[6] in 6.033-lab
Sequential Request ID problems
daemon@ATHENA.MIT.EDU (Nat Friedman)
Thu Mar 6 23:58:24 1997
Date: Thu, 6 Mar 1997 23:57:23 -0500
From: Nat Friedman <ndf@aleph1.mit.edu>
To: "Constantine P. Sapuntzakis" <csapuntz@MIT.EDU>
Cc: 6.033-lab@MIT.EDU
In-Reply-To: <199703070446.XAA14431@playdoh.mit.edu>
Constantine P. Sapuntzakis writes:
> This approach does NOT work. The problem is that there is no easy way to
> maintain monotonically increasing in two clients on a given machine/port
> where one program run the other) but coincidentally listening on the
> same machine/port. Also, in the case of a client crash, all information
Color me brash, but I think this is wrong. The network address/port pair
always uniquely identifies a socket -- how else is the data supposed to
get to the right place? There is nothing else in the protocol or kernel
socket data structure to identify the socket.
I can't find the place in Stevens where this is said, though.
-Nat
