23、Python并发编程：锁、竞态条件与全局解释器锁

Python并发编程：锁、竞态条件与全局解释器锁

1. 锁的弊端

1.1 并发程序变为顺序执行

我们来看下面这个 Python 代码示例，它用于比较并发程序和顺序程序的执行速度：

import threading
import random; random.seed(0)
import time

def update(pause_period):
    global counter
    with count_lock:
        current_counter = counter  # 读取共享资源
        time.sleep(pause_period)  # 模拟大量计算
        counter = current_counter + 1  # 更新共享资源

pause_periods = [random.randint(0, 1) for i in range(20)]

# 顺序版本
counter = 0
count_lock = threading.Lock()
start = time.perf_counter()
for i in range(20):
    update(pause_periods[i])
print('--Sequential version--')
print(f'Final counter: {counter}.')
print(f'Took {time.perf_counter() - start : .2f} seconds.')

# 并发版本
counter = 0
threads = [threading.Thread(target