Corinna Vinschen corinna-cygwin@cygwin.com
Mon Jan 18 09:16:00 GMT 2010

On Jan 17 22:35, Pierre A. Humblet wrote:
> At 05:52 PM 1/15/2010, Pierre A. Humblet wrote:
> >The scenario you describe (one packet only, with a long delay between accept
> >and WSAEventSelect) could easily be tested to settle the matter.
> >Put a sleep before fdsock !
> To close the matter, I have done just that, putting a 60 s sleep in
> :accept4 between the call to Windows accept and fdsock. Packet
> doesn't get lost :)
> Server:
> 2010_1_17.22:31:41 Listening
> 2010_1_17.22:32:43 Accepted
> 2010_1_17.22:32:43 Read 6 hello
> Client:
> 2010_1_17.22:31:43 Connecting to localhost
> 2010_1_17.22:31:43 Connected to localhost
> 2010_1_17.22:31:43 Written 6 hello
> 2010_1_17.22:32:58 Exiting

Cool, thank you!  So we can settle down with the Linux-behaviour in
terms of O_ASYNC.  In theory I'd like to make the base accept(2)
function behave like on Linux as well in terms of O_NONBLOCK, but
since that potentially breaks packages which erroneously expect BSD
semantics, it might be a bad idea...


Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

More information about the Cygwin-patches mailing list