Interrupt issue with PublishOn about reactor-core HOT 6 CLOSED

Two options:

• Release the scheduler after the terminal event has been emitted. Drawback: potentially longer retention.
• Make sure the Future.cancel() is called with false if the current thread happens to be the thread running the task. Drawback: task instance size increase, more instructions before the task body is run.

from reactor-core.

I actually wonder which interrupt @sskrla is referring to since it should be the current dispatcher strategy for SchedulerGroup at least. I guess since it was referring to Fluxion which offered dispachOn(ExecutorService) as in RSC it is this one.

EDIT: yes it was the ExecutorService scheduler implementation.

from reactor-core.

As a second issue, adding ExecutorService direct support to Flux/Mono should be considered, opened #51.

from reactor-core.

@akarnokd do you know if we face that issue still in RSC ExecutorService support ?

from reactor-core.

Due to a recent update, Rsc's observeOn shuts down the worker after onComplete has been delivered. At that point, all that's left is the end of the wrapper action on the scheduler and no user code can be executed that would throw on an interrupted thread. Otherwise, it still interrupts itself at that point which adds some overhead. I can work around that if necessary.

from reactor-core.

I'll sync it as it for now thanks @akarnokd .

from reactor-core.