Closed
Description
Hi,
We’ve recently merged a PR #152 where the finish operation instantly resumes all the pending producers and consumers (with a nil value), thus ending all the iterators.
I was wondering if we should have the same behaviour for failures. As it is a terminal event, we have the same kind of risk for infinite suspensions. I guess when sending a failure, all the pending operations should resume instantly. All the awaiting consumers should receive the failure and the future iterations should also resume instantly.
@phausler what do you think ?
The fail(_:)
function would also be not async. It makes sense since, as a terminal event, no back pressure should be applied.
Metadata
Metadata
Assignees
Labels
No labels