Re: Helpers for enhanced UI and protocols? (was WWW UI events)

Paul Burchard (burchard@horizon.math.utah.edu)
Sun, 22 Jan 1995 04:01:33 +0100


narnett@verity.com (Nick Arnett) writes:
> [ Paul Burchard <burchard@math.utah.edu> writes,
> in response to Daniel Dardailler <daniel@x.org>: ]
>
> >Agreed. While the Web needs a way to handle stateful transactions,
> >such "session" capabilities must be added with *great* care.
>
> I'm not as certain about this as I formerly was. Perhaps we
> should think of stateful protocols as something that
> should be implemented in task-specific helper
> applications.
..
> This approach seems fundamentally good to me from the
> standpoint that it helps HTML and HTTP from trying be all
> things to all people.

That's the source of my caution. I'd much rather keep HTTP stateless
than to have to deal with poorly designed, limited statefulness in
the protocol. And I've been very vocal in defending the stateless
purity of HTTP on this list.

But at the same time, we're seeing a proliferation of stateful helper
applications and protocols, with lots of overlap and duplication.
The result is that none of these systems is widely used enough to be
compelling for most users to install (or for popular clients to
support them directly). That's not an issue in the kind of
commercial niche you describe (for which a specialized application is
probably better anyway). Nevertheless, I feel that there is a broad,
common need that could be better filled by some standard than by the
current cacophony of local hacks.

W3C has identified this as an target area of work and, once they get
the ball rolling I trust them to do a good combination of design,
testing, and feedback. (I have no intention of even considering
CCI++ as a contender in this regard, at least until its designers
come clean on their proprietary intentions. Can you say "Unisys"?)

--------------------------------------------------------------------
Paul Burchard <burchard@math.utah.edu>
``I'm still learning how to count backwards from infinity...''
--------------------------------------------------------------------