Question

我有一个API侦听传入的请求并将其转储到列表中。在单独的过程中，每当追加列表时，我都想触发某种操作。我尝试实例化一个新进程（通过多重处理），但是一旦启动，它就不会更新数组的状态。

from multiprocessing import Process
import time

procs = []

def separateProcess(start, counter):
    while True:
        time.sleep(1)
        print("length of the list from separate Process: "+str(len(procs)))


if __name__ == '__main__':
    print("program started")

    counter = 0
    Process(target=separateProcess, args=(counter, counter)).start()
    print("program end")
    while True:
        counter += 1
        newstring = "string "+str(counter)
        procs.append(newstring)
        print("length of the list from main: " + str(len(procs)))
        time.sleep(2)

这是输出：

length of the list from main: 1
length of the list from separate Process: 0
length of the list from main: 2
length of the list from separate Process: 0
length of the list from separate Process: 0
length of the list from main: 3
length of the list from separate Process: 0
length of the list from separate Process: 0

Answer 1

创建一个新的子进程时，它将获得父进程地址空间的副本，但是随后的任何更改（无论是父进程还是子进程）都不会反映在另一个进程的内存中。他们每个人都有自己的专用地址空间。

您可以创建一个Manager()并使用其共享的list对象：

import time
  
from multiprocessing import Manager, Process

def separateProcess(start, counter):
    while True:
        time.sleep(1)
        print("length of the list from separate Process: "+str(len(procs)))


if __name__ == '__main__':
    m = Manager()
    procs = m.list()

    print("program started")

    counter = 0
    Process(target=separateProcess, args=(counter, counter)).start()
    print("program end")
    while True:
        counter += 1
        newstring = "string "+str(counter)
        procs.append(newstring)
        print("length of the list from main: " + str(len(procs)))
        time.sleep(2)

此方法会产生一些开销，因为它将产生一个子代进程来托管Manager服务器。

如果您可以调整工作进程逻辑以改为使用队列，请参见以下示例：

import random
import time

from multiprocessing import cpu_count, Process, Queue


def worker(q):
    for item in iter(q.get, 'STOP'):
        t = random.uniform(1, 5)
        print(f'START item: {item}')
        time.sleep(t)
        print(f'END item: {item}, ({t:.3f}s)')


def main():
    cpus = cpu_count()
    q = Queue()

    for i in range(5):
        q.put(i)

    for i in range(cpus):
        Process(target=worker, args=(q,)).start()

    for i in range(cpus):
        q.put('STOP')


if __name__ == '__main__':
    main()

聆听清单并在附加清单时做一些事情

1 个答案: