Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Python Unified Multi-tasking API (PUMA)

PUMA provides a simultaneous multi-tasking framework that takes care of managing the complexities of executing and controlling multiple threads and/or processes. PUMA abstracts each independent task of an application as a separate execution stream, called a "runnable", which can be run in a separate process or thread. PUMA provides buffers for runnables to exchange data, plus mechanisms for sending commands to runnables and receiving error status and logging information from them.

The diagram below illustrates a multi-processing example application with three tasks that respectively produce, process, and consume data.

PUMA multi-tasking example using multiple processes

Why a separate library for multi-tasking?

Python does provide built-in multi-threading and multi-processing facilities. Arguably, writing programs with these facilities in Python is more straightforward than many other programming languages. But because the Global Interpreter Lock effectively limits concurrency when using multiple threads, user code that would like to harness the available computing power for a high-performance application needs to use multiple processes. However, the common tasks of controlling launched processes, capturing errors from them, and allowing them to write to a single log output require writing boilerplate code. This is error-prone and makes it easy to write programs that stop (deadlock), which end without explanation, or whose performance is extremely poor.

PUMA aims at freeing the user to concentrate on writing their application code, by providing a framework that takes care of:

  • launching and controlling concurrent tasks;
  • capturing errors from them; and
  • maintaining a single log output for all concurrent tasks.

This separation of multi-tasking concerns naturally results in a loosely-coupled architecture with high cohesion.


The main features provided by PUMA are:

  • Runnables - Runnables allow concurrent code to be written that is able to run without changes no matter what environment (threads or processes) it is executed in.
  • Buffers - Buffers allow data to be passed between Runnables
  • ScopedAttributes - ScopedAttributes ensure that attributes behave consistently no matter the environment (threads or processes) or platform (Windows, Linux or Mac)
  • Logging - Logging provides a simple to use implementation of the challenging task of successfully logging across threads and processes.


Install directly from GitHub using pip install git+


Clone this repository and run pip install . from within your local copy.

Getting Started

For example source code demonstrating a data producer-processor-consumer pipeline as illustrated above, please see this simple demo application.

Each PUMA package includes a README file that provides a more in-depth discussion of the useful features as well as implementation details.


This project is grateful for the support from the Wellcome Trust, the Engineering and Physical Sciences Research Council (EPSRC), and the School of Biomedical Engineering and Imaging Sciences, King's College London at King's College London (KCL).


Python Unified Multi-tasking API (PUMA)







No packages published

Contributors 4