Hi, The fixes the regression we were seeing in mauve gnu.testlet.java.net.ServerSocket.AcceptTimeout. After a socket is accepted through a server socket we must explicitly reset the timeout. Since we are doing this through VMChannel now the easiest way was to do this through the java interface. That is slightly less efficient, but the easiest and cleanest solution.
2006-11-26 Mark Wielaard <[EMAIL PROTECTED]> * gnu/java/net/PlainSocketImpl.java (accept): Reset timeout on Socket. Committed, Mark diff -u -r1.15 PlainSocketImpl.java --- gnu/java/net/PlainSocketImpl.java 25 Oct 2006 00:33:26 -0000 1.15 +++ gnu/java/net/PlainSocketImpl.java 26 Nov 2006 21:27:04 -0000 @@ -342,6 +342,9 @@ that.impl.getState().setChannelFD(c.getState()); that.channel = new SocketChannelImpl(c); that.setOption(SO_REUSEADDR, Boolean.TRUE); + // Reset the inherited timeout. + that.setOption(SO_TIMEOUT, Integer.valueOf(0)); + } /**
signature.asc
Description: This is a digitally signed message part