New Case Study:See how Anthropic automated 95% of dependency reviews with Socket.Learn More
Socket
Sign inDemoInstall
Socket

offspring

Package Overview
Dependencies
Maintainers
1
Alerts
File Explorer

Advanced tools

Socket logo

Install Socket

Detect and block malicious and high-risk dependencies

Install

offspring

Objects and patterns for working with processes in Python using the multiprocessing library

  • 0.1.1
  • PyPI
  • Socket score

Maintainers
1

Offspring

.. image:: https://img.shields.io/travis/borgstrom/offspring.svg :target: https://travis-ci.org/borgstrom/offspring

.. image:: https://img.shields.io/codecov/c/github/borgstrom/offspring.svg :target: https://codecov.io/github/borgstrom/offspring

.. image:: https://img.shields.io/pypi/v/offspring.svg :target: https://pypi.python.org/pypi/offspring :alt: Latest PyPI version

This is a collection of objects and patterns for working with processes in Python using the multiprocessing library.

The main idea is that you express your unit of work as a simple method on an object and then when that object is instantiated the work will be run in a subprocess.

Use cases

Offspring was built to address the following use cases for running code in a subprocess.

Run something once


.. code-block:: python

    from offspring import Subprocess


    class MyTask(Subprocess):
        def init(self, arg1):
            # This is run in the parent process and is used to prepare your object.
            # It receives whatever arguments were supplied to the constructor.
            self.arg1 = arg1

        def run(self):
            # This will be run in the child process and completes your work.
            # ...


    MyTask('this is arg1').wait()


Run in a loop
~~~~~~~~~~~~~

.. code-block:: python

    from offspring import SubprocessLoop


    class MyTask(SubprocessLoop):
        def init(self, arg1):
            # This is the same as init for Subprocess.
            self.arg1 = arg1

        def begin(self):
            # Called before the start of the loop in your child process.
            # ...

        def loop(self):
            # Called each loop iteration in your your child process.
            # It can return a sleep duration until the next loop, or False to stop the loop.
            # ...

        def end(self):
            # Called after the end of the loop, before termination in your child process.
            # ...


    MyTask('this is arg1').wait()


Implementation details
----------------------

``.init(*args, **kwargs)``

Called when an instance of your class is created. It receives the same arguments as the __init__ method, so you are encouraged to explicitly define the arguments you expect.

.start()


Creates the subprocess.  This is automatically called unless you set ``EXPLICIT_START`` to ``True``.


``.wait()``
~~~~~~~~~~~

If you need to wait for your child process you can call ``.wait`` on your ``Subprocess`` object.  This is just a
shortcut to ``.join`` on the ``multiprocessing.Process`` object.


``.shutdown()``

This will send a TERM signal to the child process, unless TERMINATE_ON_SHUTDOWN is False, and then calls .wait() to join the child process. It is automatically called whenever the parent process exits via the atexit module.

.process


Each ``Subprocess`` object has a ``.process`` attribute that is the ``multiprocessing.Process`` object.


``WAIT_FOR_CHILD``

Defaults to False.

If set to True on your Subprocess class then a Pipe will be used to block the parent process until the child has started. This is useful when you want to ensure that your Subprocess object is started and .run is called even if the parent process exits quickly.

.. code-block:: python

class MyTask(Subprocess):
    WAIT_FOR_CHILD = True

    def run(self):
        print("This will always print")

MyTask()

The SubprocessLoop class does this to ensure that your object has begin & end called (loop may not be called as a TERM signal received during startup will prevent the loop from every actually completing other than begin & end).

TERMINATE_ON_SHUTDOWN


Defaults to ``True``.

If set to ``False`` then when ``.shutdown`` is called on a ``Subprocess`` object the child process **will not** be
terminated before being joined.  This means that the parent will block until the child completes the ``.run`` function.

.. code-block:: python

    import time

    class MyTask(Subprocess):
        TERMINATE_ON_SHUTDOWN = False

        def run(self):
            time.sleep(2)

    MyTask()


``EXPLICIT_START``
~~~~~~~~~~~~~~~~~~

Defaults to ``False``.

If set to ``True`` then when you instantiate an object you must explicitly call ``.start()`` before the child process
will be spawned.

.. code-block:: python

    class MyTask(Subprocess):
        EXPLICIT_START = True

        def run(self):
            print("Running!")


    task = MyTask()
    # Do some other work
    task.start()
    # Running! is now printed


FAQs


Did you know?

Socket

Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.

Install

Related posts

SocketSocket SOC 2 Logo

Product

  • Package Alerts
  • Integrations
  • Docs
  • Pricing
  • FAQ
  • Roadmap
  • Changelog

Packages

npm

Stay in touch

Get open source security insights delivered straight into your inbox.


  • Terms
  • Privacy
  • Security

Made with ⚡️ by Socket Inc