Dave Korn wrote:
Because SO_REUSEADDR works _correctly_ on cygwin, perhaps?
Oh, I don't know about that.
As I see it, the purpose of SO_REUSEADDR is to allow a program to bind to a port that still has WAIT states on it. (FIN_WAIT_x, TIME_WAIT, etc...see netstat(1) and the Stevens TCP/IP books.) The stack normally disallows this because packets may still be "in flight" that might be misdelivered if a program re-binds to this port before the wait states expire. A program setting this flag is saying to the stack, "yes, I'm willing to take the risk to avoid waiting". Setting this flag is important on some systems, such as Unixware, where some states can take 10 minutes or so to expire.
This is quite different from allowing two active listeners on a single port. I can't see a good reason for this. Which program accepts the connection?
In any case, this behavior is all handled by the things behind the Winsock API, not by Cygwin. I suspect that the Cygwin API is very thin at this point, just passing calls straight through to Winsock. This difference of interpretation of the BSD sockets API is not Cygwin's problem.
-- Warren Young, maintainer of the Winsock Programmer's FAQ http://tangentsoft.net/wskfaq/
-- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
