tartpy 0.2.1

Overview

The initial ideas for tartpy are copied from @dalnefre and @tristanls. At the beginning, the goal was to replicate tart.js in Python. Lately, it has diverged slightly, to experiment with the new Python asynchrony model introduced in version 3.4.

tartpy aims to be an actor library with the following features:

• it implements the pure actor model (rather than the Erlang model),

• it can be used in practical situations to express lock-free and concurrent algorithms,

• it abstracts the network transport to allow arbitrary protocols,

• it implements a capability based approach to isolate actors, using membranes (again, the ideas are due to @dalnefre and @tristanls). The semantics of membranes may differ between tartpy and tartjs.

Installing

Install with:

$ pip install git+https://github.com/waltermoreira/tartpy

or clone this repository and install with python3 setup.py install:

$ git clone https://github.com/waltermoreira/tartpy
$ cd tartpy
$ python3 setup.py install

Test

Run tests with:

$ python3 setup.py test

Examples

Run example with:

python3 tartpy/example.py

There is also an IPython notebook showing the basic properties of actors.

The project actor_model contains slides and IPython notebooks from a talk presented at TACC in April 17, 2014.

Erlang Challenge

Create a ring of M actors, sending N messages around the ring:

python3 tartpy/erlang_challenge.py M N

Starting 100000 actor ring
Construction time: 1.5192079544067383 seconds
Loop times:
  0.7743091583251953 seconds
  0.7793149948120117 seconds
  0.7702958583831787 seconds
  0.7602570056915283 seconds
  0.7704610824584961 seconds
  0.779731035232544 seconds
  0.7654228210449219 seconds
  0.7625432014465332 seconds
  0.7740719318389893 seconds
  0.7699680328369141 seconds
Average: 0.7706375122070312 seconds