joakim integrated it into TCP Prague (rather than DCTCP):
> https://github.com/JoakimMisund/net-next/tree/tcp_prague
> 
> ==Classic ECN AQM detection==
> As I said during the presentation, the code was not quite available for the interim, but we were going to publish it shortly. Olivier is now testing its inclusion in an integrated version of TCP Prague. And we are preparing publication of a large set of test results, so they should be rather more self-explanatory than they are right now.
> 
> 
> 
> 
> Bob
> 
>> 
>> Since these appear to be important components of the L4S system, can you point us towards where this code is available to evaluate?
>> 
>>  - Jonathan Morton
> 
> -- 
> ________________________________________________________________
> Bob Briscoe                               http://bobbriscoe.net/
>