Revising RFC2616 - what's happening

After the Bar Bof (summary: < 
http-wg/2006JulSep/0010.html>), I've been talking to a lot of people  
about the prospect of revising RFC2616.

Most everyone I've talked to agrees on a few things;
    * Focus on errata and clarifications
    * Don't introduce a new version of HTTP
    * Don't break compatibility (a prerequisite for an unchanged  
    * Skew towards implementation experience
    * If it's clearly uncontroversial, deprecate unused/non- 
interoperable features
    * If in doubt, don't change it (do no harm)
    * Don't try to make it perfect, just better

I've also seen a fair amount of private interest in working along  
these lines; not only from people using HTTP for machine-to-machine  
communication (e.g., Atom, AJAX) but also from traditional browser  
and server vendors.

The question now is whether that interest will become public and  
translate into activity. Jim Gettys' efforts at 2616bis didn't gain  
traction because he didn't feel that there was an appropriate pool of  
reviewers, and similarly we're concerned that while a lot of people  
might want to see this work done, we're less sure that enough people  
will actually do the work.

To help find out, a few things have been happening;
    * Yves Lafon and Julian Reschke have published an I-D that re- 
states RFC2616 using xml2rfc, so that people can verify it's a  
faithful transcription. Soon, they'll publish an -01 that  
incorporates the errata that Scott has captured in < 
NET/http-errata> (which didn't require additional discussion).
    * I'm working on an issues list that captures all of the problems  
that have popped up on the list, so that we can track proposed errata  
and clarifications. Expect to see the first revision of that soon.

We also discussed having a BoF in San Diego with the ADs, but decided  
that there hadn't yet been enough activity to make it successful.  
We're hoping that having a draft and an issues list available will  
generate discussion that will demonstrate the community's interest.

Further down the road, we'll need to figure out if, when and how a WG  
should be (re-)formed. There's also the opportunity to do interop  
work, and perhaps even a test suite (something that I've heard a few  
people express interest in), if enough people are willing to do it.

This is where we're at now; the next steps are to get the issues list  
out and start discussion, so as to give us information about the  
level of interest and commitment in moving forward. If you have  
concerns about the scope or path forward, please bring them up to the  
list; likewise, if you're willing to help, we'd love to hear about it.


Mark Nottingham

Received on Wednesday, 18 October 2006 15:30:57 UTC