参考文档:
http://www.rubyinside.com/does-the-gil-make-your-ruby-code-thread-safe-6051.html
http://www.jstorimer.com/blogs/workingwithcode/8085491-nobody-understands-the-gil
http://www.jstorimer.com/blogs/workingwithcode/8100871-nobody-understands-the-gil-part-2-implementation
http://stackoverflow.com/questions/2227169/are-python-built-in-containers-thread-safe
http://jessenoller.com/blog/2009/02/01/python-threads-and-the-global-interpreter-lock
https://wiki.python.org/moin/GlobalInterpreterLock
上面的TODO不是为了做thread safe的, add ability to block waiting on a connection to be released
,个人理解是connection在release的时候,其他方法被block。但是他的实现我没看懂,os.getpid()不是获取进程id吗?
pool正常调用顺序,get_conneciton(),release()
假设Thread A,B,那么可能出现的随机组合:(A.get_con,B.release),(A.get_con,B.get_con),(A.release,B.get_con),(A.release,B.release)
thread pool的thread safe关键点是:已经被线程A标记为使用的connection不能被B线程使用到。
如果能做到上面这点可以认为thread pool是线程安全的。
def get_connection(self):
try:
connection = self._available_connections.pop()
except IndexError:
connection = self._make_connection()
if connection and connection.connect():
self._in_use_connections.add(connection)
return connection
else:
return None
def _make_connection(self):
"Create a new connection"
if self._created_connections >= self._max_connections:
raise DbossError("Too many connections")
conn = Connection(self,self.location,self.timeout)
self._created_connections += 1
return conn
_available_connection有元素的情况:
如果_available_connection.pop()被线程A调用后,被线程B调用,线程B是否会得到和线程A一样的connection?这个是python的set来保证,据我所知,set是保证。
_available_connection没有元素的情况:
如果_available_connection.pop()被线程A调用后,被线程B调用,这个时候会触发IndexError,进入_make_connection。这里有两种情况:
1. 线程A已经执行完_make_connection(),线程B调用self._available_connections.pop()。
2. 线程A在_make_connection()执行过程中,线程B调用self._available_connections.pop()。
可以发现上述两种情况线程B都会触发IndexError异常,对结果没有影响。