#async-task #task #tokio

async_tasks_state_map

A struct for recording execution status of async tasks with async methods

5 releases (1 stable)

1.0.1 Feb 27, 2024
0.2.0 Feb 24, 2024
0.1.2 Feb 24, 2024
0.1.1 Feb 24, 2024
0.1.0 Feb 24, 2024

#1245 in Asynchronous

MIT/Apache

16KB
214 lines

Introduction

A struct for recording execution status of async tasks with async methods.

Functions:

  • Able to host Futures and query whether they are not found, running, successful, failed, or revoking.
  • Able to host Futures to revoke the succeeded Futures and make them not found.

Dependency:

  • Depend on tokio with feature rt, so cannot use other async runtimes.
  • Depend on scc for async HashMap.

Use this crate if:

  • Easy to generate an unique task_id (not necessarily String) for a future (task).
  • Don't want tasks with the same task_id to succeed more than once.
  • Require linearizability.
  • Want to revoke a task, and don't want the revoking to succeed more than once.

A recorder can only use single task_id type. The type of task_id should be:

  • Eq + Hash + Clone + Send + Sync + 'static
  • Cheap to clone (sometimes can use Arc) (only cloned once when launch).

async_tasks_recorder is another implement depending on HashSet, which is easier to iterate every task in the same state. But you should not use that crate if you only focus on iterating only one state. Instead, you can collect the tasks in certain state into an external Arc<HashSet>.

State Transition Diagram

    ┌------- Revoking ←-----┐
    ↓                       |
NotFound --> Working --> Success
               ↑ |
               | ↓
             Failed
  • Can only launch when NotFound or Failed.
  • Can only revoke when Success.

Advices

Simplified State Transition Diagram

  • In some cases, NotFound can be equivalent to Failed.
  • In most cases, Revoking can be equivalent to Success.

So you may get:

    ┌----------------------┐
    ↓                      |
Failed <--> Working --> Success

Why or why not use Arc to store task_id

If you don't use Arc, all task_id is stored in scc::HashMap. But it has to be cloned when query or update the task's state.

However, if you use Arc, only Arc is stored in scc::HashMap, and the task_id is stored in heap independently, which may cause additional memory overhead.

Make your own decision on whether to use Arc or not.

Dependencies

~3–8.5MB
~66K SLoC