Calling a super class constructor when inheriting from Thread

Question

I'm curious why I can't call super(Thread, self).__init__() instead of Thread.__init__(self) when my class inherits from Thread. Can you help me to understand the problem?

#!/usr/bin/python

from threading import Thread
from Queue import Queue
class ThreadManager(object):

    def work(self, items):
        q = Queue()
        for item in items:
            q.put(item)

        print q.qsize()

        p = Worker()
        p.start()
        p.join()

class Worker(Thread):
    def __init__(self):
        # Why doesn't this work?
        #super(Thread, self).__init__()

        Thread.__init__(self)

    def run(self):
        print 'thread running'

def main():
    items = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i']
    tm = ThreadManager()
    tm.work(items)

if __name__ == "__main__":
    main()

Answer 1

I'm curious why I can't call super(Thread, self).__init__() instead of Thread.__init__(self) when my class inherits from Thread .

Because that's not how super works. You have to pass your own type as the first parameter, so it can search for the next ancestor of that type. If you pass it Thread , you're asking for an ancestor of Thread .

If your parent class is a regular new-style Python class, doing this wrong usually means you skip over one ancestor class, which may be harmless, or may seem to work but not actually do the right thing. But threading.Thread has specific checks to make sure it gets initialized properly, so you're probably getting something like this:

AssertionError: Thread.__init__() was not called

If your parent class is a C extension class, it probably doesn't have any ancestors, and it probably doesn't implement super even if it does, so you'll usually get an error that way too.

You may want to read Python's super() considered super if you want to understand how all of this works (because the docs linked above aren't necessarily the best introductory discussion).

So, in summary:

super(Worker, self).__init__()