最近被多线程给坑了下，没意识到类变量在多线程下是共享的，还有一个就是没意识到 内存释放问题，导致越累越大

1.python 类变量 在多线程情况 下的 是共享的

2.python 类变量 在多线程情况 下的 释放是不完全的

3.python 类变量 在多线程情况 下没释放的那部分 内存 是可以重复利用的

import threading import time class Test: cache = {} @classmethod def get_value(self, key): value = Test.cache.get(key, []) return len(value) @classmethod def store_value(self, key, value): if not Test.cache.has_key(key): Test.cache[key] = range(value) else: Test.cache[key].extend(range(value)) return len(Test.cache[key]) @classmethod def release_value(self, key): if Test.cache.has_key(key): Test.cache.pop(key) return True @classmethod def print_cache(self): print 'print_cache:' for key in Test.cache: print 'key: %d, value:%d' % (key, len(Test.cache[key])) def worker(number, value): key = number % 5 print 'threading: %d, store_value: %d' % (number, Test.store_value(key, value)) time.sleep(10) print 'threading: %d, release_value: %s' % (number, Test.release_value(key)) if __name__ == '__main__': thread_num = 10 thread_pool = [] for i in range(thread_num): th = threading.Thread(target=worker,args=[i, 1000000]) thread_pool.append(th) thread_pool[i].start() for thread in thread_pool: threading.Thread.join(thread) Test.print_cache() time.sleep(10) thread_pool = [] for i in range(thread_num): th = threading.Thread(target=worker,args=[i, 100000]) thread_pool.append(th) thread_pool[i].start() for thread in thread_pool: threading.Thread.join(thread) Test.print_cache() time.sleep(10)

总结

公用的数据，除非是只读的，不然不要当类成员变量，一是会共享，二是不好释放。