Re: draft-iab-extension-recs & W3C TAG

> At the very least, a discussion of what MAY
> be extended in a protocol *without* coordination vs. what SHOULD be
> extended *with coordination* would be very useful.

I always think about this question in terms of a large sheet of
squared paper, on which every parameter value that has been assigned
is checked off. To avoid disaster, there must only be one copy of that
sheet of paper. Originally, it was held by Jon Postel.

You can hand out strips of the paper to different people, who get to
check off the boxes on their strip. For example, there are strips for
.com and .org, held by different people. Another example is that each
Ethernet manufacturer has its own OUI value.

If you can divide your parameter space up into distinct strips, this
method works without further coordination. If you can't divide it
up in that way, you must have a single point of coordination.

For protocol extensions that are algorithmic rather than just new
parameter values, I'm not sure this analysis applies, because semantic
interactions between extensions will be much more subtle. If you
care about interoperability, I would feel that uncoordinated
extensions are pretty risky.

All IMHO of course.

Regards
   Brian Carpenter




On 2011-11-04 10:21, Nick Gall wrote:
> On Mon, Oct 31, 2011 at 8:27 PM, Peter Saint-Andre <stpeter@stpeter.im>wrote:
> 
>> Dear IAB authors:
>>
>> Do you think that draft-iab-extension-recs has applicability to W3C work
>> on formats and languages?
>>
> 
> As someone who has followed the "distributed extensibility" debate from
> afar, I think draft-iab-extension-recs Section 3.4 "Protocol
> Variations"<http://bit.ly/rJvXLq>applies to the issue. I think the
> section can be read (at least by someone
> like me who is not steeped in the issues) as a repudiation of, or at least
> strong advisory against, distributed extensibility. Here is the pertinent
> advice: "In particular, the IAB considers it an essential principle of the
> protocol development process that *only one SDO maintains design authority
> for a given protocol*, with that SDO having ultimate authority over the
> allocation of protocol parameter code-points and over defining the intended
> semantics, interpretation, and actions associated with those code-points."
> (emphasis added) This language is actually quoted from RFC
> 5704 "Uncoordinated Protocol Development Considered Harmful".
> 
> As I understand it "distributed extensibility" is
> defined<http://bit.ly/vOn20K>as "The ability for a language to be
> extended by multiple parties
> *who do not explicitly coordinate with each other*." (emphasis added) I
> think it is interesting that the W3C (or at least the TAG) think that
> uncoordinated *extensibility *is a good idea, at least in some
> circumstances, while the IETF seems to feel that uncoordinated
> *development*is generally a bad idea. I googled to see if anyone had
> discussed the
> possible conflict between RFC 5704 and "distributed extensibility", but I
> didn't find anything.
> 
> I don't mean to reopen the Issue-41 can of worms, but a brief discussion of
> "distributed extensibility" (possibly in section 3.4) in an RFC on protocol
> extensibility seems warranted. At the very least, a discussion of what MAY
> be extended in a protocol *without* coordination vs. what SHOULD be
> extended *with coordination* would be very useful.
> 
> -- Nick
> 

Received on Friday, 4 November 2011 19:31:37 UTC