Re: [tcpm] Working group acceptance call for draft-touch-tcpm-2140bis

["Scharf, Michael (Nokia - DE/Stuttgart)" <[e-mail]>]

> > The feedback in the room is clearly different to your conclusion, as a
> substantial part of the community seems to see value in an informational
> document but not a BCP.
> 
> An informational doc would not appropriately include recommendations.

An informational document can document existing practice without normative recommendations and at least my take is that this is what the WG is comfortable with as there is no need to define "best", which is challenging for stack-internal implementation choices.

> > It is not uncommon that authors are convinced that their document
> warrants a more normative status than parts of the working group. But in
> that case it is up to the authors to work with the WG to reach consensus. The
> proposed charter milestone allows that.
> >
> > I do not believe that the TCPM chairs should declare rough consensus on a
> BCP document, in particular not at early stage. It is TCPM's tradition to try to
> achieve unanimous consensus.
> 
> I was not asking for that. I am claiming that there is a misunderstanding of the
> work difference between informational and BCP. The items the WG appears
> to want added would imply BCP. If BCP isn't the target, then we can excise a
> lot of the current material and jump to WGLC.

It is a misconception of the working group consensus that a BCP is needed or wanted. A subset of the working group believes that a BCP document can be worked out while there is consensus that an INFO document can be published.
 
An informational document can document implementation behavior and provide rationale for that, but probably should avoid RFC 2119 language. A BCP would make normative recommendations. As of now, it is unclear whether the WG would reach unanimous consensus on normative statements of what is "best" for implementation internals so it seems difficult to make a call right now. 

> I'm saying the WG can't "eat it's cake and have it too".  If the consensus is to
> add BCP material, then it needs to adopt as a BCP work item. If there is no
> consensus on BCP,  then the work should not progress along those lines.

The WG can only make a call on BCP when there is exact wording for the proposed BCP material. And this requires a specific text proposal with a clear separation between BCP and informational material. The authors have not proposed so far such a split between BCP and INFO and so they cannot ask the WG to make a call on that.

I think the consensus is that the WG is open to a proposal for BCP material but careful review is needed and in case no consensus will be achieved on the BCP material, the document would have to be INFO as a whole, or INFO for all parts that do not get strong consensus. I don't think the authors can ask for more from the WG with their wording in -02, which does not describe the scope of a BCP.

It is understood that a BCP would require more work in the WG but the proposed charter milestone would allow that work, provided that the authors are willing to engage.

I also think the WG would be very happy with an informational document that just updates 2140 and surveys actual implementation practice, which is basically what the authors have proposed in -02 so far. Also in that case the rationale for statements would have to be carefully reviewed. But it would be easier to deal with all cases where implementation-internal choices are different or where it is difficult to define "best", "safe", etc. This could save a lot of WG cycles on discussing implementation-internals.

If the authors (and the other supporters of BCP) would agree to a target status of INFO, I would be more than willing to change my proposed charter wording to INFO.

Michael