Proposed revisions to more carefully deal with typing of resources

At the Basel meeting  I raised some concerns that I have had for some time 
regarding the treatment of resource typing in the architecture document. 
Specifically, I feel that the existing drafts go a bit too far in implying 
that representations must in all future protocols be octet streams typed 
with Internet Media types.  I was assigned an action which is paraphrased 
in the minutes as [1]:

ACTION NM: to take a run through to see how generalizing 'representation' 
to be less constrained would look with more careful terminology, report on 
whether this looks feasible or not.

For those not at our meeting, the intention was that I would fork a 
private copy of the document and edit it to show proposed changes.   This 
note fulfulls that action;  a revised draft is attached.   This draft has 
no official status except that the TAG has committed to take a look and 
evaluate whether the changes would be (a) desirable on the merits and (b) 
sufficiently low risk given our publication schedule.

Having approached this exercise with some hesitancy I can say I am 
reasonably pleased by the result; I hope the TAG will adopt something 
close to this.  To my biased eye, the changes are subtle but important 
architecturally, and are complementary to both the flow and the intentions 
of the original.  I believe they strike a good balance in emphasizing the 
use of HTTP, octet streams and Internet media types as near universal 
current practice, encouraging their continued use in the future, yet 
allowing for the possibility that new protocols or a need to deal with the 
known shortcomings of RFC 2046 may someday present a legitimate 
requirement for other forms of representation or a different typing 
mechanism.  I believe it's important that the Web architecture anticipate 
these possibilities, and I hope this draft shows that the needed changes 
are both tractable and desirable.

In any case, I am grateful to the TAG for any consideration given at this 
late stage in the development of the architecture document.

Packaging of the attached files
-------------------------------

The files are provided in a .zip archive.  Unzip to any folder and launch 
the html file.  I'm sure there are  a few loose ends editorially, but the 
essence of the changes should be clear.

All significant changes are highlighted in color and also set off with 
markers of the form [{NOAH>} ..new text here.. {<NOAH}].  Searching for 
the string "NOAH" should get you to all significant changes.

Private copies of the usual stylesheets and .png files are provided, but 
are not modified from the official versiosn on w3c servers;  they are 
provided only so that the existing relative links will resolve.  One 
additional stylesheet and associated link has been added to define the 
highlighting of insertions and deletions made for this experimental 
branch.  Those involved in maintaining editors' drafts should know that I 
did not re-run the autogenerators for TOCs, etc., but rather edited 
everything by hand, in part to preserve change highlighting.  We can 
integrate properly if the changes are adopted.

Thank you!

Noah

P.S. To TAG members:  I would be delighted if these changes were discussed 
on the Monday call, but please recall that I am scheduled to get off an 
airplane an hour ahead of the call and there is some chance that I will 
either be late or completely absent.  Assuming the flight is on time, I'm 
optimistic about making at least the latter half.  Thank you.

[1] http://lists.w3.org/Archives/Public/www-tag/2004Oct/0035.html



--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------