Hi,

Below is a little ML code that prints out everything that gets sent to port 8080. On linux this behaves much as expected, but on MacOS, as soon as the Socket.accept call is made, it uses 100% of processor.

The code:

(*

Reproducable on Mac OS X 10.6.5 PolyML 5.4 Not reproducable on Linux :)

After calling main() the CPU usage becomes 100%. After connecting to this server socket by doing telnet localhost 8080 the CPU usage is still 100%.

*)

fun mkServerSocket () = let val server = INetSock.TCP.socket(); val _ = Socket.bind(server, INetSock.any 8080); val _ = Socket.Ctl.setREUSEADDR(server,true); val saddr = INetSock.fromAddr(Socket.Ctl.getSockName server); val _ = Socket.listen(server,128); in (saddr,server) end;

fun readLoop active_socket = let val s = Byte.bytesToString(Socket.recvVec(active_socket,80)); val _ = PolyML.print s; (* print to output *) in (PolyML.print "about to read...\n"; readLoop active_socket) end;

fun main() = let val (saddr, server_socket) = mkServerSocket () val (active_socket, active_socket_addr) = Socket.accept server_socket in readLoop active_socket end;