BREAKING: Replace implicit async function tasks with task decorator

[mosmuell]

This merge request introduces a breaking change to how tasks are defined and managed in the pydase library. It improves task management by making task definitions more explicit and by that adding code-completion capabilities. With this update, users must add the @task decorator to define an asynchronous method as a task. The previous approach where pydase automatically generated start_<method> and stop_<method> methods for task control is now deprecated.

To align with the new @task decorator approach, users need to make changes to their existing codebase. Please follow the migration guide below to update your services.

Key Changes:

1. Introduction of the @task decorator:

   • To define a method as a task within a class inheriting from pydase.DataService, you must now use the @task decorator.
   • This decorator should be applied to an asynchronous function that requires no arguments.

2. Method is converted into an object with start() and stop() methods:

   • When a method is decorated with @task, it is converted into a Task object. This object provides start() and stop() methods to control task execution.

3. Removal of start_<method> and stop_<method> method generation:

   • The old method of controlling tasks via dynamically generated start_<method> and stop_<method> methods is no longer supported.
   • Users must update their code to use the new Task object methods (start() and stop()).

4. Redefinition of Autostart Tasks:

   • The way autostart tasks are defined has changed. Previously, users added tasks to the _autostart_tasks dictionary in the __init__ method of their pydase.DataService class.
   • Now, autostart behavior is defined directly in the @task decorator with the autostart=True argument.

Example of Updated Code:

Before:

import asyncio

import pydase

class MyService(pydase.DataService):
    def __init__(self) -> None:
        super().__init__()
        self._autostart_tasks["my_task"] = ()  # Define autostart task

    async def my_task(self):
        while True:
            # Task code
            await asyncio.sleep(1)

service = MyService()
service.start_my_task()  # Old way to start the task

After:

import asyncio

import pydase
from pydase.task.decorator import task

class MyService(pydase.DataService):
    @task(autostart=True)  # Define autostart task directly in the decorator
    async def my_task(self):
        while True:
            # Task code
            await asyncio.sleep(1)

service = MyService()
service.my_task.start()  # New way to start the task

[mosmuell]

Solves #129.