^^&&^^

Edit File: multiprocessing.html

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">

<html xmlns="http://www.w3.org/1999/xhtml">
 <head>
 <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
 
 <title>16.6. multiprocessing — Process-based “threading” interface &mdash; Python 2.7.5 documentation</title>
 
 <link rel="stylesheet" href="../_static/default.css" type="text/css" />
 <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
 
 <script type="text/javascript">
 var DOCUMENTATION_OPTIONS = {
 URL_ROOT: '../',
 VERSION: '2.7.5',
 COLLAPSE_INDEX: false,
 FILE_SUFFIX: '.html',
 HAS_SOURCE: true
 };
 </script>
 <script type="text/javascript" src="../_static/jquery.js"></script>
 <script type="text/javascript" src="../_static/underscore.js"></script>
 <script type="text/javascript" src="../_static/doctools.js"></script>
 <script type="text/javascript" src="../_static/sidebar.js"></script>
 <link rel="search" type="application/opensearchdescription+xml"
 title="Search within Python 2.7.5 documentation"
 href="../_static/opensearch.xml"/>
 <link rel="author" title="About these documents" href="../about.html" />
 <link rel="copyright" title="Copyright" href="../copyright.html" />
 <link rel="top" title="Python 2.7.5 documentation" href="../index.html" />
 <link rel="up" title="16. Optional Operating System Services" href="someos.html" />
 <link rel="next" title="16.7. mmap — Memory-mapped file support" href="mmap.html" />
 <link rel="prev" title="16.5. dummy_thread — Drop-in replacement for the thread module" href="dummy_thread.html" />
 <link rel="shortcut icon" type="image/png" href="../_static/py.png" />
 <script type="text/javascript" src="../_static/copybutton.js"></script>

</head>
 <body>
 <div class="related">
 <h3>Navigation</h3>
 <ul>
 <li class="right" style="margin-right: 10px">
 <a href="../genindex.html" title="General Index"
 accesskey="I">index</a></li>
 <li class="right" >
 <a href="../py-modindex.html" title="Python Module Index"
 >modules</a> |</li>
 <li class="right" >
 <a href="mmap.html" title="16.7. mmap — Memory-mapped file support"
 accesskey="N">next</a> |</li>
 <li class="right" >
 <a href="dummy_thread.html" title="16.5. dummy_thread — Drop-in replacement for the thread module"
 accesskey="P">previous</a> |</li>
 <li><img src="../_static/py.png" alt=""
 style="vertical-align: middle; margin-top: -1px"/></li>
 <li><a href="http://www.python.org/">Python</a> &raquo;</li>
 <li>
 <a href="../index.html">Python 2.7.5 documentation</a> &raquo;
 </li>

<li><a href="index.html" >The Python Standard Library</a> &raquo;</li>
 <li><a href="someos.html" accesskey="U">16. Optional Operating System Services</a> &raquo;</li> 
 </ul>
 </div>

<div class="document">
 <div class="documentwrapper">
 <div class="bodywrapper">
 <div class="body">
 
 <div class="section" id="module-multiprocessing">
<h1>16.6. <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> &#8212; Process-based &#8220;threading&#8221; interface<a class="headerlink" href="#module-multiprocessing" title="Permalink to this headline">¶</a></h1>

New in version 2.6.
<div class="section" id="introduction">
<h2>16.6.1. Introduction<a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
<a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> is a package that supports spawning processes using an
API similar to the <a class="reference internal" href="threading.html#module-threading" title="threading: Higher-level threading interface."><tt class="xref py py-mod docutils literal">threading</tt></a> module. The <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> package
offers both local and remote concurrency, effectively side-stepping the
<a class="reference internal" href="../glossary.html#term-global-interpreter-lock">Global Interpreter Lock</a> by using subprocesses instead of threads. Due
to this, the <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> module allows the programmer to fully
leverage multiple processors on a given machine. It runs on both Unix and
Windows.
<div class="admonition warning">
Warning
Some of this package&#8217;s functionality requires a functioning shared semaphore
implementation on the host operating system. Without one, the
<tt class="xref py py-mod docutils literal">multiprocessing.synchronize</tt> module will be disabled, and attempts to
import it will result in an <a class="reference internal" href="exceptions.html#exceptions.ImportError" title="exceptions.ImportError"><tt class="xref py py-exc docutils literal">ImportError</tt></a>. See
<a class="reference external" href="http://bugs.python.org/issue3770">issue 3770</a> for additional information.
</div>
<div class="admonition note">
Note
Functionality within this package requires that the <tt class="docutils literal">__main__</tt> module be
importable by the children. This is covered in <a class="reference internal" href="#multiprocessing-programming">Programming guidelines</a>
however it is worth pointing out here. This means that some examples, such
as the <tt class="xref py py-class docutils literal">multiprocessing.Pool</tt> examples will not work in the
interactive interpreter. For example:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing import Pool
&gt;&gt;&gt; p = Pool(5)
&gt;&gt;&gt; def f(x):
... return x*x
...
&gt;&gt;&gt; p.map(f, [1,2,3])
Process PoolWorker-1:
Process PoolWorker-2:
Process PoolWorker-3:
Traceback (most recent call last):
AttributeError: &#39;module&#39; object has no attribute &#39;f&#39;
AttributeError: &#39;module&#39; object has no attribute &#39;f&#39;
AttributeError: &#39;module&#39; object has no attribute &#39;f&#39;
</pre></div>
</div>
(If you try this it will actually output three full tracebacks
interleaved in a semi-random fashion, and then you may have to
stop the master process somehow.)
</div>
<div class="section" id="the-process-class">
<h3>16.6.1.1. The <a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a> class<a class="headerlink" href="#the-process-class" title="Permalink to this headline">¶</a></h3>
In <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a>, processes are spawned by creating a <a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a>
object and then calling its <a class="reference internal" href="#multiprocessing.Process.start" title="multiprocessing.Process.start"><tt class="xref py py-meth docutils literal">start()</tt></a> method. <a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a>
follows the API of <a class="reference internal" href="threading.html#threading.Thread" title="threading.Thread"><tt class="xref py py-class docutils literal">threading.Thread</tt></a>. A trivial example of a
multiprocess program is
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Process

def f(name):
 print &#39;hello&#39;, name

def info(title):
 print title
 print &#39;module name:&#39;, __name__
 if hasattr(os, &#39;getppid&#39;): # only available on Unix
 print &#39;parent process:&#39;, os.getppid()
 print &#39;process id:&#39;, os.getpid()

def f(name):
 info(&#39;function f&#39;)
 print &#39;hello&#39;, name

def f(q):
 q.put([42, None, &#39;hello&#39;])

def f(conn):
 conn.send([42, None, &#39;hello&#39;])
 conn.close()

def f(l, i):
 l.acquire()
 print &#39;hello world&#39;, i
 l.release()

for num in range(10):
 Process(target=f, args=(lock, num)).start()
</pre></div>
</div>
Without using the lock output from the different processes is liable to get all
mixed up.
</div>
<div class="section" id="sharing-state-between-processes">
<h3>16.6.1.4. Sharing state between processes<a class="headerlink" href="#sharing-state-between-processes" title="Permalink to this headline">¶</a></h3>
As mentioned above, when doing concurrent programming it is usually best to
avoid using shared state as far as possible. This is particularly true when
using multiple processes.
However, if you really do need to use some shared data then
<a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> provides a couple of ways of doing so.
Shared memory
<blockquote>
<div>Data can be stored in a shared memory map using <a class="reference internal" href="#multiprocessing.Value" title="multiprocessing.Value"><tt class="xref py py-class docutils literal">Value</tt></a> or
<a class="reference internal" href="#multiprocessing.Array" title="multiprocessing.Array"><tt class="xref py py-class docutils literal">Array</tt></a>. For example, the following code
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Process, Value, Array

def f(n, a):
 n.value = 3.1415927
 for i in range(len(a)):
 a[i] = -a[i]

print num.value
 print arr[:]
</pre></div>
</div>
will print
<div class="highlight-python"><div class="highlight"><pre>3.1415927
[0, -1, -2, -3, -4, -5, -6, -7, -8, -9]
</pre></div>
</div>
The <tt class="docutils literal">'d'</tt> and <tt class="docutils literal">'i'</tt> arguments used when creating <tt class="docutils literal">num</tt> and <tt class="docutils literal">arr</tt> are
typecodes of the kind used by the <a class="reference internal" href="array.html#module-array" title="array: Space efficient arrays of uniformly typed numeric values."><tt class="xref py py-mod docutils literal">array</tt></a> module: <tt class="docutils literal">'d'</tt> indicates a
double precision float and <tt class="docutils literal">'i'</tt> indicates a signed integer. These shared
objects will be process and thread-safe.
For more flexibility in using shared memory one can use the
<a class="reference internal" href="#module-multiprocessing.sharedctypes" title="multiprocessing.sharedctypes: Allocate ctypes objects from shared memory."><tt class="xref py py-mod docutils literal">multiprocessing.sharedctypes</tt></a> module which supports the creation of
arbitrary ctypes objects allocated from shared memory.
</div></blockquote>
Server process
<blockquote>
<div>A manager object returned by <tt class="xref py py-func docutils literal">Manager()</tt> controls a server process which
holds Python objects and allows other processes to manipulate them using
proxies.
A manager returned by <tt class="xref py py-func docutils literal">Manager()</tt> will support types <a class="reference internal" href="functions.html#list" title="list"><tt class="xref py py-class docutils literal">list</tt></a>,
<a class="reference internal" href="stdtypes.html#dict" title="dict"><tt class="xref py py-class docutils literal">dict</tt></a>, <tt class="xref py py-class docutils literal">Namespace</tt>, <a class="reference internal" href="#multiprocessing.Lock" title="multiprocessing.Lock"><tt class="xref py py-class docutils literal">Lock</tt></a>, <a class="reference internal" href="#multiprocessing.RLock" title="multiprocessing.RLock"><tt class="xref py py-class docutils literal">RLock</tt></a>,
<a class="reference internal" href="#multiprocessing.Semaphore" title="multiprocessing.Semaphore"><tt class="xref py py-class docutils literal">Semaphore</tt></a>, <a class="reference internal" href="#multiprocessing.BoundedSemaphore" title="multiprocessing.BoundedSemaphore"><tt class="xref py py-class docutils literal">BoundedSemaphore</tt></a>, <a class="reference internal" href="#multiprocessing.Condition" title="multiprocessing.Condition"><tt class="xref py py-class docutils literal">Condition</tt></a>,
<a class="reference internal" href="#multiprocessing.Event" title="multiprocessing.Event"><tt class="xref py py-class docutils literal">Event</tt></a>, <a class="reference internal" href="#multiprocessing.Queue" title="multiprocessing.Queue"><tt class="xref py py-class docutils literal">Queue</tt></a>, <a class="reference internal" href="#multiprocessing.Value" title="multiprocessing.Value"><tt class="xref py py-class docutils literal">Value</tt></a> and <a class="reference internal" href="#multiprocessing.Array" title="multiprocessing.Array"><tt class="xref py py-class docutils literal">Array</tt></a>. For
example,
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Process, Manager

def f(d, l):
 d[1] = &#39;1&#39;
 d[&#39;2&#39;] = 2
 d[0.25] = None
 l.reverse()

d = manager.dict()
 l = manager.list(range(10))

print d
 print l
</pre></div>
</div>
will print
<div class="highlight-python"><div class="highlight"><pre>{0.25: None, 1: &#39;1&#39;, &#39;2&#39;: 2}
[9, 8, 7, 6, 5, 4, 3, 2, 1, 0]
</pre></div>
</div>
Server process managers are more flexible than using shared memory objects
because they can be made to support arbitrary object types. Also, a single
manager can be shared by processes on different computers over a network.
They are, however, slower than using shared memory.
</div></blockquote>
</div>
<div class="section" id="using-a-pool-of-workers">
<h3>16.6.1.5. Using a pool of workers<a class="headerlink" href="#using-a-pool-of-workers" title="Permalink to this headline">¶</a></h3>
The <tt class="xref py py-class docutils literal">Pool</tt> class represents a pool of worker
processes. It has methods which allows tasks to be offloaded to the worker
processes in a few different ways.
For example:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Pool

<dl class="method">
<dt id="multiprocessing.Process.start">
<tt class="descname">start</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Process.start" title="Permalink to this definition">¶</a></dt>
<dd>Start the process&#8217;s activity.
This must be called at most once per process object. It arranges for the
object&#8217;s <a class="reference internal" href="#multiprocessing.Process.run" title="multiprocessing.Process.run"><tt class="xref py py-meth docutils literal">run()</tt></a> method to be invoked in a separate process.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Process.join">
<tt class="descname">join</tt><big>(</big>[timeout]<big>)</big><a class="headerlink" href="#multiprocessing.Process.join" title="Permalink to this definition">¶</a></dt>
<dd>Block the calling thread until the process whose <a class="reference internal" href="#multiprocessing.Process.join" title="multiprocessing.Process.join"><tt class="xref py py-meth docutils literal">join()</tt></a> method is
called terminates or until the optional timeout occurs.
If timeout is <tt class="docutils literal">None</tt> then there is no timeout.
A process can be joined many times.
A process cannot join itself because this would cause a deadlock. It is
an error to attempt to join a process before it has been started.
</dd></dl>

<dl class="attribute">
<dt id="multiprocessing.Process.name">
<tt class="descname">name</tt><a class="headerlink" href="#multiprocessing.Process.name" title="Permalink to this definition">¶</a></dt>
<dd>The process&#8217;s name.
The name is a string used for identification purposes only. It has no
semantics. Multiple processes may be given the same name. The initial
name is set by the constructor.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Process.is_alive">
<tt class="descname">is_alive</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Process.is_alive" title="Permalink to this definition">¶</a></dt>
<dd>Return whether the process is alive.
Roughly, a process object is alive from the moment the <a class="reference internal" href="#multiprocessing.Process.start" title="multiprocessing.Process.start"><tt class="xref py py-meth docutils literal">start()</tt></a>
method returns until the child process terminates.
</dd></dl>

<dl class="attribute">
<dt id="multiprocessing.Process.daemon">
<tt class="descname">daemon</tt><a class="headerlink" href="#multiprocessing.Process.daemon" title="Permalink to this definition">¶</a></dt>
<dd>The process&#8217;s daemon flag, a Boolean value. This must be set before
<a class="reference internal" href="#multiprocessing.Process.start" title="multiprocessing.Process.start"><tt class="xref py py-meth docutils literal">start()</tt></a> is called.
The initial value is inherited from the creating process.
When a process exits, it attempts to terminate all of its daemonic child
processes.
Note that a daemonic process is not allowed to create child processes.
Otherwise a daemonic process would leave its children orphaned if it gets
terminated when its parent process exits. Additionally, these are not
Unix daemons or services, they are normal processes that will be
terminated (and not joined) if non-daemonic processes have exited.
</dd></dl>

In addition to the <tt class="xref py py-class docutils literal">Threading.Thread</tt> API, <a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a> objects
also support the following attributes and methods:
<dl class="attribute">
<dt id="multiprocessing.Process.pid">
<tt class="descname">pid</tt><a class="headerlink" href="#multiprocessing.Process.pid" title="Permalink to this definition">¶</a></dt>
<dd>Return the process ID. Before the process is spawned, this will be
<tt class="docutils literal">None</tt>.
</dd></dl>

<dl class="attribute">
<dt id="multiprocessing.Process.exitcode">
<tt class="descname">exitcode</tt><a class="headerlink" href="#multiprocessing.Process.exitcode" title="Permalink to this definition">¶</a></dt>
<dd>The child&#8217;s exit code. This will be <tt class="docutils literal">None</tt> if the process has not yet
terminated. A negative value -N indicates that the child was terminated
by signal N.
</dd></dl>

<dl class="attribute">
<dt id="multiprocessing.Process.authkey">
<tt class="descname">authkey</tt><a class="headerlink" href="#multiprocessing.Process.authkey" title="Permalink to this definition">¶</a></dt>
<dd>The process&#8217;s authentication key (a byte string).
When <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> is initialized the main process is assigned a
random string using <tt class="xref py py-func docutils literal">os.random()</tt>.
When a <a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a> object is created, it will inherit the
authentication key of its parent process, although this may be changed by
setting <a class="reference internal" href="#multiprocessing.Process.authkey" title="multiprocessing.Process.authkey"><tt class="xref py py-attr docutils literal">authkey</tt></a> to another byte string.
See <a class="reference internal" href="#multiprocessing-auth-keys">Authentication keys</a>.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Process.terminate">
<tt class="descname">terminate</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Process.terminate" title="Permalink to this definition">¶</a></dt>
<dd>Terminate the process. On Unix this is done using the <tt class="docutils literal">SIGTERM</tt> signal;
on Windows <tt class="xref c c-func docutils literal">TerminateProcess()</tt> is used. Note that exit handlers and
finally clauses, etc., will not be executed.
Note that descendant processes of the process will not be terminated &#8211;
they will simply become orphaned.
<div class="admonition warning">
Warning
If this method is used when the associated process is using a pipe or
queue then the pipe or queue is liable to become corrupted and may
become unusable by other process. Similarly, if the process has
acquired a lock or semaphore etc. then terminating it is liable to
cause other processes to deadlock.
</div>
</dd></dl>

Note that the <a class="reference internal" href="#multiprocessing.Process.start" title="multiprocessing.Process.start"><tt class="xref py py-meth docutils literal">start()</tt></a>, <a class="reference internal" href="#multiprocessing.Process.join" title="multiprocessing.Process.join"><tt class="xref py py-meth docutils literal">join()</tt></a>, <a class="reference internal" href="#multiprocessing.Process.is_alive" title="multiprocessing.Process.is_alive"><tt class="xref py py-meth docutils literal">is_alive()</tt></a> and
<tt class="xref py py-attr docutils literal">exit_code</tt> methods should only be called by the process that created
the process object.
Example usage of some of the methods of <a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a>:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; import multiprocessing, time, signal
&gt;&gt;&gt; p = multiprocessing.Process(target=time.sleep, args=(1000,))
&gt;&gt;&gt; print p, p.is_alive()
&lt;Process(Process-1, initial)&gt; False
&gt;&gt;&gt; p.start()
&gt;&gt;&gt; print p, p.is_alive()
&lt;Process(Process-1, started)&gt; True
&gt;&gt;&gt; p.terminate()
&gt;&gt;&gt; time.sleep(0.1)
&gt;&gt;&gt; print p, p.is_alive()
&lt;Process(Process-1, stopped[SIGTERM])&gt; False
&gt;&gt;&gt; p.exitcode == -signal.SIGTERM
True
</pre></div>
</div>
</dd></dl>

<dl class="exception">
<dt id="multiprocessing.BufferTooShort">
exception <tt class="descclassname">multiprocessing.</tt><tt class="descname">BufferTooShort</tt><a class="headerlink" href="#multiprocessing.BufferTooShort" title="Permalink to this definition">¶</a></dt>
<dd>Exception raised by <a class="reference internal" href="#multiprocessing.Connection.recv_bytes_into" title="multiprocessing.Connection.recv_bytes_into"><tt class="xref py py-meth docutils literal">Connection.recv_bytes_into()</tt></a> when the supplied
buffer object is too small for the message read.
If <tt class="docutils literal">e</tt> is an instance of <a class="reference internal" href="#multiprocessing.BufferTooShort" title="multiprocessing.BufferTooShort"><tt class="xref py py-exc docutils literal">BufferTooShort</tt></a> then <tt class="docutils literal">e.args[0]</tt> will give
the message as a byte string.
</dd></dl>

</div>
<div class="section" id="pipes-and-queues">
<h3>16.6.2.2. Pipes and Queues<a class="headerlink" href="#pipes-and-queues" title="Permalink to this headline">¶</a></h3>
When using multiple processes, one generally uses message passing for
communication between processes and avoids having to use any synchronization
primitives like locks.
For passing messages one can use <a class="reference internal" href="#multiprocessing.Pipe" title="multiprocessing.Pipe"><tt class="xref py py-func docutils literal">Pipe()</tt></a> (for a connection between two
processes) or a queue (which allows multiple producers and consumers).
The <a class="reference internal" href="#multiprocessing.Queue" title="multiprocessing.Queue"><tt class="xref py py-class docutils literal">Queue</tt></a>, <a class="reference internal" href="#multiprocessing.multiprocessing.queues.SimpleQueue" title="multiprocessing.multiprocessing.queues.SimpleQueue"><tt class="xref py py-class docutils literal">multiprocessing.queues.SimpleQueue</tt></a> and <a class="reference internal" href="#multiprocessing.JoinableQueue" title="multiprocessing.JoinableQueue"><tt class="xref py py-class docutils literal">JoinableQueue</tt></a> types are multi-producer,
multi-consumer FIFO queues modelled on the <a class="reference internal" href="queue.html#Queue.Queue" title="Queue.Queue"><tt class="xref py py-class docutils literal">Queue.Queue</tt></a> class in the
standard library. They differ in that <a class="reference internal" href="#multiprocessing.Queue" title="multiprocessing.Queue"><tt class="xref py py-class docutils literal">Queue</tt></a> lacks the
<a class="reference internal" href="queue.html#Queue.Queue.task_done" title="Queue.Queue.task_done"><tt class="xref py py-meth docutils literal">task_done()</tt></a> and <a class="reference internal" href="queue.html#Queue.Queue.join" title="Queue.Queue.join"><tt class="xref py py-meth docutils literal">join()</tt></a> methods introduced
into Python 2.5&#8217;s <a class="reference internal" href="queue.html#Queue.Queue" title="Queue.Queue"><tt class="xref py py-class docutils literal">Queue.Queue</tt></a> class.
If you use <a class="reference internal" href="#multiprocessing.JoinableQueue" title="multiprocessing.JoinableQueue"><tt class="xref py py-class docutils literal">JoinableQueue</tt></a> then you must call
<a class="reference internal" href="#multiprocessing.JoinableQueue.task_done" title="multiprocessing.JoinableQueue.task_done"><tt class="xref py py-meth docutils literal">JoinableQueue.task_done()</tt></a> for each task removed from the queue or else the
semaphore used to count the number of unfinished tasks may eventually overflow,
raising an exception.
Note that one can also create a shared queue by using a manager object &#8211; see
<a class="reference internal" href="#multiprocessing-managers">Managers</a>.
<div class="admonition note">
Note
<a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> uses the usual <a class="reference internal" href="queue.html#Queue.Empty" title="Queue.Empty"><tt class="xref py py-exc docutils literal">Queue.Empty</tt></a> and
<a class="reference internal" href="queue.html#Queue.Full" title="Queue.Full"><tt class="xref py py-exc docutils literal">Queue.Full</tt></a> exceptions to signal a timeout. They are not available in
the <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> namespace so you need to import them from
<a class="reference internal" href="queue.html#module-Queue" title="Queue: A synchronized queue class."><tt class="xref py py-mod docutils literal">Queue</tt></a>.
</div>
<div class="admonition warning">
Warning
If a process is killed using <a class="reference internal" href="#multiprocessing.Process.terminate" title="multiprocessing.Process.terminate"><tt class="xref py py-meth docutils literal">Process.terminate()</tt></a> or <a class="reference internal" href="os.html#os.kill" title="os.kill"><tt class="xref py py-func docutils literal">os.kill()</tt></a>
while it is trying to use a <a class="reference internal" href="#multiprocessing.Queue" title="multiprocessing.Queue"><tt class="xref py py-class docutils literal">Queue</tt></a>, then the data in the queue is
likely to become corrupted. This may cause any other process to get an
exception when it tries to use the queue later on.
</div>
<div class="admonition warning">
Warning
As mentioned above, if a child process has put items on a queue (and it has
not used <tt class="xref py py-meth docutils literal">JoinableQueue.cancel_join_thread()</tt>), then that process will
not terminate until all buffered items have been flushed to the pipe.
This means that if you try joining that process you may get a deadlock unless
you are sure that all items which have been put on the queue have been
consumed. Similarly, if the child process is non-daemonic then the parent
process may hang on exit when it tries to join all its non-daemonic children.
Note that a queue created using a manager does not have this issue. See
<a class="reference internal" href="#multiprocessing-programming">Programming guidelines</a>.
</div>
For an example of the usage of queues for interprocess communication see
<a class="reference internal" href="#multiprocessing-examples">Examples</a>.
<dl class="function">
<dt id="multiprocessing.Pipe">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">Pipe</tt><big>(</big>[duplex]<big>)</big><a class="headerlink" href="#multiprocessing.Pipe" title="Permalink to this definition">¶</a></dt>
<dd>Returns a pair <tt class="docutils literal">(conn1, conn2)</tt> of <a class="reference internal" href="#multiprocessing.Connection" title="multiprocessing.Connection"><tt class="xref py py-class docutils literal">Connection</tt></a> objects representing
the ends of a pipe.
If duplex is <tt class="docutils literal">True</tt> (the default) then the pipe is bidirectional. If
duplex is <tt class="docutils literal">False</tt> then the pipe is unidirectional: <tt class="docutils literal">conn1</tt> can only be
used for receiving messages and <tt class="docutils literal">conn2</tt> can only be used for sending
messages.
</dd></dl>

<dl class="class">
<dt id="multiprocessing.Queue">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">Queue</tt><big>(</big>[maxsize]<big>)</big><a class="headerlink" href="#multiprocessing.Queue" title="Permalink to this definition">¶</a></dt>
<dd>Returns a process shared queue implemented using a pipe and a few
locks/semaphores. When a process first puts an item on the queue a feeder
thread is started which transfers objects from a buffer into the pipe.
The usual <a class="reference internal" href="queue.html#Queue.Empty" title="Queue.Empty"><tt class="xref py py-exc docutils literal">Queue.Empty</tt></a> and <a class="reference internal" href="queue.html#Queue.Full" title="Queue.Full"><tt class="xref py py-exc docutils literal">Queue.Full</tt></a> exceptions from the
standard library&#8217;s <a class="reference internal" href="queue.html#module-Queue" title="Queue: A synchronized queue class."><tt class="xref py py-mod docutils literal">Queue</tt></a> module are raised to signal timeouts.
<a class="reference internal" href="#multiprocessing.Queue" title="multiprocessing.Queue"><tt class="xref py py-class docutils literal">Queue</tt></a> implements all the methods of <a class="reference internal" href="queue.html#Queue.Queue" title="Queue.Queue"><tt class="xref py py-class docutils literal">Queue.Queue</tt></a> except for
<a class="reference internal" href="queue.html#Queue.Queue.task_done" title="Queue.Queue.task_done"><tt class="xref py py-meth docutils literal">task_done()</tt></a> and <a class="reference internal" href="queue.html#Queue.Queue.join" title="Queue.Queue.join"><tt class="xref py py-meth docutils literal">join()</tt></a>.
<dl class="method">
<dt id="multiprocessing.Queue.qsize">
<tt class="descname">qsize</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Queue.qsize" title="Permalink to this definition">¶</a></dt>
<dd>Return the approximate size of the queue. Because of
multithreading/multiprocessing semantics, this number is not reliable.
Note that this may raise <a class="reference internal" href="exceptions.html#exceptions.NotImplementedError" title="exceptions.NotImplementedError"><tt class="xref py py-exc docutils literal">NotImplementedError</tt></a> on Unix platforms like
Mac OS X where <tt class="docutils literal">sem_getvalue()</tt> is not implemented.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Queue.empty">
<tt class="descname">empty</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Queue.empty" title="Permalink to this definition">¶</a></dt>
<dd>Return <tt class="docutils literal">True</tt> if the queue is empty, <tt class="docutils literal">False</tt> otherwise. Because of
multithreading/multiprocessing semantics, this is not reliable.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Queue.full">
<tt class="descname">full</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Queue.full" title="Permalink to this definition">¶</a></dt>
<dd>Return <tt class="docutils literal">True</tt> if the queue is full, <tt class="docutils literal">False</tt> otherwise. Because of
multithreading/multiprocessing semantics, this is not reliable.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Queue.put">
<tt class="descname">put</tt><big>(</big>obj[, block[, timeout]]<big>)</big><a class="headerlink" href="#multiprocessing.Queue.put" title="Permalink to this definition">¶</a></dt>
<dd>Put obj into the queue. If the optional argument block is <tt class="docutils literal">True</tt>
(the default) and timeout is <tt class="docutils literal">None</tt> (the default), block if necessary until
a free slot is available. If timeout is a positive number, it blocks at
most timeout seconds and raises the <a class="reference internal" href="queue.html#Queue.Full" title="Queue.Full"><tt class="xref py py-exc docutils literal">Queue.Full</tt></a> exception if no
free slot was available within that time. Otherwise (block is
<tt class="docutils literal">False</tt>), put an item on the queue if a free slot is immediately
available, else raise the <a class="reference internal" href="queue.html#Queue.Full" title="Queue.Full"><tt class="xref py py-exc docutils literal">Queue.Full</tt></a> exception (timeout is
ignored in that case).
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Queue.put_nowait">
<tt class="descname">put_nowait</tt><big>(</big>obj<big>)</big><a class="headerlink" href="#multiprocessing.Queue.put_nowait" title="Permalink to this definition">¶</a></dt>
<dd>Equivalent to <tt class="docutils literal">put(obj, False)</tt>.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Queue.get">
<tt class="descname">get</tt><big>(</big>[block[, timeout]]<big>)</big><a class="headerlink" href="#multiprocessing.Queue.get" title="Permalink to this definition">¶</a></dt>
<dd>Remove and return an item from the queue. If optional args block is
<tt class="docutils literal">True</tt> (the default) and timeout is <tt class="docutils literal">None</tt> (the default), block if
necessary until an item is available. If timeout is a positive number,
it blocks at most timeout seconds and raises the <a class="reference internal" href="queue.html#Queue.Empty" title="Queue.Empty"><tt class="xref py py-exc docutils literal">Queue.Empty</tt></a>
exception if no item was available within that time. Otherwise (block is
<tt class="docutils literal">False</tt>), return an item if one is immediately available, else raise the
<a class="reference internal" href="queue.html#Queue.Empty" title="Queue.Empty"><tt class="xref py py-exc docutils literal">Queue.Empty</tt></a> exception (timeout is ignored in that case).
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Queue.get_nowait">
<tt class="descname">get_nowait</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Queue.get_nowait" title="Permalink to this definition">¶</a></dt>
<dd>Equivalent to <tt class="docutils literal">get(False)</tt>.
</dd></dl>

<a class="reference internal" href="#multiprocessing.Queue" title="multiprocessing.Queue"><tt class="xref py py-class docutils literal">Queue</tt></a> has a few additional methods not found in
<a class="reference internal" href="queue.html#Queue.Queue" title="Queue.Queue"><tt class="xref py py-class docutils literal">Queue.Queue</tt></a>. These methods are usually unnecessary for most
code:
<dl class="method">
<dt id="multiprocessing.Queue.close">
<tt class="descname">close</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Queue.close" title="Permalink to this definition">¶</a></dt>
<dd>Indicate that no more data will be put on this queue by the current
process. The background thread will quit once it has flushed all buffered
data to the pipe. This is called automatically when the queue is garbage
collected.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Queue.join_thread">
<tt class="descname">join_thread</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Queue.join_thread" title="Permalink to this definition">¶</a></dt>
<dd>Join the background thread. This can only be used after <a class="reference internal" href="#multiprocessing.Queue.close" title="multiprocessing.Queue.close"><tt class="xref py py-meth docutils literal">close()</tt></a> has
been called. It blocks until the background thread exits, ensuring that
all data in the buffer has been flushed to the pipe.
By default if a process is not the creator of the queue then on exit it
will attempt to join the queue&#8217;s background thread. The process can call
<a class="reference internal" href="#multiprocessing.Queue.cancel_join_thread" title="multiprocessing.Queue.cancel_join_thread"><tt class="xref py py-meth docutils literal">cancel_join_thread()</tt></a> to make <a class="reference internal" href="#multiprocessing.Queue.join_thread" title="multiprocessing.Queue.join_thread"><tt class="xref py py-meth docutils literal">join_thread()</tt></a> do nothing.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Queue.cancel_join_thread">
<tt class="descname">cancel_join_thread</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Queue.cancel_join_thread" title="Permalink to this definition">¶</a></dt>
<dd>Prevent <a class="reference internal" href="#multiprocessing.Queue.join_thread" title="multiprocessing.Queue.join_thread"><tt class="xref py py-meth docutils literal">join_thread()</tt></a> from blocking. In particular, this prevents
the background thread from being joined automatically when the process
exits &#8211; see <a class="reference internal" href="#multiprocessing.Queue.join_thread" title="multiprocessing.Queue.join_thread"><tt class="xref py py-meth docutils literal">join_thread()</tt></a>.
</dd></dl>

</dd></dl>

<dl class="class">
<dt id="multiprocessing.multiprocessing.queues.SimpleQueue">
class <tt class="descclassname">multiprocessing.queues.</tt><tt class="descname">SimpleQueue</tt><a class="headerlink" href="#multiprocessing.multiprocessing.queues.SimpleQueue" title="Permalink to this definition">¶</a></dt>
<dd>It is a simplified <a class="reference internal" href="#multiprocessing.Queue" title="multiprocessing.Queue"><tt class="xref py py-class docutils literal">Queue</tt></a> type, very close to a locked <a class="reference internal" href="#multiprocessing.Pipe" title="multiprocessing.Pipe"><tt class="xref py py-class docutils literal">Pipe</tt></a>.
<dl class="method">
<dt id="multiprocessing.multiprocessing.queues.SimpleQueue.empty">
<tt class="descname">empty</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.multiprocessing.queues.SimpleQueue.empty" title="Permalink to this definition">¶</a></dt>
<dd>Return <tt class="docutils literal">True</tt> if the queue is empty, <tt class="docutils literal">False</tt> otherwise.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.multiprocessing.queues.SimpleQueue.get">
<tt class="descname">get</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.multiprocessing.queues.SimpleQueue.get" title="Permalink to this definition">¶</a></dt>
<dd>Remove and return an item from the queue.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.multiprocessing.queues.SimpleQueue.put">
<tt class="descname">put</tt><big>(</big>item<big>)</big><a class="headerlink" href="#multiprocessing.multiprocessing.queues.SimpleQueue.put" title="Permalink to this definition">¶</a></dt>
<dd>Put item into the queue.
</dd></dl>

</dd></dl>

<dl class="class">
<dt id="multiprocessing.JoinableQueue">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">JoinableQueue</tt><big>(</big>[maxsize]<big>)</big><a class="headerlink" href="#multiprocessing.JoinableQueue" title="Permalink to this definition">¶</a></dt>
<dd><a class="reference internal" href="#multiprocessing.JoinableQueue" title="multiprocessing.JoinableQueue"><tt class="xref py py-class docutils literal">JoinableQueue</tt></a>, a <a class="reference internal" href="#multiprocessing.Queue" title="multiprocessing.Queue"><tt class="xref py py-class docutils literal">Queue</tt></a> subclass, is a queue which
additionally has <a class="reference internal" href="#multiprocessing.JoinableQueue.task_done" title="multiprocessing.JoinableQueue.task_done"><tt class="xref py py-meth docutils literal">task_done()</tt></a> and <a class="reference internal" href="#multiprocessing.JoinableQueue.join" title="multiprocessing.JoinableQueue.join"><tt class="xref py py-meth docutils literal">join()</tt></a> methods.
<dl class="method">
<dt id="multiprocessing.JoinableQueue.task_done">
<tt class="descname">task_done</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.JoinableQueue.task_done" title="Permalink to this definition">¶</a></dt>
<dd>Indicate that a formerly enqueued task is complete. Used by queue consumer
threads. For each <a class="reference internal" href="#multiprocessing.Queue.get" title="multiprocessing.Queue.get"><tt class="xref py py-meth docutils literal">get()</tt></a> used to fetch a task, a subsequent
call to <a class="reference internal" href="#multiprocessing.JoinableQueue.task_done" title="multiprocessing.JoinableQueue.task_done"><tt class="xref py py-meth docutils literal">task_done()</tt></a> tells the queue that the processing on the task
is complete.
If a <tt class="xref py py-meth docutils literal">join()</tt> is currently blocking, it will resume when all
items have been processed (meaning that a <a class="reference internal" href="#multiprocessing.JoinableQueue.task_done" title="multiprocessing.JoinableQueue.task_done"><tt class="xref py py-meth docutils literal">task_done()</tt></a> call was
received for every item that had been <a class="reference internal" href="#multiprocessing.Queue.put" title="multiprocessing.Queue.put"><tt class="xref py py-meth docutils literal">put()</tt></a> into the queue).
Raises a <a class="reference internal" href="exceptions.html#exceptions.ValueError" title="exceptions.ValueError"><tt class="xref py py-exc docutils literal">ValueError</tt></a> if called more times than there were items
placed in the queue.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.JoinableQueue.join">
<tt class="descname">join</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.JoinableQueue.join" title="Permalink to this definition">¶</a></dt>
<dd>Block until all items in the queue have been gotten and processed.
The count of unfinished tasks goes up whenever an item is added to the
queue. The count goes down whenever a consumer thread calls
<a class="reference internal" href="#multiprocessing.JoinableQueue.task_done" title="multiprocessing.JoinableQueue.task_done"><tt class="xref py py-meth docutils literal">task_done()</tt></a> to indicate that the item was retrieved and all work on
it is complete. When the count of unfinished tasks drops to zero,
<tt class="xref py py-meth docutils literal">join()</tt> unblocks.
</dd></dl>

</dd></dl>

</div>
<div class="section" id="miscellaneous">
<h3>16.6.2.3. Miscellaneous<a class="headerlink" href="#miscellaneous" title="Permalink to this headline">¶</a></h3>
<dl class="function">
<dt id="multiprocessing.active_children">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">active_children</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.active_children" title="Permalink to this definition">¶</a></dt>
<dd>Return list of all live children of the current process.
Calling this has the side affect of &#8220;joining&#8221; any processes which have
already finished.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.cpu_count">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">cpu_count</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.cpu_count" title="Permalink to this definition">¶</a></dt>
<dd>Return the number of CPUs in the system. May raise
<a class="reference internal" href="exceptions.html#exceptions.NotImplementedError" title="exceptions.NotImplementedError"><tt class="xref py py-exc docutils literal">NotImplementedError</tt></a>.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.current_process">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">current_process</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.current_process" title="Permalink to this definition">¶</a></dt>
<dd>Return the <a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a> object corresponding to the current process.
An analogue of <a class="reference internal" href="threading.html#threading.current_thread" title="threading.current_thread"><tt class="xref py py-func docutils literal">threading.current_thread()</tt></a>.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.freeze_support">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">freeze_support</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.freeze_support" title="Permalink to this definition">¶</a></dt>
<dd>Add support for when a program which uses <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> has been
frozen to produce a Windows executable. (Has been tested with py2exe,
PyInstaller and cx_Freeze.)
One needs to call this function straight after the <tt class="docutils literal">if __name__ ==
'__main__'</tt> line of the main module. For example:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Process, freeze_support

def f():
 print &#39;hello world!&#39;

<dl class="function">
<dt id="multiprocessing.set_executable">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">set_executable</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.set_executable" title="Permalink to this definition">¶</a></dt>
<dd>Sets the path of the Python interpreter to use when starting a child process.
(By default <a class="reference internal" href="sys.html#sys.executable" title="sys.executable"><tt class="xref py py-data docutils literal">sys.executable</tt></a> is used). Embedders will probably need to
do some thing like
<div class="highlight-python"><div class="highlight"><pre>set_executable(os.path.join(sys.exec_prefix, &#39;pythonw.exe&#39;))
</pre></div>
</div>
before they can create child processes. (Windows only)
</dd></dl>

<div class="admonition note">
Note
<a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> contains no analogues of
<a class="reference internal" href="threading.html#threading.active_count" title="threading.active_count"><tt class="xref py py-func docutils literal">threading.active_count()</tt></a>, <a class="reference internal" href="threading.html#threading.enumerate" title="threading.enumerate"><tt class="xref py py-func docutils literal">threading.enumerate()</tt></a>,
<a class="reference internal" href="threading.html#threading.settrace" title="threading.settrace"><tt class="xref py py-func docutils literal">threading.settrace()</tt></a>, <a class="reference internal" href="threading.html#threading.setprofile" title="threading.setprofile"><tt class="xref py py-func docutils literal">threading.setprofile()</tt></a>,
<a class="reference internal" href="threading.html#threading.Timer" title="threading.Timer"><tt class="xref py py-class docutils literal">threading.Timer</tt></a>, or <a class="reference internal" href="threading.html#threading.local" title="threading.local"><tt class="xref py py-class docutils literal">threading.local</tt></a>.
</div>
</div>
<div class="section" id="connection-objects">
<h3>16.6.2.4. Connection Objects<a class="headerlink" href="#connection-objects" title="Permalink to this headline">¶</a></h3>
Connection objects allow the sending and receiving of picklable objects or
strings. They can be thought of as message oriented connected sockets.
Connection objects are usually created using <a class="reference internal" href="#multiprocessing.Pipe" title="multiprocessing.Pipe"><tt class="xref py py-func docutils literal">Pipe()</tt></a> &#8211; see also
<a class="reference internal" href="#multiprocessing-listeners-clients">Listeners and Clients</a>.
<dl class="class">
<dt id="multiprocessing.Connection">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">Connection</tt><a class="headerlink" href="#multiprocessing.Connection" title="Permalink to this definition">¶</a></dt>
<dd><dl class="method">
<dt id="multiprocessing.Connection.send">
<tt class="descname">send</tt><big>(</big>obj<big>)</big><a class="headerlink" href="#multiprocessing.Connection.send" title="Permalink to this definition">¶</a></dt>
<dd>Send an object to the other end of the connection which should be read
using <a class="reference internal" href="#multiprocessing.Connection.recv" title="multiprocessing.Connection.recv"><tt class="xref py py-meth docutils literal">recv()</tt></a>.
The object must be picklable. Very large pickles (approximately 32 MB+,
though it depends on the OS) may raise a <a class="reference internal" href="exceptions.html#exceptions.ValueError" title="exceptions.ValueError"><tt class="xref py py-exc docutils literal">ValueError</tt></a> exception.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Connection.recv">
<tt class="descname">recv</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Connection.recv" title="Permalink to this definition">¶</a></dt>
<dd>Return an object sent from the other end of the connection using
<a class="reference internal" href="#multiprocessing.Connection.send" title="multiprocessing.Connection.send"><tt class="xref py py-meth docutils literal">send()</tt></a>. Blocks until there its something to receive. Raises
<a class="reference internal" href="exceptions.html#exceptions.EOFError" title="exceptions.EOFError"><tt class="xref py py-exc docutils literal">EOFError</tt></a> if there is nothing left to receive
and the other end was closed.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Connection.fileno">
<tt class="descname">fileno</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Connection.fileno" title="Permalink to this definition">¶</a></dt>
<dd>Return the file descriptor or handle used by the connection.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Connection.close">
<tt class="descname">close</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.Connection.close" title="Permalink to this definition">¶</a></dt>
<dd>Close the connection.
This is called automatically when the connection is garbage collected.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Connection.poll">
<tt class="descname">poll</tt><big>(</big>[timeout]<big>)</big><a class="headerlink" href="#multiprocessing.Connection.poll" title="Permalink to this definition">¶</a></dt>
<dd>Return whether there is any data available to be read.
If timeout is not specified then it will return immediately. If
timeout is a number then this specifies the maximum time in seconds to
block. If timeout is <tt class="docutils literal">None</tt> then an infinite timeout is used.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Connection.send_bytes">
<tt class="descname">send_bytes</tt><big>(</big>buffer[, offset[, size]]<big>)</big><a class="headerlink" href="#multiprocessing.Connection.send_bytes" title="Permalink to this definition">¶</a></dt>
<dd>Send byte data from an object supporting the buffer interface as a
complete message.
If offset is given then data is read from that position in buffer. If
size is given then that many bytes will be read from buffer. Very large
buffers (approximately 32 MB+, though it depends on the OS) may raise a
<a class="reference internal" href="exceptions.html#exceptions.ValueError" title="exceptions.ValueError"><tt class="xref py py-exc docutils literal">ValueError</tt></a> exception
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Connection.recv_bytes">
<tt class="descname">recv_bytes</tt><big>(</big>[maxlength]<big>)</big><a class="headerlink" href="#multiprocessing.Connection.recv_bytes" title="Permalink to this definition">¶</a></dt>
<dd>Return a complete message of byte data sent from the other end of the
connection as a string. Blocks until there is something to receive.
Raises <a class="reference internal" href="exceptions.html#exceptions.EOFError" title="exceptions.EOFError"><tt class="xref py py-exc docutils literal">EOFError</tt></a> if there is nothing left
to receive and the other end has closed.
If maxlength is specified and the message is longer than maxlength
then <a class="reference internal" href="exceptions.html#exceptions.IOError" title="exceptions.IOError"><tt class="xref py py-exc docutils literal">IOError</tt></a> is raised and the connection will no longer be
readable.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.Connection.recv_bytes_into">
<tt class="descname">recv_bytes_into</tt><big>(</big>buffer[, offset]<big>)</big><a class="headerlink" href="#multiprocessing.Connection.recv_bytes_into" title="Permalink to this definition">¶</a></dt>
<dd>Read into buffer a complete message of byte data sent from the other end
of the connection and return the number of bytes in the message. Blocks
until there is something to receive. Raises
<a class="reference internal" href="exceptions.html#exceptions.EOFError" title="exceptions.EOFError"><tt class="xref py py-exc docutils literal">EOFError</tt></a> if there is nothing left to receive and the other end was
closed.
buffer must be an object satisfying the writable buffer interface. If
offset is given then the message will be written into the buffer from
that position. Offset must be a non-negative integer less than the
length of buffer (in bytes).
If the buffer is too short then a <a class="reference internal" href="#multiprocessing.BufferTooShort" title="multiprocessing.BufferTooShort"><tt class="xref py py-exc docutils literal">BufferTooShort</tt></a> exception is
raised and the complete message is available as <tt class="docutils literal">e.args[0]</tt> where <tt class="docutils literal">e</tt>
is the exception instance.
</dd></dl>

</dd></dl>

For example:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing import Pipe
&gt;&gt;&gt; a, b = Pipe()
&gt;&gt;&gt; a.send([1, &#39;hello&#39;, None])
&gt;&gt;&gt; b.recv()
[1, &#39;hello&#39;, None]
&gt;&gt;&gt; b.send_bytes(&#39;thank you&#39;)
&gt;&gt;&gt; a.recv_bytes()
&#39;thank you&#39;
&gt;&gt;&gt; import array
&gt;&gt;&gt; arr1 = array.array(&#39;i&#39;, range(5))
&gt;&gt;&gt; arr2 = array.array(&#39;i&#39;, [0] * 10)
&gt;&gt;&gt; a.send_bytes(arr1)
&gt;&gt;&gt; count = b.recv_bytes_into(arr2)
&gt;&gt;&gt; assert count == len(arr1) * arr1.itemsize
&gt;&gt;&gt; arr2
array(&#39;i&#39;, [0, 1, 2, 3, 4, 0, 0, 0, 0, 0])
</pre></div>
</div>
<div class="admonition warning">
Warning
The <a class="reference internal" href="#multiprocessing.Connection.recv" title="multiprocessing.Connection.recv"><tt class="xref py py-meth docutils literal">Connection.recv()</tt></a> method automatically unpickles the data it
receives, which can be a security risk unless you can trust the process
which sent the message.
Therefore, unless the connection object was produced using <a class="reference internal" href="#multiprocessing.Pipe" title="multiprocessing.Pipe"><tt class="xref py py-func docutils literal">Pipe()</tt></a> you
should only use the <a class="reference internal" href="#multiprocessing.Connection.recv" title="multiprocessing.Connection.recv"><tt class="xref py py-meth docutils literal">recv()</tt></a> and <a class="reference internal" href="#multiprocessing.Connection.send" title="multiprocessing.Connection.send"><tt class="xref py py-meth docutils literal">send()</tt></a>
methods after performing some sort of authentication. See
<a class="reference internal" href="#multiprocessing-auth-keys">Authentication keys</a>.
</div>
<div class="admonition warning">
Warning
If a process is killed while it is trying to read or write to a pipe then
the data in the pipe is likely to become corrupted, because it may become
impossible to be sure where the message boundaries lie.
</div>
</div>
<div class="section" id="synchronization-primitives">
<h3>16.6.2.5. Synchronization primitives<a class="headerlink" href="#synchronization-primitives" title="Permalink to this headline">¶</a></h3>
Generally synchronization primitives are not as necessary in a multiprocess
program as they are in a multithreaded program. See the documentation for
<a class="reference internal" href="threading.html#module-threading" title="threading: Higher-level threading interface."><tt class="xref py py-mod docutils literal">threading</tt></a> module.
Note that one can also create synchronization primitives by using a manager
object &#8211; see <a class="reference internal" href="#multiprocessing-managers">Managers</a>.
<dl class="class">
<dt id="multiprocessing.BoundedSemaphore">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">BoundedSemaphore</tt><big>(</big>[value]<big>)</big><a class="headerlink" href="#multiprocessing.BoundedSemaphore" title="Permalink to this definition">¶</a></dt>
<dd>A bounded semaphore object: a clone of <a class="reference internal" href="threading.html#threading.BoundedSemaphore" title="threading.BoundedSemaphore"><tt class="xref py py-class docutils literal">threading.BoundedSemaphore</tt></a>.
(On Mac OS X, this is indistinguishable from <a class="reference internal" href="#multiprocessing.Semaphore" title="multiprocessing.Semaphore"><tt class="xref py py-class docutils literal">Semaphore</tt></a> because
<tt class="docutils literal">sem_getvalue()</tt> is not implemented on that platform).
</dd></dl>

<dl class="class">
<dt id="multiprocessing.Condition">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">Condition</tt><big>(</big>[lock]<big>)</big><a class="headerlink" href="#multiprocessing.Condition" title="Permalink to this definition">¶</a></dt>
<dd>A condition variable: a clone of <a class="reference internal" href="threading.html#threading.Condition" title="threading.Condition"><tt class="xref py py-class docutils literal">threading.Condition</tt></a>.
If lock is specified then it should be a <a class="reference internal" href="#multiprocessing.Lock" title="multiprocessing.Lock"><tt class="xref py py-class docutils literal">Lock</tt></a> or <a class="reference internal" href="#multiprocessing.RLock" title="multiprocessing.RLock"><tt class="xref py py-class docutils literal">RLock</tt></a>
object from <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a>.
</dd></dl>

<dl class="class">
<dt id="multiprocessing.Event">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">Event</tt><a class="headerlink" href="#multiprocessing.Event" title="Permalink to this definition">¶</a></dt>
<dd>A clone of <a class="reference internal" href="threading.html#threading.Event" title="threading.Event"><tt class="xref py py-class docutils literal">threading.Event</tt></a>.
This method returns the state of the internal semaphore on exit, so it
will always return <tt class="docutils literal">True</tt> except if a timeout is given and the operation
times out.

Changed in version 2.7: Previously, the method always returned <tt class="docutils literal">None</tt>.
</dd></dl>

<dl class="class">
<dt id="multiprocessing.Lock">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">Lock</tt><a class="headerlink" href="#multiprocessing.Lock" title="Permalink to this definition">¶</a></dt>
<dd>A non-recursive lock object: a clone of <a class="reference internal" href="threading.html#threading.Lock" title="threading.Lock"><tt class="xref py py-class docutils literal">threading.Lock</tt></a>.
</dd></dl>

<dl class="class">
<dt id="multiprocessing.RLock">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">RLock</tt><a class="headerlink" href="#multiprocessing.RLock" title="Permalink to this definition">¶</a></dt>
<dd>A recursive lock object: a clone of <a class="reference internal" href="threading.html#threading.RLock" title="threading.RLock"><tt class="xref py py-class docutils literal">threading.RLock</tt></a>.
</dd></dl>

<dl class="class">
<dt id="multiprocessing.Semaphore">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">Semaphore</tt><big>(</big>[value]<big>)</big><a class="headerlink" href="#multiprocessing.Semaphore" title="Permalink to this definition">¶</a></dt>
<dd>A semaphore object: a clone of <a class="reference internal" href="threading.html#threading.Semaphore" title="threading.Semaphore"><tt class="xref py py-class docutils literal">threading.Semaphore</tt></a>.
</dd></dl>

<div class="admonition note">
Note
The <tt class="xref py py-meth docutils literal">acquire()</tt> method of <a class="reference internal" href="#multiprocessing.BoundedSemaphore" title="multiprocessing.BoundedSemaphore"><tt class="xref py py-class docutils literal">BoundedSemaphore</tt></a>, <a class="reference internal" href="#multiprocessing.Lock" title="multiprocessing.Lock"><tt class="xref py py-class docutils literal">Lock</tt></a>,
<a class="reference internal" href="#multiprocessing.RLock" title="multiprocessing.RLock"><tt class="xref py py-class docutils literal">RLock</tt></a> and <a class="reference internal" href="#multiprocessing.Semaphore" title="multiprocessing.Semaphore"><tt class="xref py py-class docutils literal">Semaphore</tt></a> has a timeout parameter not supported
by the equivalents in <a class="reference internal" href="threading.html#module-threading" title="threading: Higher-level threading interface."><tt class="xref py py-mod docutils literal">threading</tt></a>. The signature is
<tt class="docutils literal">acquire(block=True, timeout=None)</tt> with keyword parameters being
acceptable. If block is <tt class="docutils literal">True</tt> and timeout is not <tt class="docutils literal">None</tt> then it
specifies a timeout in seconds. If block is <tt class="docutils literal">False</tt> then timeout is
ignored.
On Mac OS X, <tt class="docutils literal">sem_timedwait</tt> is unsupported, so calling <tt class="docutils literal">acquire()</tt> with
a timeout will emulate that function&#8217;s behavior using a sleeping loop.
</div>
<div class="admonition note">
Note
If the SIGINT signal generated by Ctrl-C arrives while the main thread is
blocked by a call to <tt class="xref py py-meth docutils literal">BoundedSemaphore.acquire()</tt>, <tt class="xref py py-meth docutils literal">Lock.acquire()</tt>,
<tt class="xref py py-meth docutils literal">RLock.acquire()</tt>, <tt class="xref py py-meth docutils literal">Semaphore.acquire()</tt>, <tt class="xref py py-meth docutils literal">Condition.acquire()</tt>
or <tt class="xref py py-meth docutils literal">Condition.wait()</tt> then the call will be immediately interrupted and
<a class="reference internal" href="exceptions.html#exceptions.KeyboardInterrupt" title="exceptions.KeyboardInterrupt"><tt class="xref py py-exc docutils literal">KeyboardInterrupt</tt></a> will be raised.
This differs from the behaviour of <a class="reference internal" href="threading.html#module-threading" title="threading: Higher-level threading interface."><tt class="xref py py-mod docutils literal">threading</tt></a> where SIGINT will be
ignored while the equivalent blocking calls are in progress.
</div>
</div>
<div class="section" id="shared-ctypes-objects">
<h3>16.6.2.6. Shared <a class="reference internal" href="ctypes.html#module-ctypes" title="ctypes: A foreign function library for Python."><tt class="xref py py-mod docutils literal">ctypes</tt></a> Objects<a class="headerlink" href="#shared-ctypes-objects" title="Permalink to this headline">¶</a></h3>
It is possible to create shared objects using shared memory which can be
inherited by child processes.
<dl class="function">
<dt id="multiprocessing.Value">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">Value</tt><big>(</big>typecode_or_type, *args[, lock]<big>)</big><a class="headerlink" href="#multiprocessing.Value" title="Permalink to this definition">¶</a></dt>
<dd>Return a <a class="reference internal" href="ctypes.html#module-ctypes" title="ctypes: A foreign function library for Python."><tt class="xref py py-mod docutils literal">ctypes</tt></a> object allocated from shared memory. By default the
return value is actually a synchronized wrapper for the object.
typecode_or_type determines the type of the returned object: it is either a
ctypes type or a one character typecode of the kind used by the <a class="reference internal" href="array.html#module-array" title="array: Space efficient arrays of uniformly typed numeric values."><tt class="xref py py-mod docutils literal">array</tt></a>
module. *args is passed on to the constructor for the type.
If lock is <tt class="docutils literal">True</tt> (the default) then a new lock object is created to
synchronize access to the value. If lock is a <a class="reference internal" href="#multiprocessing.Lock" title="multiprocessing.Lock"><tt class="xref py py-class docutils literal">Lock</tt></a> or
<a class="reference internal" href="#multiprocessing.RLock" title="multiprocessing.RLock"><tt class="xref py py-class docutils literal">RLock</tt></a> object then that will be used to synchronize access to the
value. If lock is <tt class="docutils literal">False</tt> then access to the returned object will not be
automatically protected by a lock, so it will not necessarily be
&#8220;process-safe&#8221;.
Note that lock is a keyword-only argument.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.Array">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">Array</tt><big>(</big>typecode_or_type, size_or_initializer, *, lock=True<big>)</big><a class="headerlink" href="#multiprocessing.Array" title="Permalink to this definition">¶</a></dt>
<dd>Return a ctypes array allocated from shared memory. By default the return
value is actually a synchronized wrapper for the array.
typecode_or_type determines the type of the elements of the returned array:
it is either a ctypes type or a one character typecode of the kind used by
the <a class="reference internal" href="array.html#module-array" title="array: Space efficient arrays of uniformly typed numeric values."><tt class="xref py py-mod docutils literal">array</tt></a> module. If size_or_initializer is an integer, then it
determines the length of the array, and the array will be initially zeroed.
Otherwise, size_or_initializer is a sequence which is used to initialize
the array and whose length determines the length of the array.
If lock is <tt class="docutils literal">True</tt> (the default) then a new lock object is created to
synchronize access to the value. If lock is a <a class="reference internal" href="#multiprocessing.Lock" title="multiprocessing.Lock"><tt class="xref py py-class docutils literal">Lock</tt></a> or
<a class="reference internal" href="#multiprocessing.RLock" title="multiprocessing.RLock"><tt class="xref py py-class docutils literal">RLock</tt></a> object then that will be used to synchronize access to the
value. If lock is <tt class="docutils literal">False</tt> then access to the returned object will not be
automatically protected by a lock, so it will not necessarily be
&#8220;process-safe&#8221;.
Note that lock is a keyword only argument.
Note that an array of <a class="reference internal" href="ctypes.html#ctypes.c_char" title="ctypes.c_char"><tt class="xref py py-data docutils literal">ctypes.c_char</tt></a> has value and raw
attributes which allow one to use it to store and retrieve strings.
</dd></dl>

<div class="section" id="module-multiprocessing.sharedctypes">
<h4>16.6.2.6.1. The <a class="reference internal" href="#module-multiprocessing.sharedctypes" title="multiprocessing.sharedctypes: Allocate ctypes objects from shared memory."><tt class="xref py py-mod docutils literal">multiprocessing.sharedctypes</tt></a> module<a class="headerlink" href="#module-multiprocessing.sharedctypes" title="Permalink to this headline">¶</a></h4>
The <a class="reference internal" href="#module-multiprocessing.sharedctypes" title="multiprocessing.sharedctypes: Allocate ctypes objects from shared memory."><tt class="xref py py-mod docutils literal">multiprocessing.sharedctypes</tt></a> module provides functions for allocating
<a class="reference internal" href="ctypes.html#module-ctypes" title="ctypes: A foreign function library for Python."><tt class="xref py py-mod docutils literal">ctypes</tt></a> objects from shared memory which can be inherited by child
processes.
<div class="admonition note">
Note
Although it is possible to store a pointer in shared memory remember that
this will refer to a location in the address space of a specific process.
However, the pointer is quite likely to be invalid in the context of a second
process and trying to dereference the pointer from the second process may
cause a crash.
</div>
<dl class="function">
<dt id="multiprocessing.sharedctypes.RawArray">
<tt class="descclassname">multiprocessing.sharedctypes.</tt><tt class="descname">RawArray</tt><big>(</big>typecode_or_type, size_or_initializer<big>)</big><a class="headerlink" href="#multiprocessing.sharedctypes.RawArray" title="Permalink to this definition">¶</a></dt>
<dd>Return a ctypes array allocated from shared memory.
typecode_or_type determines the type of the elements of the returned array:
it is either a ctypes type or a one character typecode of the kind used by
the <a class="reference internal" href="array.html#module-array" title="array: Space efficient arrays of uniformly typed numeric values."><tt class="xref py py-mod docutils literal">array</tt></a> module. If size_or_initializer is an integer then it
determines the length of the array, and the array will be initially zeroed.
Otherwise size_or_initializer is a sequence which is used to initialize the
array and whose length determines the length of the array.
Note that setting and getting an element is potentially non-atomic &#8211; use
<a class="reference internal" href="#multiprocessing.sharedctypes.Array" title="multiprocessing.sharedctypes.Array"><tt class="xref py py-func docutils literal">Array()</tt></a> instead to make sure that access is automatically synchronized
using a lock.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.sharedctypes.RawValue">
<tt class="descclassname">multiprocessing.sharedctypes.</tt><tt class="descname">RawValue</tt><big>(</big>typecode_or_type, *args<big>)</big><a class="headerlink" href="#multiprocessing.sharedctypes.RawValue" title="Permalink to this definition">¶</a></dt>
<dd>Return a ctypes object allocated from shared memory.
typecode_or_type determines the type of the returned object: it is either a
ctypes type or a one character typecode of the kind used by the <a class="reference internal" href="array.html#module-array" title="array: Space efficient arrays of uniformly typed numeric values."><tt class="xref py py-mod docutils literal">array</tt></a>
module. *args is passed on to the constructor for the type.
Note that setting and getting the value is potentially non-atomic &#8211; use
<a class="reference internal" href="#multiprocessing.sharedctypes.Value" title="multiprocessing.sharedctypes.Value"><tt class="xref py py-func docutils literal">Value()</tt></a> instead to make sure that access is automatically synchronized
using a lock.
Note that an array of <a class="reference internal" href="ctypes.html#ctypes.c_char" title="ctypes.c_char"><tt class="xref py py-data docutils literal">ctypes.c_char</tt></a> has <tt class="docutils literal">value</tt> and <tt class="docutils literal">raw</tt>
attributes which allow one to use it to store and retrieve strings &#8211; see
documentation for <a class="reference internal" href="ctypes.html#module-ctypes" title="ctypes: A foreign function library for Python."><tt class="xref py py-mod docutils literal">ctypes</tt></a>.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.sharedctypes.Array">
<tt class="descclassname">multiprocessing.sharedctypes.</tt><tt class="descname">Array</tt><big>(</big>typecode_or_type, size_or_initializer, *args[, lock]<big>)</big><a class="headerlink" href="#multiprocessing.sharedctypes.Array" title="Permalink to this definition">¶</a></dt>
<dd>The same as <a class="reference internal" href="#multiprocessing.sharedctypes.RawArray" title="multiprocessing.sharedctypes.RawArray"><tt class="xref py py-func docutils literal">RawArray()</tt></a> except that depending on the value of lock a
process-safe synchronization wrapper may be returned instead of a raw ctypes
array.
If lock is <tt class="docutils literal">True</tt> (the default) then a new lock object is created to
synchronize access to the value. If lock is a <tt class="xref py py-class docutils literal">Lock</tt> or
<tt class="xref py py-class docutils literal">RLock</tt> object then that will be used to synchronize access to the
value. If lock is <tt class="docutils literal">False</tt> then access to the returned object will not be
automatically protected by a lock, so it will not necessarily be
&#8220;process-safe&#8221;.
Note that lock is a keyword-only argument.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.sharedctypes.Value">
<tt class="descclassname">multiprocessing.sharedctypes.</tt><tt class="descname">Value</tt><big>(</big>typecode_or_type, *args[, lock]<big>)</big><a class="headerlink" href="#multiprocessing.sharedctypes.Value" title="Permalink to this definition">¶</a></dt>
<dd>The same as <a class="reference internal" href="#multiprocessing.sharedctypes.RawValue" title="multiprocessing.sharedctypes.RawValue"><tt class="xref py py-func docutils literal">RawValue()</tt></a> except that depending on the value of lock a
process-safe synchronization wrapper may be returned instead of a raw ctypes
object.
If lock is <tt class="docutils literal">True</tt> (the default) then a new lock object is created to
synchronize access to the value. If lock is a <tt class="xref py py-class docutils literal">Lock</tt> or
<tt class="xref py py-class docutils literal">RLock</tt> object then that will be used to synchronize access to the
value. If lock is <tt class="docutils literal">False</tt> then access to the returned object will not be
automatically protected by a lock, so it will not necessarily be
&#8220;process-safe&#8221;.
Note that lock is a keyword-only argument.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.sharedctypes.copy">
<tt class="descclassname">multiprocessing.sharedctypes.</tt><tt class="descname">copy</tt><big>(</big>obj<big>)</big><a class="headerlink" href="#multiprocessing.sharedctypes.copy" title="Permalink to this definition">¶</a></dt>
<dd>Return a ctypes object allocated from shared memory which is a copy of the
ctypes object obj.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.sharedctypes.synchronized">
<tt class="descclassname">multiprocessing.sharedctypes.</tt><tt class="descname">synchronized</tt><big>(</big>obj[, lock]<big>)</big><a class="headerlink" href="#multiprocessing.sharedctypes.synchronized" title="Permalink to this definition">¶</a></dt>
<dd>Return a process-safe wrapper object for a ctypes object which uses lock to
synchronize access. If lock is <tt class="docutils literal">None</tt> (the default) then a
<a class="reference internal" href="#multiprocessing.RLock" title="multiprocessing.RLock"><tt class="xref py py-class docutils literal">multiprocessing.RLock</tt></a> object is created automatically.
A synchronized wrapper will have two methods in addition to those of the
object it wraps: <tt class="xref py py-meth docutils literal">get_obj()</tt> returns the wrapped object and
<tt class="xref py py-meth docutils literal">get_lock()</tt> returns the lock object used for synchronization.
Note that accessing the ctypes object through the wrapper can be a lot slower
than accessing the raw ctypes object.
</dd></dl>

The table below compares the syntax for creating shared ctypes objects from
shared memory with the normal ctypes syntax. (In the table <tt class="docutils literal">MyStruct</tt> is some
subclass of <a class="reference internal" href="ctypes.html#ctypes.Structure" title="ctypes.Structure"><tt class="xref py py-class docutils literal">ctypes.Structure</tt></a>.)
<table border="1" class="docutils">
<colgroup>
<col width="27%" />
<col width="36%" />
<col width="37%" />
</colgroup>
<thead valign="bottom">
<tr class="row-odd"><th class="head">ctypes</th>
<th class="head">sharedctypes using type</th>
<th class="head">sharedctypes using typecode</th>
</tr>
</thead>
<tbody valign="top">
<tr class="row-even"><td>c_double(2.4)</td>
<td>RawValue(c_double, 2.4)</td>
<td>RawValue(&#8216;d&#8217;, 2.4)</td>
</tr>
<tr class="row-odd"><td>MyStruct(4, 6)</td>
<td>RawValue(MyStruct, 4, 6)</td>
<td>&nbsp;</td>
</tr>
<tr class="row-even"><td>(c_short * 7)()</td>
<td>RawArray(c_short, 7)</td>
<td>RawArray(&#8216;h&#8217;, 7)</td>
</tr>
<tr class="row-odd"><td>(c_int * 3)(9, 2, 8)</td>
<td>RawArray(c_int, (9, 2, 8))</td>
<td>RawArray(&#8216;i&#8217;, (9, 2, 8))</td>
</tr>
</tbody>
</table>
Below is an example where a number of ctypes objects are modified by a child
process:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Process, Lock
from multiprocessing.sharedctypes import Value, Array
from ctypes import Structure, c_double

class Point(Structure):
 _fields_ = [(&#39;x&#39;, c_double), (&#39;y&#39;, c_double)]

def modify(n, x, s, A):
 n.value **= 2
 x.value **= 2
 s.value = s.value.upper()
 for a in A:
 a.x **= 2
 a.y **= 2

n = Value(&#39;i&#39;, 7)
 x = Value(c_double, 1.0/3.0, lock=False)
 s = Array(&#39;c&#39;, &#39;hello world&#39;, lock=lock)
 A = Array(Point, [(1.875,-6.25), (-5.75,2.0), (2.375,9.5)], lock=lock)

p = Process(target=modify, args=(n, x, s, A))
 p.start()
 p.join()

print n.value
 print x.value
 print s.value
 print [(a.x, a.y) for a in A]
</pre></div>
</div>
The results printed are
<div class="highlight-none"><div class="highlight"><pre>49
0.1111111111111111
HELLO WORLD
[(3.515625, 39.0625), (33.0625, 4.0), (5.640625, 90.25)]
</pre></div>
</div>
</div>
</div>
<div class="section" id="managers">
<h3>16.6.2.7. Managers<a class="headerlink" href="#managers" title="Permalink to this headline">¶</a></h3>
Managers provide a way to create data which can be shared between different
processes. A manager object controls a server process which manages shared
objects. Other processes can access the shared objects by using proxies.
<dl class="function">
<dt id="multiprocessing.sharedctypes.multiprocessing.Manager">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">Manager</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.sharedctypes.multiprocessing.Manager" title="Permalink to this definition">¶</a></dt>
<dd>Returns a started <a class="reference internal" href="#multiprocessing.managers.SyncManager" title="multiprocessing.managers.SyncManager"><tt class="xref py py-class docutils literal">SyncManager</tt></a> object which
can be used for sharing objects between processes. The returned manager
object corresponds to a spawned child process and has methods which will
create shared objects and return corresponding proxies.
</dd></dl>

Manager processes will be shutdown as soon as they are garbage collected or
their parent process exits. The manager classes are defined in the
<a class="reference internal" href="#module-multiprocessing.managers" title="multiprocessing.managers: Share data between process with shared objects."><tt class="xref py py-mod docutils literal">multiprocessing.managers</tt></a> module:
<dl class="class">
<dt id="multiprocessing.managers.BaseManager">
class <tt class="descclassname">multiprocessing.managers.</tt><tt class="descname">BaseManager</tt><big>(</big>[address[, authkey]]<big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseManager" title="Permalink to this definition">¶</a></dt>
<dd>Create a BaseManager object.
Once created one should call <a class="reference internal" href="#multiprocessing.managers.BaseManager.start" title="multiprocessing.managers.BaseManager.start"><tt class="xref py py-meth docutils literal">start()</tt></a> or <tt class="docutils literal">get_server().serve_forever()</tt> to ensure
that the manager object refers to a started manager process.
address is the address on which the manager process listens for new
connections. If address is <tt class="docutils literal">None</tt> then an arbitrary one is chosen.
authkey is the authentication key which will be used to check the validity
of incoming connections to the server process. If authkey is <tt class="docutils literal">None</tt> then
<tt class="docutils literal">current_process().authkey</tt>. Otherwise authkey is used and it
must be a string.
<dl class="method">
<dt id="multiprocessing.managers.BaseManager.start">
<tt class="descname">start</tt><big>(</big>[initializer[, initargs]]<big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseManager.start" title="Permalink to this definition">¶</a></dt>
<dd>Start a subprocess to start the manager. If initializer is not <tt class="docutils literal">None</tt>
then the subprocess will call <tt class="docutils literal">initializer(*initargs)</tt> when it starts.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.BaseManager.get_server">
<tt class="descname">get_server</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseManager.get_server" title="Permalink to this definition">¶</a></dt>
<dd>Returns a <tt class="xref py py-class docutils literal">Server</tt> object which represents the actual server under
the control of the Manager. The <tt class="xref py py-class docutils literal">Server</tt> object supports the
<tt class="xref py py-meth docutils literal">serve_forever()</tt> method:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing.managers import BaseManager
&gt;&gt;&gt; manager = BaseManager(address=(&#39;&#39;, 50000), authkey=&#39;abc&#39;)
&gt;&gt;&gt; server = manager.get_server()
&gt;&gt;&gt; server.serve_forever()
</pre></div>
</div>
<tt class="xref py py-class docutils literal">Server</tt> additionally has an <a class="reference internal" href="#multiprocessing.managers.BaseManager.address" title="multiprocessing.managers.BaseManager.address"><tt class="xref py py-attr docutils literal">address</tt></a> attribute.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.BaseManager.connect">
<tt class="descname">connect</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseManager.connect" title="Permalink to this definition">¶</a></dt>
<dd>Connect a local manager object to a remote manager process:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing.managers import BaseManager
&gt;&gt;&gt; m = BaseManager(address=(&#39;127.0.0.1&#39;, 5000), authkey=&#39;abc&#39;)
&gt;&gt;&gt; m.connect()
</pre></div>
</div>
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.BaseManager.shutdown">
<tt class="descname">shutdown</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseManager.shutdown" title="Permalink to this definition">¶</a></dt>
<dd>Stop the process used by the manager. This is only available if
<a class="reference internal" href="#multiprocessing.managers.BaseManager.start" title="multiprocessing.managers.BaseManager.start"><tt class="xref py py-meth docutils literal">start()</tt></a> has been used to start the server process.
This can be called multiple times.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.BaseManager.register">
<tt class="descname">register</tt><big>(</big>typeid[, callable[, proxytype[, exposed[, method_to_typeid[, create_method]]]]]<big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseManager.register" title="Permalink to this definition">¶</a></dt>
<dd>A classmethod which can be used for registering a type or callable with
the manager class.
typeid is a &#8220;type identifier&#8221; which is used to identify a particular
type of shared object. This must be a string.
callable is a callable used for creating objects for this type
identifier. If a manager instance will be created using the
<tt class="xref py py-meth docutils literal">from_address()</tt> classmethod or if the create_method argument is
<tt class="docutils literal">False</tt> then this can be left as <tt class="docutils literal">None</tt>.
proxytype is a subclass of <a class="reference internal" href="#multiprocessing.managers.BaseProxy" title="multiprocessing.managers.BaseProxy"><tt class="xref py py-class docutils literal">BaseProxy</tt></a> which is used to create
proxies for shared objects with this typeid. If <tt class="docutils literal">None</tt> then a proxy
class is created automatically.
exposed is used to specify a sequence of method names which proxies for
this typeid should be allowed to access using
<tt class="xref py py-meth docutils literal">BaseProxy._callMethod()</tt>. (If exposed is <tt class="docutils literal">None</tt> then
<tt class="xref py py-attr docutils literal">proxytype._exposed_</tt> is used instead if it exists.) In the case
where no exposed list is specified, all &#8220;public methods&#8221; of the shared
object will be accessible. (Here a &#8220;public method&#8221; means any attribute
which has a <a class="reference internal" href="../reference/datamodel.html#object.__call__" title="object.__call__"><tt class="xref py py-meth docutils literal">__call__()</tt></a> method and whose name does not begin with
<tt class="docutils literal">'_'</tt>.)
method_to_typeid is a mapping used to specify the return type of those
exposed methods which should return a proxy. It maps method names to
typeid strings. (If method_to_typeid is <tt class="docutils literal">None</tt> then
<tt class="xref py py-attr docutils literal">proxytype._method_to_typeid_</tt> is used instead if it exists.) If a
method&#8217;s name is not a key of this mapping or if the mapping is <tt class="docutils literal">None</tt>
then the object returned by the method will be copied by value.
create_method determines whether a method should be created with name
typeid which can be used to tell the server process to create a new
shared object and return a proxy for it. By default it is <tt class="docutils literal">True</tt>.
</dd></dl>

<a class="reference internal" href="#multiprocessing.managers.BaseManager" title="multiprocessing.managers.BaseManager"><tt class="xref py py-class docutils literal">BaseManager</tt></a> instances also have one read-only property:
<dl class="attribute">
<dt id="multiprocessing.managers.BaseManager.address">
<tt class="descname">address</tt><a class="headerlink" href="#multiprocessing.managers.BaseManager.address" title="Permalink to this definition">¶</a></dt>
<dd>The address used by the manager.
</dd></dl>

</dd></dl>

<dl class="class">
<dt id="multiprocessing.managers.SyncManager">
class <tt class="descclassname">multiprocessing.managers.</tt><tt class="descname">SyncManager</tt><a class="headerlink" href="#multiprocessing.managers.SyncManager" title="Permalink to this definition">¶</a></dt>
<dd>A subclass of <a class="reference internal" href="#multiprocessing.managers.BaseManager" title="multiprocessing.managers.BaseManager"><tt class="xref py py-class docutils literal">BaseManager</tt></a> which can be used for the synchronization
of processes. Objects of this type are returned by
<tt class="xref py py-func docutils literal">multiprocessing.Manager()</tt>.
It also supports creation of shared lists and dictionaries.
<dl class="method">
<dt id="multiprocessing.managers.SyncManager.BoundedSemaphore">
<tt class="descname">BoundedSemaphore</tt><big>(</big>[value]<big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.BoundedSemaphore" title="Permalink to this definition">¶</a></dt>
<dd>Create a shared <a class="reference internal" href="threading.html#threading.BoundedSemaphore" title="threading.BoundedSemaphore"><tt class="xref py py-class docutils literal">threading.BoundedSemaphore</tt></a> object and return a
proxy for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.Condition">
<tt class="descname">Condition</tt><big>(</big>[lock]<big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.Condition" title="Permalink to this definition">¶</a></dt>
<dd>Create a shared <a class="reference internal" href="threading.html#threading.Condition" title="threading.Condition"><tt class="xref py py-class docutils literal">threading.Condition</tt></a> object and return a proxy for
it.
If lock is supplied then it should be a proxy for a
<a class="reference internal" href="threading.html#threading.Lock" title="threading.Lock"><tt class="xref py py-class docutils literal">threading.Lock</tt></a> or <a class="reference internal" href="threading.html#threading.RLock" title="threading.RLock"><tt class="xref py py-class docutils literal">threading.RLock</tt></a> object.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.Event">
<tt class="descname">Event</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.Event" title="Permalink to this definition">¶</a></dt>
<dd>Create a shared <a class="reference internal" href="threading.html#threading.Event" title="threading.Event"><tt class="xref py py-class docutils literal">threading.Event</tt></a> object and return a proxy for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.Lock">
<tt class="descname">Lock</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.Lock" title="Permalink to this definition">¶</a></dt>
<dd>Create a shared <a class="reference internal" href="threading.html#threading.Lock" title="threading.Lock"><tt class="xref py py-class docutils literal">threading.Lock</tt></a> object and return a proxy for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.Namespace">
<tt class="descname">Namespace</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.Namespace" title="Permalink to this definition">¶</a></dt>
<dd>Create a shared <a class="reference internal" href="#multiprocessing.managers.SyncManager.Namespace" title="multiprocessing.managers.SyncManager.Namespace"><tt class="xref py py-class docutils literal">Namespace</tt></a> object and return a proxy for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.Queue">
<tt class="descname">Queue</tt><big>(</big>[maxsize]<big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.Queue" title="Permalink to this definition">¶</a></dt>
<dd>Create a shared <a class="reference internal" href="queue.html#Queue.Queue" title="Queue.Queue"><tt class="xref py py-class docutils literal">Queue.Queue</tt></a> object and return a proxy for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.RLock">
<tt class="descname">RLock</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.RLock" title="Permalink to this definition">¶</a></dt>
<dd>Create a shared <a class="reference internal" href="threading.html#threading.RLock" title="threading.RLock"><tt class="xref py py-class docutils literal">threading.RLock</tt></a> object and return a proxy for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.Semaphore">
<tt class="descname">Semaphore</tt><big>(</big>[value]<big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.Semaphore" title="Permalink to this definition">¶</a></dt>
<dd>Create a shared <a class="reference internal" href="threading.html#threading.Semaphore" title="threading.Semaphore"><tt class="xref py py-class docutils literal">threading.Semaphore</tt></a> object and return a proxy for
it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.Array">
<tt class="descname">Array</tt><big>(</big>typecode, sequence<big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.Array" title="Permalink to this definition">¶</a></dt>
<dd>Create an array and return a proxy for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.Value">
<tt class="descname">Value</tt><big>(</big>typecode, value<big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.Value" title="Permalink to this definition">¶</a></dt>
<dd>Create an object with a writable <tt class="docutils literal">value</tt> attribute and return a proxy
for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.dict">
<tt class="descname">dict</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.dict" title="Permalink to this definition">¶</a></dt>
<dt>
<tt class="descname">dict</tt><big>(</big>mapping<big>)</big></dt>
<dt>
<tt class="descname">dict</tt><big>(</big>sequence<big>)</big></dt>
<dd>Create a shared <tt class="docutils literal">dict</tt> object and return a proxy for it.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.SyncManager.list">
<tt class="descname">list</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.SyncManager.list" title="Permalink to this definition">¶</a></dt>
<dt>
<tt class="descname">list</tt><big>(</big>sequence<big>)</big></dt>
<dd>Create a shared <tt class="docutils literal">list</tt> object and return a proxy for it.
</dd></dl>

<div class="admonition note">
Note
Modifications to mutable values or items in dict and list proxies will not
be propagated through the manager, because the proxy has no way of knowing
when its values or items are modified. To modify such an item, you can
re-assign the modified object to the container proxy:
<div class="last highlight-python"><div class="highlight"><pre># create a list proxy and append a mutable object (a dictionary)
lproxy = manager.list()
lproxy.append({})
# now mutate the dictionary
d = lproxy[0]
d[&#39;a&#39;] = 1
d[&#39;b&#39;] = 2
# at this point, the changes to d are not yet synced, but by
# reassigning the dictionary, the proxy is notified of the change
lproxy[0] = d
</pre></div>
</div>
</div>
</dd></dl>

<div class="section" id="namespace-objects">
<h4>16.6.2.7.1. Namespace objects<a class="headerlink" href="#namespace-objects" title="Permalink to this headline">¶</a></h4>
A namespace object has no public methods, but does have writable attributes.
Its representation shows the values of its attributes.
However, when using a proxy for a namespace object, an attribute beginning with
<tt class="docutils literal">'_'</tt> will be an attribute of the proxy and not an attribute of the referent:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; manager = multiprocessing.Manager()
&gt;&gt;&gt; Global = manager.Namespace()
&gt;&gt;&gt; Global.x = 10
&gt;&gt;&gt; Global.y = &#39;hello&#39;
&gt;&gt;&gt; Global._z = 12.3 # this is an attribute of the proxy
&gt;&gt;&gt; print Global
Namespace(x=10, y=&#39;hello&#39;)
</pre></div>
</div>
</div>
<div class="section" id="customized-managers">
<h4>16.6.2.7.2. Customized managers<a class="headerlink" href="#customized-managers" title="Permalink to this headline">¶</a></h4>
To create one&#8217;s own manager, one creates a subclass of <a class="reference internal" href="#multiprocessing.managers.BaseManager" title="multiprocessing.managers.BaseManager"><tt class="xref py py-class docutils literal">BaseManager</tt></a> and
uses the <a class="reference internal" href="#multiprocessing.managers.BaseManager.register" title="multiprocessing.managers.BaseManager.register"><tt class="xref py py-meth docutils literal">register()</tt></a> classmethod to register new types or
callables with the manager class. For example:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing.managers import BaseManager

class MathsClass(object):
 def add(self, x, y):
 return x + y
 def mul(self, x, y):
 return x * y

class MyManager(BaseManager):
 pass

MyManager.register(&#39;Maths&#39;, MathsClass)

if __name__ == &#39;__main__&#39;:
 manager = MyManager()
 manager.start()
 maths = manager.Maths()
 print maths.add(4, 3) # prints 7
 print maths.mul(7, 8) # prints 56
</pre></div>
</div>
</div>
<div class="section" id="using-a-remote-manager">
<h4>16.6.2.7.3. Using a remote manager<a class="headerlink" href="#using-a-remote-manager" title="Permalink to this headline">¶</a></h4>
It is possible to run a manager server on one machine and have clients use it
from other machines (assuming that the firewalls involved allow it).
Running the following commands creates a server for a single shared queue which
remote clients can access:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing.managers import BaseManager
&gt;&gt;&gt; import Queue
&gt;&gt;&gt; queue = Queue.Queue()
&gt;&gt;&gt; class QueueManager(BaseManager): pass
&gt;&gt;&gt; QueueManager.register(&#39;get_queue&#39;, callable=lambda:queue)
&gt;&gt;&gt; m = QueueManager(address=(&#39;&#39;, 50000), authkey=&#39;abracadabra&#39;)
&gt;&gt;&gt; s = m.get_server()
&gt;&gt;&gt; s.serve_forever()
</pre></div>
</div>
One client can access the server as follows:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing.managers import BaseManager
&gt;&gt;&gt; class QueueManager(BaseManager): pass
&gt;&gt;&gt; QueueManager.register(&#39;get_queue&#39;)
&gt;&gt;&gt; m = QueueManager(address=(&#39;foo.bar.org&#39;, 50000), authkey=&#39;abracadabra&#39;)
&gt;&gt;&gt; m.connect()
&gt;&gt;&gt; queue = m.get_queue()
&gt;&gt;&gt; queue.put(&#39;hello&#39;)
</pre></div>
</div>
Another client can also use it:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing.managers import BaseManager
&gt;&gt;&gt; class QueueManager(BaseManager): pass
&gt;&gt;&gt; QueueManager.register(&#39;get_queue&#39;)
&gt;&gt;&gt; m = QueueManager(address=(&#39;foo.bar.org&#39;, 50000), authkey=&#39;abracadabra&#39;)
&gt;&gt;&gt; m.connect()
&gt;&gt;&gt; queue = m.get_queue()
&gt;&gt;&gt; queue.get()
&#39;hello&#39;
</pre></div>
</div>
Local processes can also access that queue, using the code from above on the
client to access it remotely:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing import Process, Queue
&gt;&gt;&gt; from multiprocessing.managers import BaseManager
&gt;&gt;&gt; class Worker(Process):
... def __init__(self, q):
... self.q = q
... super(Worker, self).__init__()
... def run(self):
... self.q.put(&#39;local hello&#39;)
...
&gt;&gt;&gt; queue = Queue()
&gt;&gt;&gt; w = Worker(queue)
&gt;&gt;&gt; w.start()
&gt;&gt;&gt; class QueueManager(BaseManager): pass
...
&gt;&gt;&gt; QueueManager.register(&#39;get_queue&#39;, callable=lambda: queue)
&gt;&gt;&gt; m = QueueManager(address=(&#39;&#39;, 50000), authkey=&#39;abracadabra&#39;)
&gt;&gt;&gt; s = m.get_server()
&gt;&gt;&gt; s.serve_forever()
</pre></div>
</div>
</div>
</div>
<div class="section" id="proxy-objects">
<h3>16.6.2.8. Proxy Objects<a class="headerlink" href="#proxy-objects" title="Permalink to this headline">¶</a></h3>
A proxy is an object which refers to a shared object which lives (presumably)
in a different process. The shared object is said to be the referent of the
proxy. Multiple proxy objects may have the same referent.
A proxy object has methods which invoke corresponding methods of its referent
(although not every method of the referent will necessarily be available through
the proxy). A proxy can usually be used in most of the same ways that its
referent can:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; from multiprocessing import Manager
&gt;&gt;&gt; manager = Manager()
&gt;&gt;&gt; l = manager.list([i*i for i in range(10)])
&gt;&gt;&gt; print l
[0, 1, 4, 9, 16, 25, 36, 49, 64, 81]
&gt;&gt;&gt; print repr(l)
&lt;ListProxy object, typeid &#39;list&#39; at 0x...&gt;
&gt;&gt;&gt; l[4]
16
&gt;&gt;&gt; l[2:5]
[4, 9, 16]
</pre></div>
</div>
Notice that applying <a class="reference internal" href="functions.html#str" title="str"><tt class="xref py py-func docutils literal">str()</tt></a> to a proxy will return the representation of
the referent, whereas applying <a class="reference internal" href="repr.html#module-repr" title="repr: Alternate repr() implementation with size limits."><tt class="xref py py-func docutils literal">repr()</tt></a> will return the representation of
the proxy.
An important feature of proxy objects is that they are picklable so they can be
passed between processes. Note, however, that if a proxy is sent to the
corresponding manager&#8217;s process then unpickling it will produce the referent
itself. This means, for example, that one shared object can contain a second:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; a = manager.list()
&gt;&gt;&gt; b = manager.list()
&gt;&gt;&gt; a.append(b) # referent of a now contains referent of b
&gt;&gt;&gt; print a, b
[[]] []
&gt;&gt;&gt; b.append(&#39;hello&#39;)
&gt;&gt;&gt; print a, b
[[&#39;hello&#39;]] [&#39;hello&#39;]
</pre></div>
</div>
<div class="admonition note">
Note
The proxy types in <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> do nothing to support comparisons
by value. So, for instance, we have:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; manager.list([1,2,3]) == [1,2,3]
False
</pre></div>
</div>
One should just use a copy of the referent instead when making comparisons.
</div>
<dl class="class">
<dt id="multiprocessing.managers.BaseProxy">
class <tt class="descclassname">multiprocessing.managers.</tt><tt class="descname">BaseProxy</tt><a class="headerlink" href="#multiprocessing.managers.BaseProxy" title="Permalink to this definition">¶</a></dt>
<dd>Proxy objects are instances of subclasses of <a class="reference internal" href="#multiprocessing.managers.BaseProxy" title="multiprocessing.managers.BaseProxy"><tt class="xref py py-class docutils literal">BaseProxy</tt></a>.
<dl class="method">
<dt id="multiprocessing.managers.BaseProxy._callmethod">
<tt class="descname">_callmethod</tt><big>(</big>methodname[, args[, kwds]]<big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseProxy._callmethod" title="Permalink to this definition">¶</a></dt>
<dd>Call and return the result of a method of the proxy&#8217;s referent.
If <tt class="docutils literal">proxy</tt> is a proxy whose referent is <tt class="docutils literal">obj</tt> then the expression
<div class="highlight-python"><div class="highlight"><pre>proxy._callmethod(methodname, args, kwds)
</pre></div>
</div>
will evaluate the expression
<div class="highlight-python"><div class="highlight"><pre>getattr(obj, methodname)(*args, **kwds)
</pre></div>
</div>
in the manager&#8217;s process.
The returned value will be a copy of the result of the call or a proxy to
a new shared object &#8211; see documentation for the method_to_typeid
argument of <a class="reference internal" href="#multiprocessing.managers.BaseManager.register" title="multiprocessing.managers.BaseManager.register"><tt class="xref py py-meth docutils literal">BaseManager.register()</tt></a>.
If an exception is raised by the call, then is re-raised by
<a class="reference internal" href="#multiprocessing.managers.BaseProxy._callmethod" title="multiprocessing.managers.BaseProxy._callmethod"><tt class="xref py py-meth docutils literal">_callmethod()</tt></a>. If some other exception is raised in the manager&#8217;s
process then this is converted into a <tt class="xref py py-exc docutils literal">RemoteError</tt> exception and is
raised by <a class="reference internal" href="#multiprocessing.managers.BaseProxy._callmethod" title="multiprocessing.managers.BaseProxy._callmethod"><tt class="xref py py-meth docutils literal">_callmethod()</tt></a>.
Note in particular that an exception will be raised if methodname has
not been exposed
An example of the usage of <a class="reference internal" href="#multiprocessing.managers.BaseProxy._callmethod" title="multiprocessing.managers.BaseProxy._callmethod"><tt class="xref py py-meth docutils literal">_callmethod()</tt></a>:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; l = manager.list(range(10))
&gt;&gt;&gt; l._callmethod(&#39;__len__&#39;)
10
&gt;&gt;&gt; l._callmethod(&#39;__getslice__&#39;, (2, 7)) # equiv to `l[2:7]`
[2, 3, 4, 5, 6]
&gt;&gt;&gt; l._callmethod(&#39;__getitem__&#39;, (20,)) # equiv to `l[20]`
Traceback (most recent call last):
...
IndexError: list index out of range
</pre></div>
</div>
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.BaseProxy._getvalue">
<tt class="descname">_getvalue</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseProxy._getvalue" title="Permalink to this definition">¶</a></dt>
<dd>Return a copy of the referent.
If the referent is unpicklable then this will raise an exception.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.BaseProxy.__repr__">
<tt class="descname">__repr__</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseProxy.__repr__" title="Permalink to this definition">¶</a></dt>
<dd>Return a representation of the proxy object.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.managers.BaseProxy.__str__">
<tt class="descname">__str__</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.managers.BaseProxy.__str__" title="Permalink to this definition">¶</a></dt>
<dd>Return the representation of the referent.
</dd></dl>

</dd></dl>

<div class="section" id="cleanup">
<h4>16.6.2.8.1. Cleanup<a class="headerlink" href="#cleanup" title="Permalink to this headline">¶</a></h4>
A proxy object uses a weakref callback so that when it gets garbage collected it
deregisters itself from the manager which owns its referent.
A shared object gets deleted from the manager process when there are no longer
any proxies referring to it.
</div>
</div>
<div class="section" id="module-multiprocessing.pool">
<h3>16.6.2.9. Process Pools<a class="headerlink" href="#module-multiprocessing.pool" title="Permalink to this headline">¶</a></h3>
One can create a pool of processes which will carry out tasks submitted to it
with the <tt class="xref py py-class docutils literal">Pool</tt> class.
<dl class="class">
<dt id="multiprocessing.pool.multiprocessing.Pool">
class <tt class="descclassname">multiprocessing.</tt><tt class="descname">Pool</tt><big>(</big>[processes[, initializer[, initargs[, maxtasksperchild]]]]<big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool" title="Permalink to this definition">¶</a></dt>
<dd>A process pool object which controls a pool of worker processes to which jobs
can be submitted. It supports asynchronous results with timeouts and
callbacks and has a parallel map implementation.
processes is the number of worker processes to use. If processes is
<tt class="docutils literal">None</tt> then the number returned by <tt class="xref py py-func docutils literal">cpu_count()</tt> is used. If
initializer is not <tt class="docutils literal">None</tt> then each worker process will call
<tt class="docutils literal">initializer(*initargs)</tt> when it starts.

New in version 2.7: maxtasksperchild is the number of tasks a worker process can complete
before it will exit and be replaced with a fresh worker process, to enable
unused resources to be freed. The default maxtasksperchild is None, which
means worker processes will live as long as the pool.
<div class="admonition note">
Note
Worker processes within a <tt class="xref py py-class docutils literal">Pool</tt> typically live for the complete
duration of the Pool&#8217;s work queue. A frequent pattern found in other
systems (such as Apache, mod_wsgi, etc) to free resources held by
workers is to allow a worker within a pool to complete only a set
amount of work before being exiting, being cleaned up and a new
process spawned to replace the old one. The maxtasksperchild
argument to the <tt class="xref py py-class docutils literal">Pool</tt> exposes this ability to the end user.
</div>
<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.apply">
<tt class="descname">apply</tt><big>(</big>func[, args[, kwds]]<big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.apply" title="Permalink to this definition">¶</a></dt>
<dd>Equivalent of the <a class="reference internal" href="functions.html#apply" title="apply"><tt class="xref py py-func docutils literal">apply()</tt></a> built-in function. It blocks until the
result is ready, so <a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.apply_async" title="multiprocessing.pool.multiprocessing.Pool.apply_async"><tt class="xref py py-meth docutils literal">apply_async()</tt></a> is better suited for performing
work in parallel. Additionally, func is only executed in one of the
workers of the pool.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.apply_async">
<tt class="descname">apply_async</tt><big>(</big>func[, args[, kwds[, callback]]]<big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.apply_async" title="Permalink to this definition">¶</a></dt>
<dd>A variant of the <a class="reference internal" href="functions.html#apply" title="apply"><tt class="xref py py-meth docutils literal">apply()</tt></a> method which returns a result object.
If callback is specified then it should be a callable which accepts a
single argument. When the result becomes ready callback is applied to
it (unless the call failed). callback should complete immediately since
otherwise the thread which handles the results will get blocked.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.map">
<tt class="descname">map</tt><big>(</big>func, iterable[, chunksize]<big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.map" title="Permalink to this definition">¶</a></dt>
<dd>A parallel equivalent of the <a class="reference internal" href="functions.html#map" title="map"><tt class="xref py py-func docutils literal">map()</tt></a> built-in function (it supports only
one iterable argument though). It blocks until the result is ready.
This method chops the iterable into a number of chunks which it submits to
the process pool as separate tasks. The (approximate) size of these
chunks can be specified by setting chunksize to a positive integer.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.map_async">
<tt class="descname">map_async</tt><big>(</big>func, iterable[, chunksize[, callback]]<big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.map_async" title="Permalink to this definition">¶</a></dt>
<dd>A variant of the <a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.map" title="multiprocessing.pool.multiprocessing.Pool.map"><tt class="xref py py-meth docutils literal">map()</tt></a> method which returns a result object.
If callback is specified then it should be a callable which accepts a
single argument. When the result becomes ready callback is applied to
it (unless the call failed). callback should complete immediately since
otherwise the thread which handles the results will get blocked.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.imap">
<tt class="descname">imap</tt><big>(</big>func, iterable[, chunksize]<big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.imap" title="Permalink to this definition">¶</a></dt>
<dd>An equivalent of <a class="reference internal" href="itertools.html#itertools.imap" title="itertools.imap"><tt class="xref py py-func docutils literal">itertools.imap()</tt></a>.
The chunksize argument is the same as the one used by the <a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.map" title="multiprocessing.pool.multiprocessing.Pool.map"><tt class="xref py py-meth docutils literal">map()</tt></a>
method. For very long iterables using a large value for chunksize can
make the job complete much faster than using the default value of
<tt class="docutils literal">1</tt>.
Also if chunksize is <tt class="docutils literal">1</tt> then the <tt class="xref py py-meth docutils literal">next()</tt> method of the iterator
returned by the <a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.imap" title="multiprocessing.pool.multiprocessing.Pool.imap"><tt class="xref py py-meth docutils literal">imap()</tt></a> method has an optional timeout parameter:
<tt class="docutils literal">next(timeout)</tt> will raise <tt class="xref py py-exc docutils literal">multiprocessing.TimeoutError</tt> if the
result cannot be returned within timeout seconds.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.imap_unordered">
<tt class="descname">imap_unordered</tt><big>(</big>func, iterable[, chunksize]<big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.imap_unordered" title="Permalink to this definition">¶</a></dt>
<dd>The same as <a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.imap" title="multiprocessing.pool.multiprocessing.Pool.imap"><tt class="xref py py-meth docutils literal">imap()</tt></a> except that the ordering of the results from the
returned iterator should be considered arbitrary. (Only when there is
only one worker process is the order guaranteed to be &#8220;correct&#8221;.)
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.close">
<tt class="descname">close</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.close" title="Permalink to this definition">¶</a></dt>
<dd>Prevents any more tasks from being submitted to the pool. Once all the
tasks have been completed the worker processes will exit.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.terminate">
<tt class="descname">terminate</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.terminate" title="Permalink to this definition">¶</a></dt>
<dd>Stops the worker processes immediately without completing outstanding
work. When the pool object is garbage collected <a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.terminate" title="multiprocessing.pool.multiprocessing.Pool.terminate"><tt class="xref py py-meth docutils literal">terminate()</tt></a> will be
called immediately.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.multiprocessing.Pool.join">
<tt class="descname">join</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.pool.multiprocessing.Pool.join" title="Permalink to this definition">¶</a></dt>
<dd>Wait for the worker processes to exit. One must call <a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.close" title="multiprocessing.pool.multiprocessing.Pool.close"><tt class="xref py py-meth docutils literal">close()</tt></a> or
<a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.terminate" title="multiprocessing.pool.multiprocessing.Pool.terminate"><tt class="xref py py-meth docutils literal">terminate()</tt></a> before using <a class="reference internal" href="#multiprocessing.pool.multiprocessing.Pool.join" title="multiprocessing.pool.multiprocessing.Pool.join"><tt class="xref py py-meth docutils literal">join()</tt></a>.
</dd></dl>

</dd></dl>

<dl class="class">
<dt id="multiprocessing.pool.AsyncResult">
class <tt class="descclassname">multiprocessing.pool.</tt><tt class="descname">AsyncResult</tt><a class="headerlink" href="#multiprocessing.pool.AsyncResult" title="Permalink to this definition">¶</a></dt>
<dd>The class of the result returned by <tt class="xref py py-meth docutils literal">Pool.apply_async()</tt> and
<tt class="xref py py-meth docutils literal">Pool.map_async()</tt>.
<dl class="method">
<dt id="multiprocessing.pool.AsyncResult.get">
<tt class="descname">get</tt><big>(</big>[timeout]<big>)</big><a class="headerlink" href="#multiprocessing.pool.AsyncResult.get" title="Permalink to this definition">¶</a></dt>
<dd>Return the result when it arrives. If timeout is not <tt class="docutils literal">None</tt> and the
result does not arrive within timeout seconds then
<tt class="xref py py-exc docutils literal">multiprocessing.TimeoutError</tt> is raised. If the remote call raised
an exception then that exception will be reraised by <a class="reference internal" href="#multiprocessing.pool.AsyncResult.get" title="multiprocessing.pool.AsyncResult.get"><tt class="xref py py-meth docutils literal">get()</tt></a>.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.AsyncResult.wait">
<tt class="descname">wait</tt><big>(</big>[timeout]<big>)</big><a class="headerlink" href="#multiprocessing.pool.AsyncResult.wait" title="Permalink to this definition">¶</a></dt>
<dd>Wait until the result is available or until timeout seconds pass.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.AsyncResult.ready">
<tt class="descname">ready</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.pool.AsyncResult.ready" title="Permalink to this definition">¶</a></dt>
<dd>Return whether the call has completed.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.pool.AsyncResult.successful">
<tt class="descname">successful</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.pool.AsyncResult.successful" title="Permalink to this definition">¶</a></dt>
<dd>Return whether the call completed without raising an exception. Will
raise <a class="reference internal" href="exceptions.html#exceptions.AssertionError" title="exceptions.AssertionError"><tt class="xref py py-exc docutils literal">AssertionError</tt></a> if the result is not ready.
</dd></dl>

</dd></dl>

The following example demonstrates the use of a pool:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Pool

result = pool.apply_async(f, (10,)) # evaluate &quot;f(10)&quot; asynchronously
 print result.get(timeout=1) # prints &quot;100&quot; unless your computer is *very* slow

print pool.map(f, range(10)) # prints &quot;[0, 1, 4,..., 81]&quot;

import time
 result = pool.apply_async(time.sleep, (10,))
 print result.get(timeout=1) # raises TimeoutError
</pre></div>
</div>
</div>
<div class="section" id="module-multiprocessing.connection">
<h3>16.6.2.10. Listeners and Clients<a class="headerlink" href="#module-multiprocessing.connection" title="Permalink to this headline">¶</a></h3>
Usually message passing between processes is done using queues or by using
<tt class="xref py py-class docutils literal">Connection</tt> objects returned by <tt class="xref py py-func docutils literal">Pipe()</tt>.
However, the <a class="reference internal" href="#module-multiprocessing.connection" title="multiprocessing.connection: API for dealing with sockets."><tt class="xref py py-mod docutils literal">multiprocessing.connection</tt></a> module allows some extra
flexibility. It basically gives a high level message oriented API for dealing
with sockets or Windows named pipes, and also has support for digest
authentication using the <a class="reference internal" href="hmac.html#module-hmac" title="hmac: Keyed-Hashing for Message Authentication (HMAC) implementation for Python."><tt class="xref py py-mod docutils literal">hmac</tt></a> module.
<dl class="function">
<dt id="multiprocessing.connection.deliver_challenge">
<tt class="descclassname">multiprocessing.connection.</tt><tt class="descname">deliver_challenge</tt><big>(</big>connection, authkey<big>)</big><a class="headerlink" href="#multiprocessing.connection.deliver_challenge" title="Permalink to this definition">¶</a></dt>
<dd>Send a randomly generated message to the other end of the connection and wait
for a reply.
If the reply matches the digest of the message using authkey as the key
then a welcome message is sent to the other end of the connection. Otherwise
<a class="reference internal" href="#multiprocessing.connection.AuthenticationError" title="multiprocessing.connection.AuthenticationError"><tt class="xref py py-exc docutils literal">AuthenticationError</tt></a> is raised.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.connection.answer_challenge">
<tt class="descclassname">multiprocessing.connection.</tt><tt class="descname">answer_challenge</tt><big>(</big>connection, authkey<big>)</big><a class="headerlink" href="#multiprocessing.connection.answer_challenge" title="Permalink to this definition">¶</a></dt>
<dd>Receive a message, calculate the digest of the message using authkey as the
key, and then send the digest back.
If a welcome message is not received, then <a class="reference internal" href="#multiprocessing.connection.AuthenticationError" title="multiprocessing.connection.AuthenticationError"><tt class="xref py py-exc docutils literal">AuthenticationError</tt></a> is
raised.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.connection.Client">
<tt class="descclassname">multiprocessing.connection.</tt><tt class="descname">Client</tt><big>(</big>address[, family[, authenticate[, authkey]]]<big>)</big><a class="headerlink" href="#multiprocessing.connection.Client" title="Permalink to this definition">¶</a></dt>
<dd>Attempt to set up a connection to the listener which is using address
address, returning a <a class="reference internal" href="#multiprocessing.Connection" title="multiprocessing.Connection"><tt class="xref py py-class docutils literal">Connection</tt></a>.
The type of the connection is determined by family argument, but this can
generally be omitted since it can usually be inferred from the format of
address. (See <a class="reference internal" href="#multiprocessing-address-formats">Address Formats</a>)
If authenticate is <tt class="docutils literal">True</tt> or authkey is a string then digest
authentication is used. The key used for authentication will be either
authkey or <tt class="docutils literal">current_process().authkey)</tt> if authkey is <tt class="docutils literal">None</tt>.
If authentication fails then <a class="reference internal" href="#multiprocessing.connection.AuthenticationError" title="multiprocessing.connection.AuthenticationError"><tt class="xref py py-exc docutils literal">AuthenticationError</tt></a> is raised. See
<a class="reference internal" href="#multiprocessing-auth-keys">Authentication keys</a>.
</dd></dl>

<dl class="class">
<dt id="multiprocessing.connection.Listener">
class <tt class="descclassname">multiprocessing.connection.</tt><tt class="descname">Listener</tt><big>(</big>[address[, family[, backlog[, authenticate[, authkey]]]]]<big>)</big><a class="headerlink" href="#multiprocessing.connection.Listener" title="Permalink to this definition">¶</a></dt>
<dd>A wrapper for a bound socket or Windows named pipe which is &#8216;listening&#8217; for
connections.
address is the address to be used by the bound socket or named pipe of the
listener object.
<div class="admonition note">
Note
If an address of &#8216;0.0.0.0&#8217; is used, the address will not be a connectable
end point on Windows. If you require a connectable end-point,
you should use &#8216;127.0.0.1&#8217;.
</div>
family is the type of socket (or named pipe) to use. This can be one of
the strings <tt class="docutils literal">'AF_INET'</tt> (for a TCP socket), <tt class="docutils literal">'AF_UNIX'</tt> (for a Unix
domain socket) or <tt class="docutils literal">'AF_PIPE'</tt> (for a Windows named pipe). Of these only
the first is guaranteed to be available. If family is <tt class="docutils literal">None</tt> then the
family is inferred from the format of address. If address is also
<tt class="docutils literal">None</tt> then a default is chosen. This default is the family which is
assumed to be the fastest available. See
<a class="reference internal" href="#multiprocessing-address-formats">Address Formats</a>. Note that if family is
<tt class="docutils literal">'AF_UNIX'</tt> and address is <tt class="docutils literal">None</tt> then the socket will be created in a
private temporary directory created using <a class="reference internal" href="tempfile.html#tempfile.mkstemp" title="tempfile.mkstemp"><tt class="xref py py-func docutils literal">tempfile.mkstemp()</tt></a>.
If the listener object uses a socket then backlog (1 by default) is passed
to the <tt class="xref py py-meth docutils literal">listen()</tt> method of the socket once it has been bound.
If authenticate is <tt class="docutils literal">True</tt> (<tt class="docutils literal">False</tt> by default) or authkey is not
<tt class="docutils literal">None</tt> then digest authentication is used.
If authkey is a string then it will be used as the authentication key;
otherwise it must be None.
If authkey is <tt class="docutils literal">None</tt> and authenticate is <tt class="docutils literal">True</tt> then
<tt class="docutils literal">current_process().authkey</tt> is used as the authentication key. If
authkey is <tt class="docutils literal">None</tt> and authenticate is <tt class="docutils literal">False</tt> then no
authentication is done. If authentication fails then
<a class="reference internal" href="#multiprocessing.connection.AuthenticationError" title="multiprocessing.connection.AuthenticationError"><tt class="xref py py-exc docutils literal">AuthenticationError</tt></a> is raised. See <a class="reference internal" href="#multiprocessing-auth-keys">Authentication keys</a>.
<dl class="method">
<dt id="multiprocessing.connection.Listener.accept">
<tt class="descname">accept</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.connection.Listener.accept" title="Permalink to this definition">¶</a></dt>
<dd>Accept a connection on the bound socket or named pipe of the listener
object and return a <tt class="xref py py-class docutils literal">Connection</tt> object. If authentication is
attempted and fails, then <a class="reference internal" href="#multiprocessing.connection.AuthenticationError" title="multiprocessing.connection.AuthenticationError"><tt class="xref py py-exc docutils literal">AuthenticationError</tt></a> is raised.
</dd></dl>

<dl class="method">
<dt id="multiprocessing.connection.Listener.close">
<tt class="descname">close</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.connection.Listener.close" title="Permalink to this definition">¶</a></dt>
<dd>Close the bound socket or named pipe of the listener object. This is
called automatically when the listener is garbage collected. However it
is advisable to call it explicitly.
</dd></dl>

Listener objects have the following read-only properties:
<dl class="attribute">
<dt id="multiprocessing.connection.Listener.address">
<tt class="descname">address</tt><a class="headerlink" href="#multiprocessing.connection.Listener.address" title="Permalink to this definition">¶</a></dt>
<dd>The address which is being used by the Listener object.
</dd></dl>

<dl class="attribute">
<dt id="multiprocessing.connection.Listener.last_accepted">
<tt class="descname">last_accepted</tt><a class="headerlink" href="#multiprocessing.connection.Listener.last_accepted" title="Permalink to this definition">¶</a></dt>
<dd>The address from which the last accepted connection came. If this is
unavailable then it is <tt class="docutils literal">None</tt>.
</dd></dl>

</dd></dl>

The module defines two exceptions:
<dl class="exception">
<dt id="multiprocessing.connection.AuthenticationError">
exception <tt class="descclassname">multiprocessing.connection.</tt><tt class="descname">AuthenticationError</tt><a class="headerlink" href="#multiprocessing.connection.AuthenticationError" title="Permalink to this definition">¶</a></dt>
<dd>Exception raised when there is an authentication error.
</dd></dl>

Examples
The following server code creates a listener which uses <tt class="docutils literal">'secret password'</tt> as
an authentication key. It then waits for a connection and sends some data to
the client:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing.connection import Listener
from array import array

address = (&#39;localhost&#39;, 6000) # family is deduced to be &#39;AF_INET&#39;
listener = Listener(address, authkey=&#39;secret password&#39;)

conn = listener.accept()
print &#39;connection accepted from&#39;, listener.last_accepted

conn.send([2.25, None, &#39;junk&#39;, float])

conn.send_bytes(&#39;hello&#39;)

conn.send_bytes(array(&#39;i&#39;, [42, 1729]))

conn.close()
listener.close()
</pre></div>
</div>
The following code connects to the server and receives some data from the
server:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing.connection import Client
from array import array

address = (&#39;localhost&#39;, 6000)
conn = Client(address, authkey=&#39;secret password&#39;)

print conn.recv() # =&gt; [2.25, None, &#39;junk&#39;, float]

print conn.recv_bytes() # =&gt; &#39;hello&#39;

arr = array(&#39;i&#39;, [0, 0, 0, 0, 0])
print conn.recv_bytes_into(arr) # =&gt; 8
print arr # =&gt; array(&#39;i&#39;, [42, 1729, 0, 0, 0])

conn.close()
</pre></div>
</div>
<div class="section" id="address-formats">
<h4>16.6.2.10.1. Address Formats<a class="headerlink" href="#address-formats" title="Permalink to this headline">¶</a></h4>
<ul>
<li>An <tt class="docutils literal">'AF_INET'</tt> address is a tuple of the form <tt class="docutils literal">(hostname, port)</tt> where
hostname is a string and port is an integer.
</li>
<li>An <tt class="docutils literal">'AF_UNIX'</tt> address is a string representing a filename on the
filesystem.
</li>
<li><dl class="first docutils">
<dt>An <tt class="docutils literal">'AF_PIPE'</tt> address is a string of the form</dt>
<dd><tt class="samp docutils literal">r'\\.\pipe\PipeName'</tt>. To use <a class="reference internal" href="#multiprocessing.connection.Client" title="multiprocessing.connection.Client"><tt class="xref py py-func docutils literal">Client()</tt></a> to connect to a named
pipe on a remote computer called ServerName one should use an address of the
form <tt class="samp docutils literal">r'\\ServerName\pipe\PipeName'</tt> instead.
</dd>
</dl>
</li>
</ul>
Note that any string beginning with two backslashes is assumed by default to be
an <tt class="docutils literal">'AF_PIPE'</tt> address rather than an <tt class="docutils literal">'AF_UNIX'</tt> address.
</div>
</div>
<div class="section" id="authentication-keys">
<h3>16.6.2.11. Authentication keys<a class="headerlink" href="#authentication-keys" title="Permalink to this headline">¶</a></h3>
When one uses <tt class="xref py py-meth docutils literal">Connection.recv()</tt>, the data received is automatically
unpickled. Unfortunately unpickling data from an untrusted source is a security
risk. Therefore <a class="reference internal" href="#multiprocessing.connection.Listener" title="multiprocessing.connection.Listener"><tt class="xref py py-class docutils literal">Listener</tt></a> and <a class="reference internal" href="#multiprocessing.connection.Client" title="multiprocessing.connection.Client"><tt class="xref py py-func docutils literal">Client()</tt></a> use the <a class="reference internal" href="hmac.html#module-hmac" title="hmac: Keyed-Hashing for Message Authentication (HMAC) implementation for Python."><tt class="xref py py-mod docutils literal">hmac</tt></a> module
to provide digest authentication.
An authentication key is a string which can be thought of as a password: once a
connection is established both ends will demand proof that the other knows the
authentication key. (Demonstrating that both ends are using the same key does
not involve sending the key over the connection.)
If authentication is requested but do authentication key is specified then the
return value of <tt class="docutils literal">current_process().authkey</tt> is used (see
<a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a>). This value will automatically inherited by
any <a class="reference internal" href="#multiprocessing.Process" title="multiprocessing.Process"><tt class="xref py py-class docutils literal">Process</tt></a> object that the current process creates.
This means that (by default) all processes of a multi-process program will share
a single authentication key which can be used when setting up connections
between themselves.
Suitable authentication keys can also be generated by using <a class="reference internal" href="os.html#os.urandom" title="os.urandom"><tt class="xref py py-func docutils literal">os.urandom()</tt></a>.
</div>
<div class="section" id="logging">
<h3>16.6.2.12. Logging<a class="headerlink" href="#logging" title="Permalink to this headline">¶</a></h3>
Some support for logging is available. Note, however, that the <a class="reference internal" href="logging.html#module-logging" title="logging: Flexible event logging system for applications."><tt class="xref py py-mod docutils literal">logging</tt></a>
package does not use process shared locks so it is possible (depending on the
handler type) for messages from different processes to get mixed up.
<dl class="function">
<dt id="multiprocessing.get_logger">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">get_logger</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.get_logger" title="Permalink to this definition">¶</a></dt>
<dd>Returns the logger used by <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a>. If necessary, a new one
will be created.
When first created the logger has level <tt class="xref py py-data docutils literal">logging.NOTSET</tt> and no
default handler. Messages sent to this logger will not by default propagate
to the root logger.
Note that on Windows child processes will only inherit the level of the
parent process&#8217;s logger &#8211; any other customization of the logger will not be
inherited.
</dd></dl>

<dl class="function">
<dt id="multiprocessing.log_to_stderr">
<tt class="descclassname">multiprocessing.</tt><tt class="descname">log_to_stderr</tt><big>(</big><big>)</big><a class="headerlink" href="#multiprocessing.log_to_stderr" title="Permalink to this definition">¶</a></dt>
<dd>This function performs a call to <a class="reference internal" href="#multiprocessing.get_logger" title="multiprocessing.get_logger"><tt class="xref py py-func docutils literal">get_logger()</tt></a> but in addition to
returning the logger created by get_logger, it adds a handler which sends
output to <a class="reference internal" href="sys.html#sys.stderr" title="sys.stderr"><tt class="xref py py-data docutils literal">sys.stderr</tt></a> using format
<tt class="docutils literal">'[%(levelname)s/%(processName)s] %(message)s'</tt>.
</dd></dl>

Below is an example session with logging turned on:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; import multiprocessing, logging
&gt;&gt;&gt; logger = multiprocessing.log_to_stderr()
&gt;&gt;&gt; logger.setLevel(logging.INFO)
&gt;&gt;&gt; logger.warning(&#39;doomed&#39;)
[WARNING/MainProcess] doomed
&gt;&gt;&gt; m = multiprocessing.Manager()
[INFO/SyncManager-...] child process calling self.run()
[INFO/SyncManager-...] created temp directory /.../pymp-...
[INFO/SyncManager-...] manager serving at &#39;/.../listener-...&#39;
&gt;&gt;&gt; del m
[INFO/MainProcess] sending shutdown message to manager
[INFO/SyncManager-...] manager exiting with exitcode 0
</pre></div>
</div>
In addition to having these two logging functions, the multiprocessing also
exposes two additional logging level attributes. These are <tt class="xref py py-const docutils literal">SUBWARNING</tt>
and <tt class="xref py py-const docutils literal">SUBDEBUG</tt>. The table below illustrates where theses fit in the
normal level hierarchy.
<table border="1" class="docutils">
<colgroup>
<col width="50%" />
<col width="50%" />
</colgroup>
<thead valign="bottom">
<tr class="row-odd"><th class="head">Level</th>
<th class="head">Numeric value</th>
</tr>
</thead>
<tbody valign="top">
<tr class="row-even"><td><tt class="docutils literal">SUBWARNING</tt></td>
<td>25</td>
</tr>
<tr class="row-odd"><td><tt class="docutils literal">SUBDEBUG</tt></td>
<td>5</td>
</tr>
</tbody>
</table>
For a full table of logging levels, see the <a class="reference internal" href="logging.html#module-logging" title="logging: Flexible event logging system for applications."><tt class="xref py py-mod docutils literal">logging</tt></a> module.
These additional logging levels are used primarily for certain debug messages
within the multiprocessing module. Below is the same example as above, except
with <tt class="xref py py-const docutils literal">SUBDEBUG</tt> enabled:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; import multiprocessing, logging
&gt;&gt;&gt; logger = multiprocessing.log_to_stderr()
&gt;&gt;&gt; logger.setLevel(multiprocessing.SUBDEBUG)
&gt;&gt;&gt; logger.warning(&#39;doomed&#39;)
[WARNING/MainProcess] doomed
&gt;&gt;&gt; m = multiprocessing.Manager()
[INFO/SyncManager-...] child process calling self.run()
[INFO/SyncManager-...] created temp directory /.../pymp-...
[INFO/SyncManager-...] manager serving at &#39;/.../pymp-djGBXN/listener-...&#39;
&gt;&gt;&gt; del m
[SUBDEBUG/MainProcess] finalizer calling ...
[INFO/MainProcess] sending shutdown message to manager
[DEBUG/SyncManager-...] manager received shutdown message
[SUBDEBUG/SyncManager-...] calling &lt;Finalize object, callback=unlink, ...
[SUBDEBUG/SyncManager-...] finalizer calling &lt;built-in function unlink&gt; ...
[SUBDEBUG/SyncManager-...] calling &lt;Finalize object, dead&gt;
[SUBDEBUG/SyncManager-...] finalizer calling &lt;function rmtree at 0x5aa730&gt; ...
[INFO/SyncManager-...] manager exiting with exitcode 0
</pre></div>
</div>
</div>
<div class="section" id="module-multiprocessing.dummy">
<h3>16.6.2.13. The <a class="reference internal" href="#module-multiprocessing.dummy" title="multiprocessing.dummy: Dumb wrapper around threading."><tt class="xref py py-mod docutils literal">multiprocessing.dummy</tt></a> module<a class="headerlink" href="#module-multiprocessing.dummy" title="Permalink to this headline">¶</a></h3>
<a class="reference internal" href="#module-multiprocessing.dummy" title="multiprocessing.dummy: Dumb wrapper around threading."><tt class="xref py py-mod docutils literal">multiprocessing.dummy</tt></a> replicates the API of <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> but is
no more than a wrapper around the <a class="reference internal" href="threading.html#module-threading" title="threading: Higher-level threading interface."><tt class="xref py py-mod docutils literal">threading</tt></a> module.
</div>
</div>
<div class="section" id="programming-guidelines">
<h2>16.6.3. Programming guidelines<a class="headerlink" href="#programming-guidelines" title="Permalink to this headline">¶</a></h2>
There are certain guidelines and idioms which should be adhered to when using
<a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a>.
<div class="section" id="all-platforms">
<h3>16.6.3.1. All platforms<a class="headerlink" href="#all-platforms" title="Permalink to this headline">¶</a></h3>
Avoid shared state
<blockquote>
<div>As far as possible one should try to avoid shifting large amounts of data
between processes.
It is probably best to stick to using queues or pipes for communication
between processes rather than using the lower level synchronization
primitives from the <a class="reference internal" href="threading.html#module-threading" title="threading: Higher-level threading interface."><tt class="xref py py-mod docutils literal">threading</tt></a> module.
</div></blockquote>
Picklability
<blockquote>
<div>Ensure that the arguments to the methods of proxies are picklable.</div></blockquote>
Thread safety of proxies
<blockquote>
<div>Do not use a proxy object from more than one thread unless you protect it
with a lock.
(There is never a problem with different processes using the same proxy.)
</div></blockquote>
Joining zombie processes
<blockquote>
<div>On Unix when a process finishes but has not been joined it becomes a zombie.
There should never be very many because each time a new process starts (or
<tt class="xref py py-func docutils literal">active_children()</tt> is called) all completed processes which have not
yet been joined will be joined. Also calling a finished process&#8217;s
<tt class="xref py py-meth docutils literal">Process.is_alive()</tt> will join the process. Even so it is probably good
practice to explicitly join all the processes that you start.</div></blockquote>
Better to inherit than pickle/unpickle
<blockquote>
<div>On Windows many types from <a class="reference internal" href="#module-multiprocessing" title="multiprocessing: Process-based &quot;threading&quot; interface."><tt class="xref py py-mod docutils literal">multiprocessing</tt></a> need to be picklable so
that child processes can use them. However, one should generally avoid
sending shared objects to other processes using pipes or queues. Instead
you should arrange the program so that a process which needs access to a
shared resource created elsewhere can inherit it from an ancestor process.</div></blockquote>
Avoid terminating processes
<blockquote>
<div>Using the <tt class="xref py py-meth docutils literal">Process.terminate()</tt> method to stop a process is liable to
cause any shared resources (such as locks, semaphores, pipes and queues)
currently being used by the process to become broken or unavailable to other
processes.
Therefore it is probably best to only consider using
<tt class="xref py py-meth docutils literal">Process.terminate()</tt> on processes which never use any shared resources.
</div></blockquote>
Joining processes that use queues
<blockquote>
<div>Bear in mind that a process that has put items in a queue will wait before
terminating until all the buffered items are fed by the &#8220;feeder&#8221; thread to
the underlying pipe. (The child process can call the
<a class="reference internal" href="#multiprocessing.Queue.cancel_join_thread" title="multiprocessing.Queue.cancel_join_thread"><tt class="xref py py-meth docutils literal">cancel_join_thread()</tt></a> method of the queue to avoid this behaviour.)
This means that whenever you use a queue you need to make sure that all
items which have been put on the queue will eventually be removed before the
process is joined. Otherwise you cannot be sure that processes which have
put items on the queue will terminate. Remember also that non-daemonic
processes will be automatically be joined.
An example which will deadlock is the following:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Process, Queue

def f(q):
 q.put(&#39;X&#39; * 1000000)

def f():
 ... do something using &quot;lock&quot; ...

def f(l):
 ... do something using &quot;l&quot; ...

def foo():
 print &#39;hello&#39;

p = Process(target=foo)
p.start()
</pre></div>
</div>
Instead one should protect the &#8220;entry point&#8221; of the program by using <tt class="docutils literal">if
__name__ == '__main__':</tt> as follows:
<div class="highlight-python"><div class="highlight"><pre>from multiprocessing import Process, freeze_support

def foo():
 print &#39;hello&#39;

from multiprocessing import freeze_support
from multiprocessing.managers import BaseManager, BaseProxy
import operator

##

class Foo(object):
 def f(self):
 print &#39;you called Foo.f()&#39;
 def g(self):
 print &#39;you called Foo.g()&#39;
 def _h(self):
 print &#39;you called Foo._h()&#39;

# A simple generator function
def baz():
 for i in xrange(10):
 yield i*i

# Proxy type for generator objects
class GeneratorProxy(BaseProxy):
 _exposed_ = (&#39;next&#39;, &#39;__next__&#39;)
 def __iter__(self):
 return self
 def next(self):
 return self._callmethod(&#39;next&#39;)
 def __next__(self):
 return self._callmethod(&#39;__next__&#39;)

# Function to return the operator module
def get_operator_module():
 return operator

##

class MyManager(BaseManager):
 pass

# register the Foo class; make `f()` and `g()` accessible via proxy
MyManager.register(&#39;Foo1&#39;, Foo)

# register the Foo class; make `g()` and `_h()` accessible via proxy
MyManager.register(&#39;Foo2&#39;, Foo, exposed=(&#39;g&#39;, &#39;_h&#39;))

# register the generator function baz; use `GeneratorProxy` to make proxies
MyManager.register(&#39;baz&#39;, baz, proxytype=GeneratorProxy)

# register get_operator_module(); make public functions accessible via proxy
MyManager.register(&#39;operator&#39;, get_operator_module)

##

def test():
 manager = MyManager()
 manager.start()

print &#39;-&#39; * 20

f1 = manager.Foo1()
 f1.f()
 f1.g()
 assert not hasattr(f1, &#39;_h&#39;)
 assert sorted(f1._exposed_) == sorted([&#39;f&#39;, &#39;g&#39;])

print &#39;-&#39; * 20

f2 = manager.Foo2()
 f2.g()
 f2._h()
 assert not hasattr(f2, &#39;f&#39;)
 assert sorted(f2._exposed_) == sorted([&#39;g&#39;, &#39;_h&#39;])

print &#39;-&#39; * 20

it = manager.baz()
 for i in it:
 print &#39;&lt;%d&gt;&#39; % i,
 print

print &#39;-&#39; * 20

op = manager.operator()
 print &#39;op.add(23, 45) =&#39;, op.add(23, 45)
 print &#39;op.pow(2, 94) =&#39;, op.pow(2, 94)
 print &#39;op.getslice(range(10), 2, 6) =&#39;, op.getslice(range(10), 2, 6)
 print &#39;op.repeat(range(5), 3) =&#39;, op.repeat(range(5), 3)
 print &#39;op._exposed_ =&#39;, op._exposed_

##

import multiprocessing
import time
import random
import sys

#
# Functions used by test code
#

def calculate(func, args):
 result = func(*args)
 return &#39;%s says that %s%s = %s&#39; % (
 multiprocessing.current_process().name,
 func.__name__, args, result
 )

def calculatestar(args):
 return calculate(*args)

def mul(a, b):
 time.sleep(0.5*random.random())
 return a * b

def plus(a, b):
 time.sleep(0.5*random.random())
 return a + b

def f(x):
 return 1.0 / (x-5.0)

def pow3(x):
 return x**3

def noop(x):
 pass

#
# Test code
#

def test():
 print &#39;cpu_count() = %d\n&#39; % multiprocessing.cpu_count()

#
 # Create pool
 #

PROCESSES = 4
 print &#39;Creating pool with %d processes\n&#39; % PROCESSES
 pool = multiprocessing.Pool(PROCESSES)
 print &#39;pool = %s&#39; % pool
 print

#
 # Tests
 #

TASKS = [(mul, (i, 7)) for i in range(10)] + \
 [(plus, (i, 8)) for i in range(10)]

results = [pool.apply_async(calculate, t) for t in TASKS]
 imap_it = pool.imap(calculatestar, TASKS)
 imap_unordered_it = pool.imap_unordered(calculatestar, TASKS)

print &#39;Ordered results using pool.apply_async():&#39;
 for r in results:
 print &#39;\t&#39;, r.get()
 print

print &#39;Ordered results using pool.imap():&#39;
 for x in imap_it:
 print &#39;\t&#39;, x
 print

print &#39;Unordered results using pool.imap_unordered():&#39;
 for x in imap_unordered_it:
 print &#39;\t&#39;, x
 print

print &#39;Ordered results using pool.map() --- will block till complete:&#39;
 for x in pool.map(calculatestar, TASKS):
 print &#39;\t&#39;, x
 print

#
 # Simple benchmarks
 #

N = 100000
 print &#39;def pow3(x): return x**3&#39;

t = time.time()
 B = pool.map(pow3, xrange(N))
 print &#39;\tpool.map(pow3, xrange(%d)):\n\t\t%s seconds&#39; % \
 (N, time.time() - t)

t = time.time()
 C = list(pool.imap(pow3, xrange(N), chunksize=N//8))
 print &#39;\tlist(pool.imap(pow3, xrange(%d), chunksize=%d)):\n\t\t%s&#39; \
 &#39; seconds&#39; % (N, N//8, time.time() - t)

assert A == B == C, (len(A), len(B), len(C))
 print

L = [None] * 1000000
 print &#39;def noop(x): pass&#39;
 print &#39;L = [None] * 1000000&#39;

t = time.time()
 C = list(pool.imap(noop, L, chunksize=len(L)//8))
 print &#39;\tlist(pool.imap(noop, L, chunksize=%d)):\n\t\t%s seconds&#39; % \
 (len(L)//8, time.time() - t)

del A, B, C, L

#
 # Test error handling
 #

print &#39;Testing error handling:&#39;

try:
 print pool.apply(f, (5,))
 except ZeroDivisionError:
 print &#39;\tGot ZeroDivisionError as expected from pool.apply()&#39;
 else:
 raise AssertionError(&#39;expected ZeroDivisionError&#39;)

try:
 print pool.map(f, range(10))
 except ZeroDivisionError:
 print &#39;\tGot ZeroDivisionError as expected from pool.map()&#39;
 else:
 raise AssertionError(&#39;expected ZeroDivisionError&#39;)

try:
 print list(pool.imap(f, range(10)))
 except ZeroDivisionError:
 print &#39;\tGot ZeroDivisionError as expected from list(pool.imap())&#39;
 else:
 raise AssertionError(&#39;expected ZeroDivisionError&#39;)

it = pool.imap(f, range(10))
 for i in range(10):
 try:
 x = it.next()
 except ZeroDivisionError:
 if i == 5:
 pass
 except StopIteration:
 break
 else:
 if i == 5:
 raise AssertionError(&#39;expected ZeroDivisionError&#39;)

assert i == 9
 print &#39;\tGot ZeroDivisionError as expected from IMapIterator.next()&#39;
 print

#
 # Testing timeouts
 #

print &#39;Testing ApplyResult.get() with timeout:&#39;,
 res = pool.apply_async(calculate, TASKS[0])
 while 1:
 sys.stdout.flush()
 try:
 sys.stdout.write(&#39;\n\t%s&#39; % res.get(0.02))
 break
 except multiprocessing.TimeoutError:
 sys.stdout.write(&#39;.&#39;)
 print
 print

print &#39;Testing IMapIterator.next() with timeout:&#39;,
 it = pool.imap(calculatestar, TASKS)
 while 1:
 sys.stdout.flush()
 try:
 sys.stdout.write(&#39;\n\t%s&#39; % it.next(0.02))
 except StopIteration:
 break
 except multiprocessing.TimeoutError:
 sys.stdout.write(&#39;.&#39;)
 print
 print

#
 # Testing callback
 #

print &#39;Testing callback:&#39;

A = []
 B = [56, 0, 1, 8, 27, 64, 125, 216, 343, 512, 729]

r = pool.apply_async(mul, (7, 8), callback=A.append)
 r.wait()

r = pool.map_async(pow3, range(10), callback=A.extend)
 r.wait()

if A == B:
 print &#39;\tcallbacks succeeded\n&#39;
 else:
 print &#39;\t*** callbacks failed\n\t\t%s != %s\n&#39; % (A, B)

#
 # Check there are no outstanding tasks
 #

assert not pool._cache, &#39;cache = %r&#39; % pool._cache

#
 # Check close() methods
 #

print &#39;Testing close():&#39;

for worker in pool._pool:
 assert worker.is_alive()

result = pool.apply_async(time.sleep, [0.5])
 pool.close()
 pool.join()

assert result.get() is None

for worker in pool._pool:
 assert not worker.is_alive()

print &#39;\tclose() succeeded\n&#39;

#
 # Check terminate() method
 #

print &#39;Testing terminate():&#39;

pool = multiprocessing.Pool(2)
 DELTA = 0.1
 ignore = pool.apply(pow3, [2])
 results = [pool.apply_async(time.sleep, [DELTA]) for i in range(100)]
 pool.terminate()
 pool.join()

for worker in pool._pool:
 assert not worker.is_alive()

print &#39;\tterminate() succeeded\n&#39;

#
 # Check garbage collection
 #

print &#39;Testing garbage collection:&#39;

pool = multiprocessing.Pool(2)
 DELTA = 0.1
 processes = pool._pool
 ignore = pool.apply(pow3, [2])
 results = [pool.apply_async(time.sleep, [DELTA]) for i in range(100)]

results = pool = None

time.sleep(DELTA * 2)

for worker in processes:
 assert not worker.is_alive()

print &#39;\tgarbage collection succeeded\n&#39;

assert len(sys.argv) in (1, 2)

test()
</pre></div>
</div>
Synchronization types like locks, conditions and queues:
<div class="highlight-python"><div class="highlight"><pre>#
# A test file for the `multiprocessing` package
#
# Copyright (c) 2006-2008, R Oudkerk
# All rights reserved.
#

import time, sys, random
from Queue import Empty

import multiprocessing # may get overwritten

#### TEST_VALUE

def value_func(running, mutex):
 random.seed()
 time.sleep(random.random()*4)

mutex.acquire()
 print &#39;\n\t\t\t&#39; + str(multiprocessing.current_process()) + &#39; has finished&#39;
 running.value -= 1
 mutex.release()

def test_value():
 TASKS = 10
 running = multiprocessing.Value(&#39;i&#39;, TASKS)
 mutex = multiprocessing.Lock()

for i in range(TASKS):
 p = multiprocessing.Process(target=value_func, args=(running, mutex))
 p.start()

while running.value &gt; 0:
 time.sleep(0.08)
 mutex.acquire()
 print running.value,
 sys.stdout.flush()
 mutex.release()

print
 print &#39;No more running processes&#39;

#### TEST_QUEUE

def queue_func(queue):
 for i in range(30):
 time.sleep(0.5 * random.random())
 queue.put(i*i)
 queue.put(&#39;STOP&#39;)

def test_queue():
 q = multiprocessing.Queue()

o = None
 while o != &#39;STOP&#39;:
 try:
 o = q.get(timeout=0.3)
 print o,
 sys.stdout.flush()
 except Empty:
 print &#39;TIMEOUT&#39;

print

#### TEST_CONDITION

def condition_func(cond):
 cond.acquire()
 print &#39;\t&#39; + str(cond)
 time.sleep(2)
 print &#39;\tchild is notifying&#39;
 print &#39;\t&#39; + str(cond)
 cond.notify()
 cond.release()

def test_condition():
 cond = multiprocessing.Condition()

cond.acquire()
 print cond
 cond.acquire()
 print cond

p.start()

print &#39;main is waiting&#39;
 cond.wait()
 print &#39;main has woken up&#39;

print cond
 cond.release()
 print cond
 cond.release()

p.join()
 print cond

#### TEST_SEMAPHORE

def semaphore_func(sema, mutex, running):
 sema.acquire()

mutex.acquire()
 running.value += 1
 print running.value, &#39;tasks are running&#39;
 mutex.release()

random.seed()
 time.sleep(random.random()*2)

mutex.acquire()
 running.value -= 1
 print &#39;%s has finished&#39; % multiprocessing.current_process()
 mutex.release()

sema.release()

def test_semaphore():
 sema = multiprocessing.Semaphore(3)
 mutex = multiprocessing.RLock()
 running = multiprocessing.Value(&#39;i&#39;, 0)

processes = [
 multiprocessing.Process(target=semaphore_func,
 args=(sema, mutex, running))
 for i in range(10)
 ]

for p in processes:
 p.start()

#### TEST_JOIN_TIMEOUT

def join_timeout_func():
 print &#39;\tchild sleeping&#39;
 time.sleep(5.5)
 print &#39;\n\tchild terminating&#39;

def test_join_timeout():
 p = multiprocessing.Process(target=join_timeout_func)
 p.start()

print &#39;waiting for process to finish&#39;

while 1:
 p.join(timeout=1)
 if not p.is_alive():
 break
 print &#39;.&#39;,
 sys.stdout.flush()

#### TEST_EVENT

def event_func(event):
 print &#39;\t%r is waiting&#39; % multiprocessing.current_process()
 event.wait()
 print &#39;\t%r has woken up&#39; % multiprocessing.current_process()

def test_event():
 event = multiprocessing.Event()

processes = [multiprocessing.Process(target=event_func, args=(event,))
 for i in range(5)]

print &#39;main is sleeping&#39;
 time.sleep(2)

print &#39;main is setting event&#39;
 event.set()

#### TEST_SHAREDVALUES

def sharedvalues_func(values, arrays, shared_values, shared_arrays):
 for i in range(len(values)):
 v = values[i][1]
 sv = shared_values[i].value
 assert v == sv

for i in range(len(values)):
 a = arrays[i][1]
 sa = list(shared_arrays[i][:])
 assert a == sa

print &#39;Tests passed&#39;

def test_sharedvalues():
 values = [
 (&#39;i&#39;, 10),
 (&#39;h&#39;, -2),
 (&#39;d&#39;, 1.25)
 ]
 arrays = [
 (&#39;i&#39;, range(100)),
 (&#39;d&#39;, [0.25 * i for i in range(100)]),
 (&#39;H&#39;, range(1000))
 ]

shared_values = [multiprocessing.Value(id, v) for id, v in values]
 shared_arrays = [multiprocessing.Array(id, a) for id, a in arrays]

p = multiprocessing.Process(
 target=sharedvalues_func,
 args=(values, arrays, shared_values, shared_arrays)
 )
 p.start()
 p.join()

assert p.exitcode == 0

####

def test(namespace=multiprocessing):
 global multiprocessing

multiprocessing = namespace

for func in [ test_value, test_queue, test_condition,
 test_semaphore, test_join_timeout, test_event,
 test_sharedvalues ]:

print &#39;\n\t######## %s\n&#39; % func.__name__
 func()

ignore = multiprocessing.active_children() # cleanup any old processes
 if hasattr(multiprocessing, &#39;_debug_info&#39;):
 info = multiprocessing._debug_info()
 if info:
 print info
 raise ValueError(&#39;there should be no positive refcounts left&#39;)

if len(sys.argv) == 1 or sys.argv[1] == &#39;processes&#39;:
 print &#39; Using processes &#39;.center(79, &#39;-&#39;)
 namespace = multiprocessing
 elif sys.argv[1] == &#39;manager&#39;:
 print &#39; Using processes and a manager &#39;.center(79, &#39;-&#39;)
 namespace = multiprocessing.Manager()
 namespace.Process = multiprocessing.Process
 namespace.current_process = multiprocessing.current_process
 namespace.active_children = multiprocessing.active_children
 elif sys.argv[1] == &#39;threads&#39;:
 print &#39; Using threads &#39;.center(79, &#39;-&#39;)
 import multiprocessing.dummy as namespace
 else:
 print &#39;Usage:\n\t%s [processes | manager | threads]&#39; % sys.argv[0]
 raise SystemExit(2)

test(namespace)
</pre></div>
</div>
An example showing how to use queues to feed tasks to a collection of worker
processes and collect the results:
<div class="highlight-python"><div class="highlight"><pre>#
# Simple example which uses a pool of workers to carry out some tasks.
#
# Notice that the results will probably not come out of the output
# queue in the same in the same order as the corresponding tasks were
# put on the input queue. If it is important to get the results back
# in the original order then consider using `Pool.map()` or
# `Pool.imap()` (which will save on the amount of code needed anyway).
#
# Copyright (c) 2006-2008, R Oudkerk
# All rights reserved.
#

import time
import random

#
# Function run by worker processes
#

def worker(input, output):
 for func, args in iter(input.get, &#39;STOP&#39;):
 result = calculate(func, args)
 output.put(result)

#
# Function used to calculate result
#

def calculate(func, args):
 result = func(*args)
 return &#39;%s says that %s%s = %s&#39; % \
 (current_process().name, func.__name__, args, result)

#
# Functions referenced by tasks
#

#
#
#

def test():
 NUMBER_OF_PROCESSES = 4
 TASKS1 = [(mul, (i, 7)) for i in range(20)]
 TASKS2 = [(plus, (i, 8)) for i in range(10)]

# Create queues
 task_queue = Queue()
 done_queue = Queue()

# Submit tasks
 for task in TASKS1:
 task_queue.put(task)

# Start worker processes
 for i in range(NUMBER_OF_PROCESSES):
 Process(target=worker, args=(task_queue, done_queue)).start()

# Get and print results
 print &#39;Unordered results:&#39;
 for i in range(len(TASKS1)):
 print &#39;\t&#39;, done_queue.get()

# Add more tasks using `put()`
 for task in TASKS2:
 task_queue.put(task)

# Get and print some more results
 for i in range(len(TASKS2)):
 print &#39;\t&#39;, done_queue.get()

# Tell child processes to stop
 for i in range(NUMBER_OF_PROCESSES):
 task_queue.put(&#39;STOP&#39;)

import os
import sys

from multiprocessing import Process, current_process, freeze_support
from BaseHTTPServer import HTTPServer
from SimpleHTTPServer import SimpleHTTPRequestHandler

def note(format, *args):
 sys.stderr.write(&#39;[%s]\t%s\n&#39; % (current_process().name, format%args))

class RequestHandler(SimpleHTTPRequestHandler):
 # we override log_message() to show which process is handling the request
 def log_message(self, format, *args):
 note(format, *args)

def serve_forever(server):
 note(&#39;starting server&#39;)
 try:
 server.serve_forever()
 except KeyboardInterrupt:
 pass

def runpool(address, number_of_processes):
 # create a single server object -- children will each inherit a copy
 server = HTTPServer(address, RequestHandler)

# create child processes to act as workers
 for i in range(number_of_processes-1):
 Process(target=serve_forever, args=(server,)).start()

# main process also acts as a worker
 serve_forever(server)

def test():
 DIR = os.path.join(os.path.dirname(__file__), &#39;..&#39;)
 ADDRESS = (&#39;localhost&#39;, 8000)
 NUMBER_OF_PROCESSES = 4

print &#39;Serving at http://%s:%d using %d worker processes&#39; % \
 (ADDRESS[0], ADDRESS[1], NUMBER_OF_PROCESSES)
 print &#39;To exit press Ctrl-&#39; + [&#39;C&#39;, &#39;Break&#39;][sys.platform==&#39;win32&#39;]

os.chdir(DIR)
 runpool(ADDRESS, NUMBER_OF_PROCESSES)

import time, sys, multiprocessing, threading, Queue, gc

if sys.platform == &#39;win32&#39;:
 _timer = time.clock
else:
 _timer = time.time

delta = 1

#### TEST_QUEUESPEED

def queuespeed_func(q, c, iterations):
 a = &#39;0&#39; * 256
 c.acquire()
 c.notify()
 c.release()

for i in xrange(iterations):
 q.put(a)

q.put(&#39;STOP&#39;)

def test_queuespeed(Process, q, c):
 elapsed = 0
 iterations = 1

while elapsed &lt; delta:
 iterations *= 2

p = Process(target=queuespeed_func, args=(q, c, iterations))
 c.acquire()
 p.start()
 c.wait()
 c.release()

result = None
 t = _timer()

while result != &#39;STOP&#39;:
 result = q.get()

elapsed = _timer() - t

p.join()

print iterations, &#39;objects passed through the queue in&#39;, elapsed, &#39;seconds&#39;
 print &#39;average number/sec:&#39;, iterations/elapsed

#### TEST_PIPESPEED

def pipe_func(c, cond, iterations):
 a = &#39;0&#39; * 256
 cond.acquire()
 cond.notify()
 cond.release()

for i in xrange(iterations):
 c.send(a)

c.send(&#39;STOP&#39;)

def test_pipespeed():
 c, d = multiprocessing.Pipe()
 cond = multiprocessing.Condition()
 elapsed = 0
 iterations = 1

p = multiprocessing.Process(target=pipe_func,
 args=(d, cond, iterations))
 cond.acquire()
 p.start()
 cond.wait()
 cond.release()

result = None
 t = _timer()

while result != &#39;STOP&#39;:
 result = c.recv()

elapsed = _timer() - t
 p.join()

print iterations, &#39;objects passed through connection in&#39;,elapsed,&#39;seconds&#39;
 print &#39;average number/sec:&#39;, iterations/elapsed

#### TEST_SEQSPEED

def test_seqspeed(seq):
 elapsed = 0
 iterations = 1

t = _timer()

for i in xrange(iterations):
 a = seq[5]

elapsed = _timer()-t

print iterations, &#39;iterations in&#39;, elapsed, &#39;seconds&#39;
 print &#39;average number/sec:&#39;, iterations/elapsed

#### TEST_LOCK

def test_lockspeed(l):
 elapsed = 0
 iterations = 1

t = _timer()

for i in xrange(iterations):
 l.acquire()
 l.release()

elapsed = _timer()-t

#### TEST_CONDITION

def conditionspeed_func(c, N):
 c.acquire()
 c.notify()

for i in xrange(N):
 c.wait()
 c.notify()

c.release()

def test_conditionspeed(Process, c):
 elapsed = 0
 iterations = 1

c.acquire()
 p = Process(target=conditionspeed_func, args=(c, iterations))
 p.start()

c.wait()

t = _timer()

for i in xrange(iterations):
 c.notify()
 c.wait()

elapsed = _timer()-t

c.release()
 p.join()

print iterations * 2, &#39;waits in&#39;, elapsed, &#39;seconds&#39;
 print &#39;average number/sec:&#39;, iterations * 2 / elapsed

####

def test():
 manager = multiprocessing.Manager()

gc.disable()

print &#39;\n\t######## testing Queue.Queue\n&#39;
 test_queuespeed(threading.Thread, Queue.Queue(),
 threading.Condition())
 print &#39;\n\t######## testing multiprocessing.Queue\n&#39;
 test_queuespeed(multiprocessing.Process, multiprocessing.Queue(),
 multiprocessing.Condition())
 print &#39;\n\t######## testing Queue managed by server process\n&#39;
 test_queuespeed(multiprocessing.Process, manager.Queue(),
 manager.Condition())
 print &#39;\n\t######## testing multiprocessing.Pipe\n&#39;
 test_pipespeed()

print

print &#39;\n\t######## testing list\n&#39;
 test_seqspeed(range(10))
 print &#39;\n\t######## testing list managed by server process\n&#39;
 test_seqspeed(manager.list(range(10)))
 print &#39;\n\t######## testing Array(&quot;i&quot;, ..., lock=False)\n&#39;
 test_seqspeed(multiprocessing.Array(&#39;i&#39;, range(10), lock=False))
 print &#39;\n\t######## testing Array(&quot;i&quot;, ..., lock=True)\n&#39;
 test_seqspeed(multiprocessing.Array(&#39;i&#39;, range(10), lock=True))

print

print &#39;\n\t######## testing threading.Lock\n&#39;
 test_lockspeed(threading.Lock())
 print &#39;\n\t######## testing threading.RLock\n&#39;
 test_lockspeed(threading.RLock())
 print &#39;\n\t######## testing multiprocessing.Lock\n&#39;
 test_lockspeed(multiprocessing.Lock())
 print &#39;\n\t######## testing multiprocessing.RLock\n&#39;
 test_lockspeed(multiprocessing.RLock())
 print &#39;\n\t######## testing lock managed by server process\n&#39;
 test_lockspeed(manager.Lock())
 print &#39;\n\t######## testing rlock managed by server process\n&#39;
 test_lockspeed(manager.RLock())

print

print &#39;\n\t######## testing threading.Condition\n&#39;
 test_conditionspeed(threading.Thread, threading.Condition())
 print &#39;\n\t######## testing multiprocessing.Condition\n&#39;
 test_conditionspeed(multiprocessing.Process, multiprocessing.Condition())
 print &#39;\n\t######## testing condition managed by a server process\n&#39;
 test_conditionspeed(multiprocessing.Process, manager.Condition())

gc.enable()

</div>
 </div>
 </div>
 <div class="sphinxsidebar">
 <div class="sphinxsidebarwrapper">
 <h3><a href="../contents.html">Table Of Contents</a></h3>
 <ul>
<li><a class="reference internal" href="#">16.6. <tt class="docutils literal">multiprocessing</tt> &#8212; Process-based &#8220;threading&#8221; interface</a><ul>
<li><a class="reference internal" href="#introduction">16.6.1. Introduction</a><ul>
<li><a class="reference internal" href="#the-process-class">16.6.1.1. The <tt class="docutils literal">Process</tt> class</a></li>
<li><a class="reference internal" href="#exchanging-objects-between-processes">16.6.1.2. Exchanging objects between processes</a></li>
<li><a class="reference internal" href="#synchronization-between-processes">16.6.1.3. Synchronization between processes</a></li>
<li><a class="reference internal" href="#sharing-state-between-processes">16.6.1.4. Sharing state between processes</a></li>
<li><a class="reference internal" href="#using-a-pool-of-workers">16.6.1.5. Using a pool of workers</a></li>
</ul>
</li>
<li><a class="reference internal" href="#reference">16.6.2. Reference</a><ul>
<li><a class="reference internal" href="#process-and-exceptions">16.6.2.1. <tt class="docutils literal">Process</tt> and exceptions</a></li>
<li><a class="reference internal" href="#pipes-and-queues">16.6.2.2. Pipes and Queues</a></li>
<li><a class="reference internal" href="#miscellaneous">16.6.2.3. Miscellaneous</a></li>
<li><a class="reference internal" href="#connection-objects">16.6.2.4. Connection Objects</a></li>
<li><a class="reference internal" href="#synchronization-primitives">16.6.2.5. Synchronization primitives</a></li>
<li><a class="reference internal" href="#shared-ctypes-objects">16.6.2.6. Shared <tt class="docutils literal">ctypes</tt> Objects</a><ul>
<li><a class="reference internal" href="#module-multiprocessing.sharedctypes">16.6.2.6.1. The <tt class="docutils literal">multiprocessing.sharedctypes</tt> module</a></li>
</ul>
</li>
<li><a class="reference internal" href="#managers">16.6.2.7. Managers</a><ul>
<li><a class="reference internal" href="#namespace-objects">16.6.2.7.1. Namespace objects</a></li>
<li><a class="reference internal" href="#customized-managers">16.6.2.7.2. Customized managers</a></li>
<li><a class="reference internal" href="#using-a-remote-manager">16.6.2.7.3. Using a remote manager</a></li>
</ul>
</li>
<li><a class="reference internal" href="#proxy-objects">16.6.2.8. Proxy Objects</a><ul>
<li><a class="reference internal" href="#cleanup">16.6.2.8.1. Cleanup</a></li>
</ul>
</li>
<li><a class="reference internal" href="#module-multiprocessing.pool">16.6.2.9. Process Pools</a></li>
<li><a class="reference internal" href="#module-multiprocessing.connection">16.6.2.10. Listeners and Clients</a><ul>
<li><a class="reference internal" href="#address-formats">16.6.2.10.1. Address Formats</a></li>
</ul>
</li>
<li><a class="reference internal" href="#authentication-keys">16.6.2.11. Authentication keys</a></li>
<li><a class="reference internal" href="#logging">16.6.2.12. Logging</a></li>
<li><a class="reference internal" href="#module-multiprocessing.dummy">16.6.2.13. The <tt class="docutils literal">multiprocessing.dummy</tt> module</a></li>
</ul>
</li>
<li><a class="reference internal" href="#programming-guidelines">16.6.3. Programming guidelines</a><ul>
<li><a class="reference internal" href="#all-platforms">16.6.3.1. All platforms</a></li>
<li><a class="reference internal" href="#windows">16.6.3.2. Windows</a></li>
</ul>
</li>
<li><a class="reference internal" href="#examples">16.6.4. Examples</a></li>
</ul>
</li>
</ul>

<h4>Previous topic</h4>
 <a href="dummy_thread.html"
 title="previous chapter">16.5. <tt class="docutils literal">dummy_thread</tt> &#8212; Drop-in replacement for the <tt class="docutils literal">thread</tt> module</a>
 <h4>Next topic</h4>
 <a href="mmap.html"
 title="next chapter">16.7. <tt class="docutils literal">mmap</tt> &#8212; Memory-mapped file support</a>
<h3>This Page</h3>
<ul class="this-page-menu">
 <li><a href="../bugs.html">Report a Bug</a></li>
 <li><a href="../_sources/library/multiprocessing.txt"
 rel="nofollow">Show Source</a></li>
</ul>

<div id="searchbox" style="display: none">
 <h3>Quick search</h3>
 <form class="search" action="../search.html" method="get">
 <input type="text" name="q" />
 <input type="submit" value="Go" />
 <input type="hidden" name="check_keywords" value="yes" />
 <input type="hidden" name="area" value="default" />
 </form>
 
 Enter search terms or a module, class or function name.
 
</div>
<script type="text/javascript">$('#searchbox').show(0);</script>
 </div>
 </div>
 <div class="clearer"></div>
 </div>
 <div class="related">
 <h3>Navigation</h3>
 <ul>
 <li class="right" style="margin-right: 10px">
 <a href="../genindex.html" title="General Index"
 >index</a></li>
 <li class="right" >
 <a href="../py-modindex.html" title="Python Module Index"
 >modules</a> |</li>
 <li class="right" >
 <a href="mmap.html" title="16.7. mmap — Memory-mapped file support"
 >next</a> |</li>
 <li class="right" >
 <a href="dummy_thread.html" title="16.5. dummy_thread — Drop-in replacement for the thread module"
 >previous</a> |</li>
 <li><img src="../_static/py.png" alt=""
 style="vertical-align: middle; margin-top: -1px"/></li>
 <li><a href="http://www.python.org/">Python</a> &raquo;</li>
 <li>
 <a href="../index.html">Python 2.7.5 documentation</a> &raquo;
 </li>

<li><a href="index.html" >The Python Standard Library</a> &raquo;</li>
 <li><a href="someos.html" >16. Optional Operating System Services</a> &raquo;</li> 
 </ul>
 </div>
 <div class="footer">
 &copy; <a href="../copyright.html">Copyright</a> 1990-2020, Python Software Foundation.
 
 The Python Software Foundation is a non-profit corporation.
 <a href="http://www.python.org/psf/donations/">Please donate.</a>
 
 Last updated on Oct 13, 2020.
 <a href="../bugs.html">Found a bug</a>?
 
 Created using <a href="http://sphinx.pocoo.org/">Sphinx</a> 1.1.3.
 </div>

</body>
</html>