Mark RubySingleThreadExecutor as a SerialExecutorService

[meineerde]

This resolves #1069.

However, I'm still not 100% sure if the RubySingleThreadExecutor is actually serialized. From my understanding of the code, it is, but there may be edge-cases I'm missing (such as when the thread dies).

As such, I'd still appreciate if someone who knows the (intended) RubyThreadPoolExecutor semantics to review this.

[bensheldon]

The old implementation doesn't look substantially different than the new one does. Both call Queue#pop and rescue from Exception, albeit the old implementation would swallow the exception whereas the new one will create a new worker thread; that doesn't change the behavior in my eyes.

Old implementation:

concurrent-ruby/lib/concurrent/executor/ruby_single_thread_executor.rb

Lines 66 to 78 in 7c2909f

	def work
	loop do
	task = @queue.pop
	break if task == :stop
	begin
	task.last.call(*task.first)
	rescue => ex
	# let it fail
	log DEBUG, ex
	end
	end
	stopped_event.set
	end

Current implementation:

concurrent-ruby/lib/concurrent-ruby/concurrent/executor/ruby_thread_pool_executor.rb

Lines 338 to 356 in eae2851

	def create_worker(queue, pool, idletime)
	Thread.new(queue, pool, idletime) do |my_queue, my_pool, my_idletime|
	catch(:stop) do
	loop do
	case message = my_queue.pop
	when :stop
	my_pool.remove_busy_worker(self)
	throw :stop
	else
	task, args = message
	run_task my_pool, task, args
	my_pool.ready_worker(self, Concurrent.monotonic_time)
	end
	end
	end
	end
	end

I'll leave this PR open a little longer to see if anyone else has thoughts, otherwise I can accept it.

[bensheldon]

I should also note I didn't see any test assertions of this behavior. The only place I saw #serialized? meaningfully change test behavior is right here, though I honestly think that the test is wrong and should actually be omitting on immediacy not serialized.

concurrent-ruby/spec/concurrent/executor/executor_service_shared.rb

Lines 202 to 211 in eae2851

	it 'returns false when shutdown fails to complete before timeout' do
	unless subject.serialized?
	latch = Concurrent::CountDownLatch.new 1
	100.times{ subject.post{ latch.wait } }
	sleep(0.1)
	subject.shutdown
	expect(subject.wait_for_termination(0.01)).to be_falsey
	latch.count_down
	end
	end

[meineerde]

That's what I saw too when checking if/how to adapt tests. On the other hand, the serialized? method also doesn't seem to be used that much anywhere else either.

To be honest, my original intend for my question in #1069 was to ensure that the semantics of a RubySingleThreadExecutor were in fact to serialize tasks as I want to reply on that. The correct return value of the serialized? method is then a nice bonus on top :)

Still, if this holds, it may be worthwhile to update the RubySingleThreadExecutor itself to implement serialized? as something like

def serialized?
  @max_length <= 1
end

I believe the same should be valid for the JavaThreadPoolExecutor, but I'm less familiar with the exact semantics of those. The documentation however seems to imply this.