Re: [Live-devel] live555 taskSchedule thread was hung

2013-06-25 Thread Ross Finlayson
The existing code is correct. The reason for this code is to ensure that sending a (RTP or RTCP) packet over a TCP connection - which involves two separate writes: one for a header; another for the packet data itself - occurs as an atomic operation. I.e., we want either neither of the writes t

[Live-devel] live555 taskSchedule thread was hung

2013-06-25 Thread 谢永达
As my testing, the function RTPInterface::sendRTPorRTCPPacketOverTCP may hung taskSchedule's thread, I found it was at source code line 307, and I modify it from " if (!sendDataOverTCP(socketNum, packet, packetSize, True)) break;" to "if (!sendDataOverTCP(socketNum, packet, packetSize, F