Re: HTTP Session Extension draft

On July 5th, 1995, Ted Hardie writes:
>I notice in the draft that you use Connection: maintain in section 1.2
>and the bit about server-proxies, but use Session: maintain in the
>examples.  I personally prefer the Session: maintain, but I can see
>reasons to use both.
Good eye! I missed that one in editing. If I recall correctly I switched my
proposal to the Connection: language some time back at a couple peoples
suggestion. I don't really care, but its going to stay Connection: maintain
unless I get mail from a couple people requesting that I change it, or mail
from one person with a good reason. :) Send in your ballots now. 

>
>I also wonder how difficult it would be to allow either side to open
>the negotiation for a maintained session.  Could we allow the server
>to send "Session: maintain" in response to a request which (from the
>servers' point of view) is likely to be followed by requests for which
>a session-orientation is useful?  Many servers would like to have
>"click-trail" traces of web users' traversals; if you allow the server
>to initiate a session, the current kludges using cgi scripts and
>hidden variables could be eliminated.  The browser could respond to
>such a request by a server with a simple Session: maintain
>confirmation.
I don't think that would work. The main reason is that most clients are
waiting for the server to close the connection to know when they have
reached the end of a response. Therefore when a non-compliant client
connects to a server and the server tries to open the negotiation, the
client will be stuck in limbo. If this extension gets incorporated into HTTP
1.1 or something where compliance is necessary in order to advertise a
client as being HTTP 1.1, then maybe we could have servers initiate the
sessions. For right now it seems to me that clients should always negotiate
the Session Extension and then just close the connection if they have
nothing more to request.

I should also add that I got some excelent mail from Jeffrey Mogul pointing
out that servers, even after they have negotiated the Session Extension are
allowed to close the connection at any time, and that they can choose to not
send a content length or a multipart message, but just close the connection
and the end of a response.

Alex Hopmann
ResNova Software, Inc.
hopmann@holonet.net

Received on Wednesday, 5 July 1995 13:33:48 UTC