问题
This is for Python 3.2.2. I'm only just learning how Python and multiprocessing work and this simple example has been tripping me:
from multiprocessing import Pipe, Process
def f(r):
print(r.recv())
if __name__ == '__main__':
q, r = Pipe()
p = Process(target=f, args=(r,))
p.start()
q.send([42, None, 'hello'])
p.join()
The main thread creates a new Process
, p
, and sends r
, a bidirectional connection object, to function f()
. When process p
is started, I expect r.recv()
to block (which, as far as I understand, means that this process will wait indefinitely until something comes through the pipe) until the main process sends some objects through with q.send
.
Then p.join()
should make the main process wait until p
has run its course.
But nothing whatsoever happens. If I add a print
statement to f()
, nothing happens there, either, as if f()
never even runs and p.start()
is nonfunctional.
Can you explain why this won't work and what might be fixed?
回答1:
I know it's been a while, but for others with this problem, you have the ends of your pipe reversed. You're trying to use the receiving end to send, and trying to receive with the sending end. I find that adding duplex=True to the Pipe constructor makes it much easier to deal with the different ends.
Source: https://docs.python.org/2/library/multiprocessing.html#pipes-and-queues
回答2:
From experience I've found that I can't print from processes that I've started. You could try reversing your program:
from multiprocessing import Pipe, Process
def f(r):
r.send([42, None, 'hello'])
if __name__ == '__main__':
q, r = Pipe()
p = Process(target=f, args=(r,))
p.start()
print(q.recv())
p.join()
来源:https://stackoverflow.com/questions/10442870/python-multiprocessing-pipe-will-not-recv-properly