>We have a multithreaded application (using live555 / linux) I hope you've read the FAQ about threads.
>Under circumstances, some of our customers want to use for their video >sources two different multicast addresses (let's say 239.61.87.46 and >239.61.87.46) FWIW, those are the same multicast address :-) >, but the same port (say 8000) This is a bad idea, as you discovered. >Unfortunately, when reading the two RTP streams, live555 seems to mix >the two streams into one. This is usually an operating system problem. Some OSs do not handle this (different multicast addresses using the same port number) properly. That's why it's a bad idea to use the same port number with different multicast addresses. >The solution is not to bind with INADDR_ANY, but with the desired >multicast address: > > // Whereas following works: socket will only see datagrams from >'multicast_addr' > saddr.sin_addr.s_addr = inet_addr(multicast_addr); > status = bind(sock, (struct sockaddr *)&saddr, sizeof(struct >sockaddr_in)); No, that's not the way bind() is supposed to work (if it works at all for you, then it's only by accident, I think). I've never before heard of anyone (successfully) calling bind() on a multicast address. >By default, ReceivingInterfaceAddr is INADDR_ANY: that's why we get the >two video streams mixed on the video output. >The problem is that ReceivingInterfaceAddr is a global variable. We >would need here one different value (i.e. multicast address) for each >reader. No, if you set ReceivingInterfaceAddr at all, then it needs to be set to a *unicast* address - the IP address of the desired network interface, not a multicast address. In any case, you probably should not be running the code multi-threaded within a single address space. Instead, run multiple address spaces (i.e., one per receiver). Again, please read the FAQ on threads. -- Ross Finlayson Live Networks, Inc. http://www.live555.com/ _______________________________________________ live-devel mailing list live-devel@lists.live555.com http://lists.live555.com/mailman/listinfo/live-devel
