- From: Paul Leach <paulle@microsoft.com>
- Date: Fri, 16 Feb 96 12:16:14 PST
- To: carlos@patora.mrec.ar
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Carlos said: ] > ] ] > ] How does the client signal, after being continuously pushing requests ] > ] into the pipe, that it doesn't want anything else ? ] > ] > I'm not sure I understand the question. I think the answer is that ] > when the client has no more requests to make, and has recieved all the ] > responses it wants to wait for, it closes the connection. ] ] It could even shutdown the connection for writing when it's sure it has ] sent all its requests. But, and this is my personal taste, I'd like the ] client to send a last request (may be empty) indicating that it doesn't want ] anything else from the server, rather than just shutting down the connection. ] ] I don't like to rely on the return codes of the server's ] transport API to realize that the client doesn't want anything else. ] In other words, the semantics of "the connection has been closed" ] and "the client stops pushing requests into the pipeline" may differ. This is an issue with the current draft of 1.1, without the addition of multiplexing, if the client uses pipelining. The persistent connection subgroup was very careful to check that pipelining could be implemented in 1.1 (i.e, was legal). All I'm trying to say is that it is an orthogonal issue to multiplexing -- and that if its really a problem, we should fix it regardless of what we do for multiplexing, and use the solution in the multiplexing case as well. Any comments from the persistent connection folks? (Others, too, of course...) Paul
Received on Friday, 16 February 1996 12:14:48 UTC