Switching scheduler to Main Thread

[alek5k]

Hi there, just wondering how I can schedule work on specific threads, specifically the program's main thread? For example, I want to do certain map operations (here I use do_action) on separate threads, then use on_next the main thread. My reference to CurrentThreadScheduler seems to change once the ThreadPoolSchedulers are used - is this intended behaviour?

Here is some example code:

    import reactivex as rx
    from reactivex import operators as ops
    from reactivex.scheduler import ThreadPoolScheduler, CurrentThreadScheduler
    from threading import current_thread

    print(f"starting thread: {current_thread().name}")

    current_thread_scheduler = CurrentThreadScheduler()
    threadpool1 = ThreadPoolScheduler(1)
    threadpool2 = ThreadPoolScheduler(1)

    rx.just(1).pipe(
        ops.observe_on(threadpool1),
        ops.do_action(lambda _: print(f"step 1: {current_thread().name}")),
        ops.observe_on(threadpool2),
        ops.do_action(lambda _: print(f"step 2: {current_thread().name}")),
        ops.observe_on(current_thread_scheduler),
        ops.do_action(lambda _: print(f"step 3: {current_thread().name}")),
    ).subscribe(on_next=lambda _: print(f"on_next: {current_thread().name}"))

Output:

starting thread: MainThread
step 1: ThreadPoolExecutor-0_0
step 2: ThreadPoolExecutor-1_0
step 3: ThreadPoolExecutor-1_0
on_next: ThreadPoolExecutor-1_0

Desired Output:

starting thread: MainThread
step 1: ThreadPoolExecutor-0_0
step 2: ThreadPoolExecutor-1_0
step 3: MainThread
on_next: MainThread

• OS: Windows10
• RxPY: 4.0.2
• Python version 3.9

[Petrus Nguyễn Thái Học]

I think in this case, current_thread_scheduler == threadpool2 is expected behavior,
current_thread means the thread on_next is called in it

[Romain Picard]

This is not possible without an event loop. See the issue #245 for more details.