Re: [Tsv-art] do we need a checklist or guidelines?

Hi Joe,

I would say that the points under c) are usually already discussed and considered in the IETF (in tcpm where we sooner or later have all these proposals presented at least once). However, I don’t see a need to further formalize anything here because I don’t see how this could help.

I also think that „attack“ is really not the right word here. People are free and should be encouraged to bring their idea to the IETF/tcpm and then have a discussion with the experts there figuring out if their proposed solution is the right approach or if there are other better alternatives. Just trying to send these people away doesn’t help.

Mirja

> Am 19.07.2017 um 20:16 schrieb Joe Touch <touch@isi.edu>:
> 
> Hi, all,
> 
> I'm seeing increasingly bizzare attempts to modify TCP.
> 
> We currently have two requirements for TCP mods, AFAICT:
> 
>    a) passes TSV-chair, TSV-ART, and possibly TCPM, TSVWG, etc. review
> 
>    b) includes "TCP-friendliness" (mostly for congestion control)
> 
> I propose it's time for some new criteria, as follows. Can others let me
> know if this would be useful to document somehow, whether formally as an
> ID or at least as a TSV-ART checklist?
> 
>    c) compliant with existing TCP requirements, notably:
> 
>        1) compatible with already-defined TCP options
> 
>        2) fail-safe fallback to legacy endpoints (notably not requiring
> a new connection attempt with different parameters)
> 
> Criteria "c)" is currently under "attack" (imo) by some in the LWIP and
> MPTCP groups, the former which hasn't addressed this issue at all and
> the latter appears to be actively rejecting it.
> 
> Thoughts?
> 
> I'd be glad to write this up further, with justification.
> 
> Joe
> 
> _______________________________________________
> Tsv-art mailing list
> Tsv-art@ietf.org
> https://www.ietf.org/mailman/listinfo/tsv-art