- From: Daniel DuBois <ddubois@spyglass.com>
- Date: Mon, 7 Aug 95 19:46:47 -0500
- To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
!!!! This is not about Session-IDs. It's about persistant connections. Is the flipping from "Connection:" to "Session:" in the Alex Hopman session extension draft intentional? Even from a third reading, it doesn't seem so. However, this could be a potentially beneficial typo. I would like to see BOTH a Connection: header and a Session: header, the first indicating that only the connection is persistent, the second indicating that not only is the connection persistent, but various headers including the Accept: and WWW-Authenticate: are persistent/saved as well. Implementation of the acceptance array storing is annoying at best, because it brings information that naturally pertains to 'request objects' up to a 'connection object' level, and could get really ugly. The benefits are probably negligible relative to the TCP overhead, certainly for Accept: headers, and maybe even for WWW-Authenticate: computations. Anyway, having both Connection: and Session: and would greatly speed up wide-spread implementation (of the former at least), which is what we all want, right? PS: Oh, and if this WAS an intentional change of terms, indicating exactly what I described above, please help me remove my foot from my mouth. PSS: This goes without saying, but we don't have to have Connection: and Session:. There could be "Connection: maintain" and "Connection: maintain, accept", "Connection: maintain, authenticate", etc... Or any other of a variety of syntaxes yielding the behavior I described in paragraph 1. ----- Dan DuBois, Software Animal ddubois@spyglass.com (708) 505-1010 x532 http://www.spyglass.com/~ddubois/
Received on Monday, 7 August 1995 17:48:53 UTC