Recherche avancée

Sur d’autres sites (9294)

• Fixing "RTP : dropping old packet received too late" in FFMPEG

  25 juin 2014, par user985030

  I have been using FFMPEG 0.6 for years with no problems and recently ported much of my code to 2.2 ; however, there is still a problem that I cannot resolve after fixing many of the deprecated functions. I am generating simulated video and then using RTSP to unicast this generated stream. The problem is that when I change the height and width of my video data, I basically recreate a new stream to send the subscribed client. The algorithm I used to do this in 0.6 worked like a charm, never had any problems. Now that I have upgraded, I get "RTP : dropping old packet received too late" as soon as I change my frame size. I think I have been dropping packets all along, but the new code is causing connection issues for me. The packets being dropped in the past were negligible and I really didn’t care if I missed them as long as the stream eventually corrected itself. Is there a flag that I can set to not drop these packets ? Or at least recover more quickly ? I believe that this has something to do with receiving packets out of order. There is a section of code that does a diff in FFMPEG in the rtpdec.c file in the function rtp_parts_one_packet. The only reference I found similar to my issue is here :

  http://en.it-usenet.org/thread/16949/6708/#post6707. Any tips would be greatly appreciated. In the meantime, I am just going to patch the FFMPEG code to not do the following check :

  if (diff < 0) {
  
  
  
  /* Packet older than the previously emitted one, drop */
  
  
  
  av_log(s->st ? s->st->codec : NULL,        AV_LOG_WARNING,
  
  
  
  "RTP: dropping old packet received too late\n");
  
  
  
  return -1;
  
  
  
  }

  By commenting out the above code, I am able to run my streaming application like I used to, but I have a feeling that I am not doing something correctly but I am not sure what it is. Thanks for any advice !

• Revision d203203cc5 : Merge "Fix out of boundary memory read in fuzz test on vpxdec"

  16 juin 2014, par Jingning Han

  Changed Paths :
   Modify /test/test-data.sha1

  
   Modify /test/test.mk

  
   Modify /vp9/vp9_dx_iface.c

  
  
  Merge "Fix out of boundary memory read in fuzz test on vpxdec"

• Python ThreadedTCPServer : "Name or service not known"

  11 avril 2014, par Hal

  I was developing a ThreadedTCPServer to communicate with a PHP application also residing in this same machine. This is suppose to receive requests from this PHP app and to convert some videos locally using ffmpeg.

  Here's the code :

  # -*- coding: utf-8 -*- 
  
  import os
  
  import socket
  
  import threading
  
  import logging.config
  
  import SocketServer, time
  
  from queuev2 import QueueServer
  
  
  
  logging.basicConfig(format='[%(asctime)s.%(msecs).03d] %(message)s', datefmt='%Y-%m-%d %H:%M:%S', filename=os.path.join(os.path.dirname(os.path.realpath(__file__)), 'converter.log'), level=logging.INFO)
  
  
  
  class ThreadedTCPRequestHandler(SocketServer.BaseRequestHandler):
  
  
  
      def handle(self):
  
          data = self.request.recv(1024)
  
          cur_thread = threading.current_thread()
  
          response = "{}: {}".format(cur_thread.name, data)
  
          videoPool.add(data)
  
          print "Output! %s" % data
  
          self.request.sendall(response)
  
  
  
  class ThreadedTCPServer(SocketServer.ThreadingMixIn, SocketServer.TCPServer):
  
      pass
  
  
  
  if __name__ == "__main__":
  
  
  
      logging.info("Initializing...")
  
      videoPool = QueueServer()
  
      HOST, PORT = "localhost", 6666
  
  
  
      server = ThreadedTCPServer((HOST, PORT), ThreadedTCPRequestHandler)
  
      ip, port = server.server_address
  
  
  
      # Start a thread with the server -- that thread will then start one
  
      # more thread for each request
  
      server_thread = threading.Thread(target=server.serve_forever)
  
  
  
      # Exit the server thread when the main thread terminates
  
      server_thread.daemon = True
  
      server_thread.start()
  
  
  
      print("Server loop running in thread: %s" % server_thread.name)
  
  
  
      # "Groundhog day" time
  
      while True:
  
          time.sleep(999)
  
          pass
  
  
  
      #server.shutdown()

  This works well in my development laptop, but on the server i'm getting the following error :

  Traceback (most recent call last):
  
    File "server.py", line 31, in <module>
  
      server = ThreadedTCPServer((HOST, PORT), ThreadedTCPRequestHandler)
  
    File "/usr/lib/python2.7/SocketServer.py", line 408, in __init__
  
      self.server_bind()
  
    File "/usr/lib/python2.7/SocketServer.py", line 419, in server_bind
  
      self.socket.bind(self.server_address)
  
    File "/usr/lib/python2.7/socket.py", line 224, in meth
  
      return getattr(self._sock,name)(*args)
  
  socket.gaierror: [Errno -2] Name or service not known
  
  </module>

  I'm guessing it has to do with the port I'm using (6666), but I've tried others and it hasn't been working. Would Unix Domain Sockets be of use here ? Can you give me an example ?