Noah Davids <(E-Mail Removed)> wrote:
> I believe that if the client application has set SO_LINGER as a
> socket option with a timeout of 0 it will send a reset when the
> connection is closed.
> I am pretty sure that Windows will do this, not sure about Linux.
My understanding is that just about any stack's socket interface is
supposed to behave that way with SO_LINGER set to a timeout of 0.
While I generally frown upon it with extreme prejudice, both HP-UX
11.X and Solaris (IIRC Solaris still has it) has an ndd option
(tcp_discon and/or tcp_discon_by_addr) that will effectively do the
same thing to a local endpoint, and I believe cause it to issue an RST
segment to the remote. So, if one half of your connection were on
such a platform, you could get by without special case code in the
application itself, or needing to have knowledge of the current
sequence numbers for the connection etc that one needs to "forge" the
RST segment from an external application.
rick jones
--
oxymoron n, Hummer H2 with California Save Our Coasts and Oceans plates
these opinions are mine, all mine; HP might not want them anyway...

feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...