RE: HTTP futures workshop at WWW7

Hi Mike,

Rohit, Henrik, Roy, Henry and I discussed this topic during the LA IETF 
last week, and came to the conclusion that the workshop is still very much 
worth holding.

I can understand your concerns about the community of participants not 
being large enough or different enough from the HTTP-NG group to have a 
good discussion during the workshop, but there are several reason why I 
feel the workshop is still a very good idea.

Addressing the lack of diversity concern, given that Rohit and I aren't 
part of the HTTP-NG effort yet have significant views on this topic, and 
since I Hugh Davis will bring a new an interesting perspective from the 
Open Hypermedia community, I tend to think there will be sufficient 
diversity of opinion at the workshop.  Furthermore, we expect to gather a 
few latecomers to the workshop based on discussions we had at the LA IETF.

One area where I think there can be much fruitful discussion at the 
workshop are the impact(s) of providing WebDAV support in HTTP-NG, since I 
don't feel the HTTP-NG group has had a full DAV advocate, and our 
experience with WebDAV is that there are several suboptimal choices that 
are made when you "bolt-on" to a non-DAV aware protocol like HTTP/1.1. 
Providing WebDAV support is far more complicated than the bullet on this 
topic in the "Goals" document.  I'm currently working on my position paper 
on this very subject, but support for WebDAV has far reaching effects, 
including:

- support for operations which affect more than one resource (e.g., have a 
source and a destination like MOVE and COPY)
- operations which operate on a hierarchy of resources
- operations which have a source and a destination and which operate over a 
hierarchy (e.g., COPY hierarchy A to B)
- improved error reporting (the error could be on the source, or the 
destination, or both simultaneously)
- error reporting precedence rules for handling cases where multiple 
simultaneous errors can occur
- interactions between parameter encoding issues like extensibility, 
internationalization and ease of rapid parsing for security issues, or 
server efficiency
- strong precondition evaluation support
- caching of parts of resources, like properties
- support for searching, and a new kind of resource, the "arbiter"
- a data model which distinguishes between active (e.g., CGI, ASP, etc. and 
static resources)

Many of these issues are not explicity addressed in the "Goals" document.

So, my feeling is, though the turnout is less than expected, the "right" 
people are attending, and I expect the resulting discussion to be of high 
quality, more than justifying the expenditure of a day.

- Jim

On Tuesday, April 07, 1998 2:16 PM, Mike_Spreitzer.PARC@xerox.com 
[SMTP:Mike_Spreitzer.PARC@xerox.com] wrote:
> I'm given to understand that the signups for this workshop consist mainly 
of
> HTTP-NG folks and Australian on-lookers (I saw Jim's note saying Hugh 
Davis
> will be there for part of the day, but this is the only exception to the 
above
> rule that I'm aware of).  This leads me to doubt whether much actual work 
can
> be done.  Perhaps we should disband this workshop, freeing the 
participants to
> find a more productive use of the time?
>

Received on Tuesday, 7 April 1998 18:33:44 UTC