- From: Dave Kristol <dmk@allegra.att.com>
- Date: Sat, 9 Dec 95 20:27:20 EST
- To: sankar@fcrao1.phast.umass.edu
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
"Sankar Virdhagriswaran, Crystaliz Inc." <sankar@fcrao1.phast.umass.edu> wrote: > > Seems like the state-info proposal covers what one would need for the > moment. While it pushes the really difficult aspect of implementing > transactional concepts to CGI bin scripts, one could argue that is how it > should be (since the notion of transaction varies in different application > areas). Let me try to clarify something. The value for the State-Info header is opaque. But just because the specification says it's opaque does not mean a server can't structure the information in some way specific to it, such as making the first part a path. The server could further assume that State-Info headers that return to it have that structure and use the path information to control behavior. There's a false presumption that a CGI-like mechanism must be used for all implementations of State-Info. The mechanism is general and can be implemented and used many ways. Dave Kristol
Received on Saturday, 9 December 1995 17:30:19 UTC