Re: Extension to HTTP V1.x (was HTTP RFC)

Christopher McRae ([email protected])
Mon, 16 May 94 19:43:30 -0700


Jared Rhine writes:
> If the protocol is completely stateless (as any MIME-oriented MGET proposal
> is), it will have to resend the entire query, forcing the server to do all
> the processing again. Only in a stateful protocol can this be implemented
> cleanly and efficiently.

The HIDDEN attribute of FORMs was added as a means for passing out-of-band
information such as session id's back and forth. The intent, I believe, is
that such info is just part of the data as far as HTTP is concerned, so the
protocol remains stateless.

Browser developers should anticipate the likelihood that this attribute may
be very large relative to the values of most other attributes. That is, don't
assume that an attribute value is always going to be small.

I don't think the HIDDEN mechanism solves all the problems with our current
client-server transaction model, however. I believe that browsers will need to
provide and advertise a 'callback port' when they access certain services.
This would avoid the client having to wait and/or poll a server when some
request takes a long time to process.

Chris
-----------------------------------------------------------------------
Christopher McRae [email protected]
President, SIGWEB 415/242-9623
Project Manager, O'Reilly & Associates, Publishers 510/540-6036