TAG weekly meeting

 

Scribe Ed Rice

Chair Norm

Date: 08 March 2005

RESOLTUION: MINUTES ACCEPTED FROM F2F MEETING

Periodic report

Resoltuion: accepted Periodic report, will be cleaned up and distributed.

<scribe> ACTION: Norm to clean up and distribute Periodic Report [recorded in http://www.w3.org/2005/03/08-tagmem-minutes.html#action01]

<Noah> http://lists.w3.org/Archives/Public/www-tag/2005Feb/0013.html

Technical discussion. Releationship of URI Schemes to protocols

<Noah> http://larry.masinter.net/draft-hansen-2717bis-2718bis-uri-guidelines-03.html

Noah presented topic

outstanding questions;

- when you write the spec for a new URI scheme what much/should/want you do in order to specify the new URI scheme?

<Zakim> timbl, you wanted to suggest that this is important for evolvability. A scheme is an abstract scheme, and a set of names lasts longer than a given network protocol.

<Zakim> Norm, you wanted to ask about the tension between defining operations on the scheme and the number of schemes

Tim: really what we're saying is that invarient is for the name space.
... Post was not defined to be a catch-all but now it's used as a SOAP catch all.

Norm: The question was about new operations.

<Zakim> Noah, you wanted to respond to Tim: we need to explain when doing p2p when you should use http: scheme and when p2p:

Noah: Tim made the case that you could keep a scheme like HTTP and keep the operations that go with it.
... Suggests, we should give guidence: HTTP is the last scheme or when you should move to P2P:

Ed: caution in saying we should never say never.

<timbl> Noah: "I'm going to ascribe to Tim more than I think he belives"

<timbl> (and he did!) :-)

Roy: http: scheme is intended to map out an identification of resources intended on an HTTP server.

<Norm> Norm hears that a key point is that the resources are identified by http: and the system is free to use another protocol to get representations of *those resources* if it happens to know of one

<Zakim> Noah, you wanted to ask Roy for clarification: must p2p resources also be available from http server (at least in principle) to use http: scheme for them?

<timbl> It really helps to have clear distinctions in an architecture between different sorts of thing. Resources, Methods, TCP ports, IP addresses, telnet services, mailboxes are all fairly well defined. The architecture is strengthened by the things you can't do with those things as well as the things you can do. If you get confused between e.g. documents and endpoints the arch becomes a worse foundation, you can't predict what you'll be asked to do, etc. You then

tim: you update the architecture by defining blocks which have clearly defined functionality.

<Zakim> Noah, you wanted to ask, do resources keep their names as their capabilities grow?

<timbl> For example DAACP

<timbl> as an HTTP-like protocol which someone felt had to be invented and be different from HTTP.

Tim: Sometimes we do need to, and have, evolved standards
... in general, we should not slowly evolve HTTP mappings, but we should be able to evolve the mapping under HTTP

Roy: Disagrees with Tim
... HTTP has evolved significantly

<Noah> Noah is glad to see the p2p issue getting positive attention. Some may remember me giving talks a few years ago suggesting that the P2P community needed to integrate with the web and vice versa

Tim: The role is the same, but the main purpose is the same.

<Norm> Roy: what distinguishes the HTTP of '93 and onwards is its ability to act as a proxy protocol

<timbl> URI-Scheme-Protocol

<Norm> uriSchemesProtocolsOperations-xx?

<Roy> protocolSchemes ?

naming of the issue (suggestions)

<Norm> uriSchemesProtocolsOps-xx?

<timbl> SchemesProtocols

<Norm> schemesProtocols-48

<Noah> Thus, I hereby propose to open a TAG issue

<Noah> regarding the relationship of URI schemes to protocols and on-the-wire

<Noah> application semantics.

adtoption of Schemesprotocols-48

<Noah> regarding the relationship of URI schemes to protocols, operations and on-the-wire

<Noah> application semantics.

Resolution: Adoption of SchemesProtocols-48 noted.

<scribe> ACTION: Noah to own draft skeleton of SchemesProtocols-48 finding and send around for comments. [recorded in http://www.w3.org/2005/03/08-tagmem-minutes.html#action02]

<scribe> ACTION: Noah to send announcment note to www-tag to announce SchemesProtocols-48 issue and gather feedback. [recorded in http://www.w3.org/2005/03/08-tagmem-minutes.html#action03]

Nsuri document needs to be revisited

Tim to outline reasons for revisiting.

<Norm> http://lists.w3.org/Archives/Public/www-tag/2005Feb/0114.html

<Norm> http://www.w3.org/1999/10/nsuri

Tim: Note: To give a specification the necessary stability, provisions two and three of the template MUST NOT be used in a Proposed [Edited] Recommendation or Recommendation.
... is too strong.

<Zakim> Noah, you wanted to suggest that we have issues at two levels, namespaces in general and w3c namespaces in particular

Noah: suggests there is a lot of controversy about this. Some think a name space should never be changed. Tim is referring to the w3c document.
... bigger issue, TAG should put some effort into evolution of namespaces

<Noah> I'm specifically saying we should be careful not to offend the sensibilities of those who will feel that an important architectural issue has been teed up but not resolved.

<scribe> ACTION: Tim to provide a draft of new URI doc and ask the proposed policy around www-tag [recorded in http://www.w3.org/2005/03/08-tagmem-minutes.html#action04]

<Noah> Should Tim's action be about "URI doc" or namespace doc? I think the latter. I can correct the action if you all agree.

<timbl> http://www.w3.org/1999/10/nsuri

educational materials

at the f2f Dan took an action to follow-up

need to setup infrasturction/pages/processes prior to soliciting feedback.

meeting adjourned