eisenmp 1.0

How it works

You write two functions and two modules. Let’s name them Manager.py and Worker.py.

1. Manager.py defines custom objects and starts a new process, target is module_loader.py (loader)

2. loader imports your (independent) Worker.py module from file system without side effects

3. loader sits in a loop and calls your Worker.py entry function until queue is empty

4. q_feeder iterator sends a STOP message. Your Worker reads STOP, return False and exit.

5. module_loader module then puts a STOP Worker message in all other input queues.

6. Next Worker reads STOP, exit …

7. Loader runs idle and awaits the internal STOP Process message.

Generators

Variables transfer to worker

The worker should be more flexible. Thus, it needs more structured information.

• Information collector is the ModuleConfiguration class instance. The class can carry any name.

• All instance attributes are stored in a dictionary __dict__. (self.foo will be kwargs.foo)

• The instance dictionary is argument (kwargs) to call the eisenmp start method.

• Kwargs is updated further with queue information and the START_SEQUENCE_NUM of the process, before process start.

Default Queues

• Input worker lists, Output result and stop lists

• Tools, Print, Info

How to run the examples?

Fast PyPi package. eisenmp will be installed also.

In depth. Clone both repos.

eisenmp and eisenmp_examples

Install in editable pip mode.

cd eisenmp
pip3 install -e .  # uninstall (linux pip3) with: pip3 uninstall eisenmp

cd eisenmp_examples
pip3 install -e .  # uninstall with: pip3 uninstall eisenmp_examples

Run the examples. Either local Ajax Web server or terminal menu.

eisenmp_url
# or
eisenmp_menu

Get the documentation on readthedocs https://eisenmp.readthedocs.io/