Re: [Rswg] Making progress on evolving the XML format

On Mon, Feb 13, 2023 at 2:48 PM Jay Daley <exec-director@ietf.org> wrote:

> Seems a practical way forward.  A few points below:
>
> > On 13 Feb 2023, at 20:28, Eric Rescorla <ekr@rtfm.com> wrote:
> >
> > 
> > Hi RSWG members,
> >
> > Pete and I have been chatting about the best way forward for evolving
> > the XMLv3 grammar. We think that there's broad agreement that:
> >
> > 1. We need some document that describes the "as-is" xmlv3 format.
> >    - That document should be published in some form that implies
> >      it doesn't have consensus.
> >    - The WG should do the work of making sure that happens
>
> Would we assign that a version number? If so then would it replace v3 or
> be a v3.1 with no documents ever having been published in v3?  Before
> answering this you might want to read further on.
>

This proposal takes no position on this topic. I simply used "3.1" because
that seemed
to be the version with the most traction. Perhaps we should say "NG" :)

> > 2. We need to work on a new format (v3.1?) that will have consensus.
> >
> > I think people are assuming that the new format will need an RFC but
> > there is not agreement on whether the "as-is" should be.
> >
> >
> > Pete and I would like to propose the following way forward, based
> > on a suggestion from Martin Thomson in January.
> >
> > - Adopt draft-irse-draft-irse-xml2rfcv3-implemented as a WG document
> >   to use as the basis for v3.1.
>
> Wouldn’t that draft also be the basis for 1. above?
>

Yes, in the appendix although initially the appendix part of the doc would
be empty.

> - Everything that doesn't match RFC 7990 will be marked as
> >   "provisional". We go through all of these and either adopt them as
> >   having consensus or decline to make the change, in which case we
> >   move them to an Appendix.
>
> If we did remove something, then documents in the new grammar would no
> longer be backwards compatible (unless that thing had never been used) with
> those in the as-is grammar. I would prefer then that we numbered it v4 to
> signal that. If we were to keep the version number 3.1 then we would have
>
> - v3/7990 which was never used for a document
>

I do not personally know this to be true, but I am prepared to believe it
is.

- version ”as-is” which all the current documents are in
>
- v3.1 which none of the existing documents are in and which all future
> ones will be
>

I agree with you that there are these three versions. I think what numbers
to assign
to them is an orthogonal question.

-Ekr

> Jay
>
> --
> Jay Daley
> IETF Executive Director
> >
> > - If necessary, we can also update the I-D to make consensus changes
> >   to the as-is version on an "emergency" basis, though hopefully
> >   this will not be needed often.
> >
> > At the end of this process, we will have a document which describes
> > the consensus format and also an appendix which describes how "as
> > implemented" differs from the consensus format. These can be published
> > together as an RFC, which will provide archival forms of both but
> > without requiring that we publish an RFC of the current document while
> > the consensus version is still in flux. In the meantime, the ID
> > can serve as the definition of the current format.
> >
> > What do people think of this proposal?
> > -Ekr
> > --
> > rswg mailing list
> > rswg@rfc-editor.org
> > https://mailman.rfc-editor.org/mailman/listinfo/rswg
>