codep 0.0.1

A light-weight framework for defining and running dependent tasks.

codep

A light-weight framework for defining and running dependent tasks. By building a dependency graph between tasks, the runner makes sure that tasks that are required by multiple dependants are only run once and reuses the results.

Usage

In the example below the calculate_circumference and calculate_volume tasks both depend on the acquire_radius and calculate_pi tasks. The runner will infer that from the dependency graph and run them before.

import codep
import immutables


@codep.make_partial()
def acquire_radius(state: immutables.Map) -> immutables.Map:
    return state.set("radius", 6_371)


@codep.make_partial()
def calculate_pi(state: immutables.Map) -> immutables.Map:
    return state.set("pi", 3.14)


@codep.make_partial(depends=(calculate_pi, acquire_radius))
def calculate_circumference(state: immutables.Map) -> immutables.Map:
    return state.set("circumference", 2 * state["pi"] * state["radius"])


@codep.make_partial(depends=(acquire_radius, calculate_pi))
def calculate_volume(state: immutables.Map) -> immutables.Map:
    return state.set("volume", (4/3) * state["pi"] * state["radius"] ** 3)


if __name__ == '__main__':
    r1, r2 = codep.run(calculate_circumference, calculate_volume)
    circumference = r1.state["circumference"]
    volume = r2.state["volume"]
    print(
        f"The circumference of earth is {circumference} km and its volume is " 
        f"{volume} km^3"
    )

Todo

• Make Partial a Generic, and make tasks only able to return a value. Partial.apply should instead assign the result to the state which becomes a mapping Type[Partial] -> Any.
• Remove print(), use logging
• Concurrent task execution