asyncframes - Parallel programming for software engineers

Version asyncframes v1 asyncframes v2
Docs   Documentation Status
Download pypi_v1 pypi_v2
Source github_v1 github_v2
Status travis_v1 coverage_v1 travis_v2 coverage_v2
License license_v1 license_v2

asyncframes is a coroutine library for Python and a reference implementation of the Frame Hierarchy Programming Model (FHM). The goal of FHM is to help programmers design clean and scalable parallel programs. The main features of asyncframes are:

  • Hierarchical code design
  • Inherent and scalable parallelism
  • Architecture independence
  • Extensibility through frame classes (a class whose lifetime is bound to the execution of a frame)

Introduction

In the Frame Hierarchy Programming Model (FHM) a program is represented as a dynamic tree of frames. A frame is a suspendable function (a coroutine) with an object oriented context (the frame class) that only exists until the function returns. Frames can be used to represent both temporal processes (using the coroutine) and physical or conceptual objects (using the frame class).

Each FHM program has exactly one root frame. The root frame can recursively spawn child frames. Each child frame runs in parallel unless it’s awaiting another frame or an awaitable event. Frames of type Frame run on a single thread. They use cooperative multitasking to simulate parallelism. Frames of type PFrame run on any of the threads available in the event loop’s thread pool. Frame and PFrame are frame classes. They can be sub-classed to create specialized frame classes with encapsulated data.

Indices and tables