Ticket #99 (defect)

On shutdown, the server tries to finish all CherryPy server threads, by signaling them. However, if the application has other "user-defined" threads, the code fails.

2005/01/26 10:02:11 HTTP INFO <Ctrl-C> hit: shutting down
Unhandled exception in thread started by <bound method Thread.__bootstrap of <Thread(expireCache, stopped daemon)>>
Traceback (most recent call last):
  File "/usr/lib/python2.3/threading.py", line 451, in __bootstrap
    self.__stop()
  File "/usr/lib/python2.3/threading.py", line 460, in __stop
    self.__block.notifyAll()
  File "/usr/lib/python2.3/threading.py", line 256, in notifyAll
    self.notify(len(self.__waiters))
  File "/usr/lib/python2.3/threading.py", line 238, in notify
    currentThread() # for side-effect
TypeError: 'NoneType' object is not callable

In the example above, there is one extra thread that is started by the CacheFilter?. It's a single thread, and it's set to run as a Daemon, so the server shutdown code doesn't need to worry about it. Neverthless, it tries to signal it and fail. In this case, the user thread is a single callable, and it does not support the same signaling protocol that is used by CherryPy itself to do the shutdown.

I believe that there are a few solutions for this problem:

• Server shutdown code doesn't have to worry about any other threads that are not CP core ones, and specially, 'daemonic' ones.
• Alternatively, the server shutdown code may try to signal all threads that 'conform' to the correct protocol, or that check the correct events.
• CherryPy can require that all user threads support a standard shutdown signaling protocols. (I believe that this is impossible to enforce in practice, though).