Merge branch 'cleanup' into async

Byron · Byron · commit 29eb123beb1c · 2010-06-11T16:47:40.000+02:00
diff --git a/lib/git/async/__init__.py b/lib/git/async/__init__.py
@@ -1 +1,30 @@
 """Initialize the multi-processing package"""
+
+#{ Initialization
+def _init_atexit():
+	"""Setup an at-exit job to be sure our workers are shutdown correctly before
+	the interpreter quits"""
+	import atexit
+	import thread
+	atexit.register(thread.do_terminate_threads)
+	
+def _init_signals():
+	"""Assure we shutdown our threads correctly when being interrupted"""
+	import signal
+	import thread
+	
+	prev_handler = signal.getsignal(signal.SIGINT)
+	def thread_interrupt_handler(signum, frame):
+		thread.do_terminate_threads()
+		if callable(prev_handler):
+			prev_handler(signum, frame)
+			raise KeyboardInterrupt()
+		# END call previous handler
+	# END signal handler
+	signal.signal(signal.SIGINT, thread_interrupt_handler)
+
+
+#} END init
+
+_init_atexit()
+_init_signals()
diff --git a/lib/git/async/pool.py b/lib/git/async/pool.py
@@ -95,10 +95,6 @@ def read(self, count=0, block=True, timeout=None):
 		# provided enough - its better to have some possibly empty task runs 
 		# than having and empty queue that blocks.
 		
-		# NOTE: TODO: that case is only possible if one Task could be connected 
-		# to multiple input channels in a manner known by the system. Currently
-		# this is not possible, but should be implemented at some point.
-		
 		# if the user tries to use us to read from a done task, we will never 
 		# compute as all produced items are already in the channel
 		task = self._task_ref()
@@ -260,8 +256,6 @@ def _prepare_channel_read(self, task, count):
 			# the following loops are kind of unrolled - code duplication
 			# should make things execute faster. Putting the if statements 
 			# into the loop would be less code, but ... slower
-			# DEBUG
-			# print actual_count, numchunks, chunksize, remainder, task._out_writer.size()
 			if self._num_workers:
 				# respect the chunk size, and split the task up if we want 
 				# to process too much. This can be defined per task
diff --git a/lib/git/async/task.py b/lib/git/async/task.py
@@ -88,11 +88,7 @@ def process(self, count=0):
 		self._num_writers += 1
 		self._wlock.release()
 		
-		#print "%r: reading %i" % (self.id, count)
-		#if hasattr(self, 'reader'):
-		#	print "from", self.reader().channel
 		items = self._read(count)
-		#print "%r: done reading %i items" % (self.id, len(items))
 		
 		try:
 			try:
@@ -117,7 +113,6 @@ def process(self, count=0):
 				self._wlock.release()
 			# END handle writer count
 		except Exception, e:
-			print >> sys.stderr, "task %s error:" % self.id, type(e), str(e)	# TODO: REMOVE DEBUG, or make it use logging
 			# be sure our task is not scheduled again
 			self.set_done()
 			
@@ -164,7 +159,6 @@ def process(self, count=0):
 			self._wlock.acquire()
 			try:
 				if self._num_writers == 0:
-					# print "Closing channel of %r" % self.id, len(self._out_writer.channel.queue), self._out_writer.channel
 					self.close()
 				# END handle writers
 			finally:
diff --git a/lib/git/async/thread.py b/lib/git/async/thread.py
@@ -136,16 +136,21 @@ def stop(cls, *args):
 	
 	def run(self):
 		"""Process input tasks until we receive the quit signal"""
-		print self.name, "starts processing"	# DEBUG
-		
 		gettask = self.inq.get
 		while True:
 			if self._should_terminate():
 				break
 			# END check for stop request
 			
-			# we wait and block - to terminate, send the 'stop' method
+			# note: during shutdown, this turns None in the middle of waiting 
+			# for an item to be put onto it - we can't du anything about it - 
+			# even if we catch everything and break gracefully, the parent 
+			# call will think we failed with an empty exception.
+			# Hence we just don't do anything about it. Alternatively
+			# we could override the start method to get our own bootstrapping, 
+			# which would mean repeating plenty of code in of the threading module.
 			tasktuple = gettask()
+				
 			# needing exactly one function, and one arg
 			routine, arg = tasktuple
 			
@@ -161,7 +166,7 @@ def run(self):
 						rval = routine(arg)
 					else:
 						# ignore unknown items
-						print >> sys.stderr, "%s: task %s was not understood - terminating" % (self.getName(), str(tasktuple))
+						sys.stderr.write("%s: task %s was not understood - terminating\n" % (self.getName(), str(tasktuple)))
 						break
 					# END make routine call
 				finally:
@@ -171,18 +176,22 @@ def run(self):
 					del(routine)
 					del(tasktuple)
 			except StopProcessing:
-				print self.name, "stops processing"	# DEBUG
 				break
 			except Exception,e:
-				print >> sys.stderr, "%s: Task %s raised unhandled exception: %s - this really shouldn't happen !" % (self.getName(), str(tasktuple), str(e))
+				sys.stderr.write("%s: Task %s raised unhandled exception: %s - this really shouldn't happen !\n" % (self.getName(), str(tasktuple), str(e)))
 				continue	# just continue 
 			# END routine exception handling
 		
 			# END handle routine release
 		# END endless loop
 	
 	def stop_and_join(self):
-		"""Send stop message to ourselves"""
+		"""Send stop message to ourselves - we don't block, the thread will terminate 
+		once it has finished processing its input queue to receive our termination
+		event"""
+		# DONT call superclass as it will try to join - join's don't work for 
+		# some reason, as python apparently doesn't switch threads (so often)
+		# while waiting ... I don't know, but the threads respond properly, 
+		# but only if dear python switches to them
 		self.inq.put((self.stop, None))
-		super(WorkerThread, self).stop_and_join()
 #} END classes
diff --git a/lib/git/async/util.py b/lib/git/async/util.py
@@ -2,7 +2,6 @@
 
 from threading import (
 	Lock,
-	current_thread,
 	_allocate_lock,
 	_Condition, 
 	_sleep,
diff --git a/test/git/async/test_pool.py b/test/git/async/test_pool.py
@@ -19,7 +19,6 @@ class TestThreadPool(TestBase):
 	
 	def _assert_single_task(self, p, async=False):
 		"""Performs testing in a synchronized environment"""
-		# return	# DEBUG TODO: Fixme deactivated it
 		print >> sys.stderr, "Threadpool: Starting single task (async = %i) with %i threads" % (async, p.size())
 		null_tasks = p.num_tasks()		# in case we had some before
 		
@@ -373,10 +372,7 @@ def _assert_async_dependent_tasks(self, pool):
 		
 	
 	
-	
-	# for some reason, sometimes it has multiple workerthreads already when he 
-	# enters the method ... dunno yet, pools should clean up themselvess
-	#@terminate_threads
+	@terminate_threads
 	def test_base(self):
 		assert len(threading.enumerate()) == 1
 		
@@ -463,10 +459,11 @@ def test_base(self):
 		# threads per core
 		p.set_size(4)
 		self._assert_single_task(p, True)
+		
+		
 		# DEPENDENT TASK ASYNC MODE
 		###########################
 		self._assert_async_dependent_tasks(p)
 		
 		print >> sys.stderr, "Done with everything"
 		
-		# TODO: test multi-pool connections 
