早就听说gevent基于事件的异步处理能力 效率多么高，一直在项目中也很少用到，今天先来没事就学习了些简单的用法。

有个官方的教程写的很不错 中文版的地址为：http://xlambda.com/gevent-tutorial/ 学习gevent很不错的资料。

具体的理论这里不怎么说了，只是有些了解，具体的原理还不能解释的很清楚。不过协程这种概念在golang里面很多。

写了一个访问网络，使用同步操作，gevent 和 多线程对比的例子。

#!/usr/bin/python
# -*- coding: utf-8 -*-
# python2.7x
# gevent_urllib2.py
# author: orangelliu
# date: 2014-08-20

import gevent.monkey
gevent.monkey.patch_socket()

import gevent
import urllib2
import json
import threading

def fetch(pid):
	response = urllib2.urlopen('http://www.orangleliu.info')
	result = response.read()
	btypes = len(result)

	print 'process %s : %s'%(pid, btypes)

def synchronous():
	for i in range(10):
		fetch(i)

def asynchonous():
	threads = []
	for i in range(10):
		threads.append(gevent.spawn(fetch,i))
	gevent.joinall(threads)

def mulithread():
	threads = []
	for i in range(10):
		th = threading.Thread(target=fetch, args=(i,))
		threads.append(th)

	for thread in threads:
		thread.start()

	for thread in threads:
		threading.Thread.join(thread)

import time
print 'sync....'
ss = time.time()
synchronous()
print 'sync time is %s'%(time.time()-ss)

print 'async'
sa = time.time()
asynchonous()
print 'async time is %s'%(time.time()-sa)

print 'async'
sm = time.time()
mulithread()
print 'thread time is %s'%(time.time()-sm)

但是考虑到gevent的开销很小，所以还是很具有性价比的。

还有从结果中可以看到gevent和多线程都会有上下文切换，所以执行结果的线程id是乱序的，这个很好理解。

sync....
process 0 : 8657
process 1 : 8657
process 2 : 8657
process 3 : 8657
process 4 : 8657
process 5 : 8657
process 6 : 8657
process 7 : 8657
process 8 : 8657
process 9 : 8657
sync time is 2.7610001564
async
process 8 : 8657
process 7 : 8657
process 6 : 8657
process 2 : 8657
process 5 : 8657
process 3 : 8657
process 0 : 8657
process 4 : 8657
process 1 : 8657
process 9 : 8657
async time is 1.50199985504
async
process 0 : 8657
process 1 : 8657
process 3 : 8657
process 4 : 8657
process 5 : 8657
process 7 : 8657
process 9 : 8657
process 8 : 8657
process 6 : 8657
process 2 : 8657
thread time is 0.986000061035