Continuation-Monad

A Python library implementing stack-safe continuations based on schedulers, ensuring deadlock-free asynchronous computations.

Features

• Trampoline-based: Stack-safe execution of the continuation monad through trampolining
• Continuation certificate: Ensures the execution of a Flowable completes, albeit with some computational overhead.
• Scheduler-based: Explicit control over execution context through different scheduler implementations
• Composable operations: Monadic operators for clean functional pipelines

Installation

You can install Continuation-Monad using pip:

pip install continuationmonad

Example

import continuationmonad


def count_down(count: int):
    print(f'{count=}')

    if count == 0:
        return continuationmonad.from_(count)
    
    else:
        # schedule recursive call on the trampoline
        return continuationmonad.tail_rec(lambda: count_down(count - 1))

# runs continuation and returns 0
result = continuationmonad.run(
    count_down(5)
)

Schedulers and Trampolines

The library provides the following scheduler implementations

• init_main_scheduler: Initializes a scheduler that runs in the main thread. It starts execution via the run() method, which blocks until stop() is called to terminate the scheduler.
• init_event_loop_scheduler: Creates a scheduler that operates on a dedicated background thread. This is useful for offloading tasks from the main thread or isolating execution contexts.

  scheduler = continuationmonad.init_event_loop_scheduler()
  

• init_trampoline: Returns a lightweight scheduler that only implements the schedule() method. It does not support delayed execution methods like schedule_relative() or schedule_absolute(). This scheduler runs tasks immediately in a loop until its queue is exhausted.

Operations

Creating Continuation Monads

• defer - creates a continuation monad that defers the subscription until a source is connected (see defer example)
• from_: Create a continuation monad from a value:

  c = continuationmonad.from_(5)
  

• get_trampoline - retrieve trampoline associated with the continuation monad

  c = continuationmonad.get_trampoline()
  

• schedule_on - schedule elements emitted by the source on a dedicated scheduler

  c = continuationmonad.schedule_on(scheduler)
  

• schedule_trampoline - schedule item on trampoline

  c = continuationmonad.schedule_trampoline()
  

• tail_rec - recursively call a function on trampoline

Transforming operators

• connect - connects the continuation monad to deferred one or multiple subscribers (see defer example)
• flat_map - apply a function to the item emitted by the source and flattens the result
• map - map the item emitted by the source by applying the given function

Combining operators

• zip - create a new continuation monad from two (or more) continuation monads by combining their items in a tuple

  c = continuationmonad.zip((
      continuationmonad.from_(1),
      continuationmonad.from_(2),
  ))
  

Other operators

• fork - runs the continuation monad on a specified trampoline (see defer example)

  c = continuationmonad.fork(
      source=source,
      scheduler=scheduler,
      weight=1,               # specify continuation weight
  )
  

• run - runs the continuation monad on a new trampoline and returns its result

  # execute a continuation monad
  result = continuationmonad.run(source)