之前用python的多线程，总是处理不好进程和线程之间的关系。后来发现了join和setdaemon函数，才终于弄明白。下面总结一下。
1.使用join函数后，主进程会在调用join的地方等待子线程结束，然后才接着往下执行。
join使用实例如下：
import time
import random
import threading
class worker(threading.thread):
def __init__(self):
threading.thread.__init__(self)
def run(self):
t = random.randint(1,10)
time.sleep(t)
print "this is " self.getname() ";i sleep %d second."%(t)
tsk = []
for i in xrange(0,5):
time.sleep(0.1)
thread = worker()
thread.start()
tsk.append(thread)
for tt in tsk:
tt.join()
print "this is the end of main thread."
运行结果如下：
# python testjoin.py this is thread-3;i sleep 2 second. this is thread-1;i sleep 4 second. this is thread-2;i sleep 7 second. this is thread-4;i sleep 7 second. this is thread-5;i sleep 7 second. this is the end of main thread.
这里创建了5个子线程，每个线程随机等待1-10秒后打印退出；主线程分别等待5个子线程结束。最后结果是先显示各个子线程，再显示主进程的结果。
2. 如果使用的setdaemon函数，则与join相反，主进程结束的时候不会等待子线程。
setdaemon函数使用实例：
import time
import random
import threading
class worker(threading.thread):
def __init__(self):
threading.thread.__init__(self)
def run(self):
t = random.randint(1,10)
time.sleep(t)
print "this is " self.getname() ";i sleep %d second."%(t)
tsk = []
for i in xrange(0,5):
time.sleep(0.1)
thread = worker()
thread.setdaemon(true)
thread.start()
tsk.append(thread)
print "this is the end of main thread."
这里设置主进程为守护进程，当主进程结束的时候，子线程被中止
运行结果如下：
#python testsetdaemon.py
this is the end of main thread.
3、如果没有使用join和setdaemon函数，则主进程在创建子线程后，直接运行后面的代码，主程序一直挂起，直到子线程结束才能结束。
import time
import random
import threading
class worker(threading.thread):
def __init__(self):
threading.thread.__init__(self)
def run(self):
t = random.randint(1,10)
time.sleep(t)
print "this is " self.getname() ";i sleep %d second."%(t)
tsk = []
for i in xrange(0,5):
time.sleep(0.1)
thread = worker()
thread.start()
tsk.append(thread)
print "this is the end of main thread."
运行结果如下：
# python testthread.py this is the end of main thread. this is thread-4;i sleep 1 second. this is thread-3;i sleep 7 second. this is thread-5;i sleep 7 second. this is thread-1;i sleep 10 second. this is thread-2;i sleep 10 second.
补充知识：python thread和process对比
原因：进程和线程的差距（方向不同，之针对这个实例）
# coding=utf-8
import logging
import multiprocessing
import os
import time
from threading import thread
logging.basicconfig(
level=logging.info,
format="%(asctime)s 【 %(process)d 】 %(processname)s %(message)s"
)
def func (i):
# logging.info(f'子：{os.getpid()},t{i}')
return f'子：{os.getpid()},t{i}'
def main (ctx):
start01 = time.time()
ts = [thread(target=func, args=(i,)) for i in range(100)]
[t.start() for t in ts]
[t.join() for t in ts]
end01 = time.time() - start01
logging.info(f"线程花费的时间：{end01}秒")
start02 = time.time()
ps = [ctx.process(target=func, args=(i,)) for i in range(100)]
[p.start() for p in ps]
[p.join() for p in ps]
end02 = time.time() - start02
logging.info(f"进程花费的时间：{end02}秒")
if __name__ == '__main__':
# windows 启动方式
multiprocessing.set_start_method('spawn')
# 获取上下文
ctx = multiprocessing.get_context('spawn')
# 检查这是否是冻结的可执行文件中的伪分支进程。
ctx.freeze_support()
main(ctx)
输出：
2019-10-06 14:17:22,729 【 7412 】 mainprocess 线程花费的时间：0.012967586517333984秒
2019-10-06 14:17:25,671 【 7412 】 mainprocess 进程花费的时间：2.9418249130249023秒
