- From: Graham Klyne <GK@ninebynine.org>
- Date: Wed, 09 Jan 2002 09:52:38 +0000
- To: "Ian B. Jacobs" <ij@w3.org>
- Cc: www-tag@w3.org
At 06:18 PM 1/8/02 -0500, Ian B. Jacobs wrote: >The Chair observed that several TAG participants considered Roy >Fielding's dissertation to be good material on Web Protocols, >and invited the remaining participants to review chapters 4 >and 5 and indicate whether they should form a basis for a written >Web architecture. I've taken a look at those chapters, and would like to offer some observations... It strikes me that the primary focus of this material is on protocol interactions, with only passing reference to information representation issues. I think both of these issues (should) have comparable weight in an articulation of Web architecture. Much of W3C's current work is focused on information representations in XML (and RDF). Section 5.2.1 (of Roy's thesis) starts to address the significance of representations (emphasizing the importance of "move the data", rather than taking the processing description to the data). But having acknowledged the importance, the focus remains on protocol interactions, with little development of architectural principles for the content itself. What discussion therte is seems to focus on HTTP/MIME-like protocol elements, as in "metadata is in the form of name-value pairs", something that XML/RDF are allowing us to develop beyond - especially as ideas filter back to the IETF protocol community. This is not a criticism of Roy's thesis [chaps 4,5] per se, but a suggestion that they're only part of what should be addressed by "web architecture". (I suggest that information representation issues must include some account of transformations that can be said to preserve the intended content in some way, such as rendering of HTML, or entailment in RDF.) .. Concerning the discussion of "connector types" in section 5.2.2 - the terminology here seems slightly counter to what is being described. I think the discussion is about interface abstractions, but that's not always clear in the text. The distinction between connectors and components seems a bit fuzzy to me. .. #g
Received on Wednesday, 9 January 2002 05:00:49 UTC