Re: [Live-devel] Late binding fix in 2010.01.15 version not yielding video...

2010-01-22 Thread Ross Finlayson
OK, I have now installed a new version (2010.01.22) of the software that should fix this problem. Note to Rob Krakora and Alex Nemec: Please download and test this, to see whether it works OK (and still avoids getting any "select()" errors). I can't test it myself. Important note: As a res

Re: [Live-devel] Late binding fix in 2010.01.15 version not yielding video...

2010-01-21 Thread Robert Krakora
Ross: No multicast packets are received with the 2010-01-15 version for which you provided the fix. I looked at the TCP/IP stack source and it appears that once a socket is bound it cannot be rebound. As a test I commented out the 'bind' code block in the setupDatagramSocket() function in the "G

Re: [Live-devel] Late binding fix in 2010.01.15 version not yielding video...

2010-01-21 Thread Robert Krakora
My bad, SO_REUSEADDR allows binding to an address that is in a TIME_WAIT state while SO_REUSEPORT allows multiple processes to bind to the same address. This has nothing to do with rebinding...I will keep looking... On Thu, Jan 21, 2010 at 10:58 AM, Robert Krakora wrote: > Ross: > > It looks li

Re: [Live-devel] Late binding fix in 2010.01.15 version not yielding video...

2010-01-21 Thread Robert Krakora
Ross: It looks like kernel 2.6.18-128.1.10.el5 utilized in CentOS 5.3 does not implement SO_REUSEPORT (see commented out SO_REUSEPORT below from "socket.h" file for this kernel and the comment about it's needed implementation in *bold red*). Otherwise I believe your fix would have worked. I did

Re: [Live-devel] Late binding fix in 2010.01.15 version not yielding video...

2010-01-21 Thread Robert Krakora
Ross: I will debug this and let you know what is going on. Multicast data was arriving per Wireshark. However, I did not capture the RTSP traffic...I cannot see the video as the location is remote. It was bad judgement on my part to indicate to you that it was working prematurely. I will have

Re: [Live-devel] Late binding fix in 2010.01.15 version not yielding video...

2010-01-20 Thread Ross Finlayson
The fix with which you provided me for the multicast destination address embedded in the SETUP response does not appear to be yielding video. Well, beforehand you told me that it worked OK! Anyway, the fix was to reimplement the "Socket::changePort()" function (in "groupsock/NetInterface.cpp"

[Live-devel] Late binding fix in 2010.01.15 version not yielding video...

2010-01-20 Thread Robert Krakora
Ross: The fix with which you provided me for the multicast destination address embedded in the SETUP response does not appear to be yielding video. The RTSP server and clients and hence the camera that I am testing against are at a remote location where nobody was available until today. Since I