Re: New feature negotiation syntax

On Tue, 10 Jun 1997, Jim Gettys wrote:

> For example, I have a (personal) belief that only the client can
> have enough information to make an informed choice, and that having
> a proxy try to short circuit the first round trip will ultimately be futile.

Unless I've missed something in the discussion or missunderstood my
reading of Koen's drafts, TCN provides a mechanism which allows a 
client to express absolute conditions as well as preferences. If
after application of the algorithm, there is only one choice, then
the client is provided with the correct response, otherwise the client
must examine the alternatives and apply additional knowledge it has
about its environment, including possibly asking the user, to then
request a specific choice.

> I've therefore been very skeptical of any q-factor algorithm being
> ultimately useful, and am mostly interested in conveying information to
> a client so it can make a choice on its own.  Hence, Alternates and

In the TCN proposal, the q-factors bound the choices and may result in
a single choice.

> the like are very usefull in my view, and some way to encode features.

> Until there is shared belief on the requirements (or at least some subset
> of total requirements), I'm pessimistic about making progress...

I think the key phrase here is "shared belief" which I've just realized
may describe the situation. I feel there has been an ongoing undercurrent
in the WG since its beginning that content variant handling was kludgy 
at best, that the facilities we have defined conflict with effective
caching, etc. We frequently pan the use of the User Agent field to
control content but provide NO alternative to the application developer.

So like Koen, I'd felt there was already a shared belief that better
negotiation support was required. What is perhaps missing is the level of
shared belief that there is a problem to be solved that is important
enough to solve that we as individuals will spend the time reviewing the
drafts and caching implications.

What we don't seem to have is interest in the issue. To that end, perhaps
a problem/requirements 'document' would help move the issue forward.

Many moons ago Dave Raggett brought up the idea of some kind of features
and mis-features (bugs?) repository which could be some kind of shared
intellegence about specific instances of user agents. Various ideas were
tossed about about who whould store such data, etc. Perhaps we could
come to closure on the requirement for TCN if there were an accepted
mechanism for translating the User Agent field value into a set of
TCN features for those User-agents which didn't declare themselves.
Even the recomendation that server implementors provide the installation
with a configuration option for associating TCN values with UA values? 

It would still be better of the UA would tell the server which local
options the user had activated (Java? JavaScript? Etc.).

Dave Morris

Received on Tuesday, 10 June 1997 11:29:15 UTC