AT

• a timeout can be specified in milliseconds
• an appropriated maximum timeout exists and an exception is thrown if exceeded
• if there are available tasks, the response is performed immediately
• if there are currently no available tasks, the request is added to the long polling thread
• the long polling thread...
  • is a singleton, so all requests are handled inside this thread instance
  • falls asleep for an appropriated period of time if there are no requests pending
  • is woken up by a new request