Re: SOCKET RECEIVE techniques...

New Message Reply Date view Thread view Subject view Author view Other groups

Subject: Re: SOCKET RECEIVE techniques...
ifranzki@googlemail.com
Date: Thu Aug 14 2008 - 12:15:23 EDT 

Hi Randy,

<snip> If the TCP/IP stacks at each end aren't taking care of this for
the TCP/IP application, then who is going to do it? The TCP/IP
application
program? </snip>

You are right, the TCP/IP stack of course should take care about the
ACKs (thats what a TCP/IP stack is for). But why should the sending
application have to wait for that?

<snip> But use of ACKs guarantees delivery between TCP/IP stacks and
this allows the *TCP/IP application* not to
have to worry about that part - and that is a very good thing for
everyone developing TCP/IP applications.  </snip>

You state it currectly, the TCP/IP application should not have to
worry about it, thus it should not be put into wait. The stack will
handle the ACK processing asynchronously, without the application to
worry about. If the packet gets lost, the stack will retransmit it
(again, the application does not have to worry about this). If the
packet gets completely lost, the stack will flag the connection as
being broken.

Again, no other TCP/IP stack that exists in this world (at least non
that I am aware of) does wait for the ACK at send..... So it can't be
too wrong what these stacks do....

Kind regards,
Ingo Franzki, IBM

New Message Reply Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b25 : Sat Aug 30 2008 - 00:35:13 EDT