python线程池ThreadPoolExecutor用法

线程池，进程池

python的多线程并不是完全鸡肋的存在，得分情况来看。在IO密集型任务下，能提高多倍效率。在CPU密集型任务下，使用多进程也能规避GIL锁。

python3标准库concurrent.futures比原Thread封装更高，多线程concurrent.futures.ThreadPoolExecutor，多进程concurrent.futures.ProcessPoolExecutor

利用concurrent.futures.Future来进行各种便捷的数据交互，包括处理异常，都在result()中再次抛出。

模板

import time
from concurrent import futures
from concurrent.futures import ThreadPoolExecutor

def display(args):
    print(time.strftime('[%H:%M:%S]', time.localtime()), end=' ')
    print(args)

def task(n):
    """只是休眠"""
    display('begin sleep {}s.'.format(n))
    time.sleep(n)
    display('ended sleep {}s.'.format(n))

def do_many_task_inorder():
    """多线程
    按任务发布顺序依次等待完成
    """
    tasks = [5, 4, 3, 2, 1]
    with ThreadPoolExecutor(max_workers=3) as executor:
        future_list = [executor.submit(task, arg) for arg in tasks]

        display('非阻塞运行')

        for future in future_list:
            display(future)

        display('统一结束(有序)')

        for future in future_list:
            display(future.result())

def do_many_task_disorder():
    """多线程执行
    先完成先显示
    """
    tasks = [5, 4, 3, 2, 1]
    with ThreadPoolExecutor(max_workers=3) as executor:
        future_list = [executor.submit(task, arg) for arg in tasks]

        display('非阻塞运行')

        for future in future_list:
            display(future)

        display('统一结束(无序)')

        done_iter = futures.as_completed(future_list)  # generator

        for done in done_iter:
            display(done)

if __name__ == '__main__':
    do_many_task_inorder()
    do_many_task_disorder()