> -----Original Message----- > From: Alan D. Cabrera [mailto:[EMAIL PROTECTED] > Sent: Monday, October 11, 2004 10:00 PM > To: [EMAIL PROTECTED]; [EMAIL PROTECTED] > Subject: RE: o.a.g.network.protocol.SocketProtocol and Error Recovery > > > I think that it needs an error state so that when someone tries to > access the dead protocol, it will throw a ProtocolException. Remember, > that when you are in a serviceRead you're not in the stack, per se. > > I don't think that we need any changes to the Protocol interface but I'm > interested in what you were thinking. >
It may be as simple as the close method in SocketProtocol calling the teardown method on the up protocol (similarly for other protocols). > > Regards, > Alan > > > -----Original Message----- > > From: David Farb [mailto:[EMAIL PROTECTED] > > Sent: Monday, October 11, 2004 9:45 PM > > To: Dev > > Subject: o.a.g.network.protocol.SocketProtocol and Error Recovery > > > > > > o.a.g.network.protocol.SocketProtocol does not percolate a client > error > > up the protocol stack when the client disconnects. For example, when > > serviceRead gets an IOException, the socketChannel is closed, but > > the up protocol is not informed. > > > > Is this by design? If so how is the up protocol to find out that the > > client > > has gone away? > > > > If this is an error, I can probably come up with a solution, but I > would > > be interested in opinions on how best to fix it, as it probably will > > require > > changes to the o.a.g.network.protocol.Protocol interface. > > > > Thanks > > David Farb > > > >
