Re: returning to the previous state

At 2:31 PM -0700 5/19/97, Jim Lawson Williams wrote:
>I guess, since my reading of that spec. is different, it needs clarification.
>The way I read it is
>
>  "The writer (i.e. client, sender) cannot consider the session successfully
>   closed until the reader (i.e. server, receiver) has acknowledged session-
>   end with a close_notify reply.  If, however, the reader decides to close
>   the connection prematurely for whatever reason, then it need not
>   necessarily wait for the writer's matching close_notify."

The difference being:
"you don't have to wait for a response after saying you're going to tear
down the connection in the normal fashion"
vs.
"if you're about to die, say so first, but you don't have to wait around
for a response"

I'd like to see some clarification from the TLS WG on which this means,
because the second one makes much more sense and probably will be invoked
much more rarely than the first. That is, the second interpretation is for
emergencies, and the first is for sloppy programming.

--Paul E. Hoffman, Director
--Internet Mail Consortium