You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In #504 an ordering issue for writes was solved by having writes enqueued on the computation scheduler. Unfortunately there seems to be a glitch when the emitted messages comes from the event loop sometimes and from other threads sometimes. When another thread than the event loop is used, the BackpressureManagingHandler starts to enqueue writes, but if the next write happens on the event loop then the write is not enqueued. When this happens, there is a chance that the previous write has not been processed on the computation scheduler yet, which makes the writes out of order.
A possible solution is to continue to enqueue messages until the queue is drained. PR coming up.
The text was updated successfully, but these errors were encountered:
Continue to enqueue writes until queue is drained, to avoid that writes comes out of order when writes are triggered from both the event loop and other threads.
Solves issue ReactiveX#605
In #504 an ordering issue for writes was solved by having writes enqueued on the computation scheduler. Unfortunately there seems to be a glitch when the emitted messages comes from the event loop sometimes and from other threads sometimes. When another thread than the event loop is used, the BackpressureManagingHandler starts to enqueue writes, but if the next write happens on the event loop then the write is not enqueued. When this happens, there is a chance that the previous write has not been processed on the computation scheduler yet, which makes the writes out of order.
A possible solution is to continue to enqueue messages until the queue is drained. PR coming up.
The text was updated successfully, but these errors were encountered: