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
Similar to #39 I think we want to help user code deal with unexpected errors being raised or process exits. This can often come from things like a statement timeout from Ecto, or trying to call a process that has crashed etc. These situations are often things that will self-correct in a well functioning BEAM application. So sending a NACK (or maybe letting the message timeout?) makes sense, rather than crashing the whole consumer.
The text was updated successfully, but these errors were encountered:
Hmm, I would definitely not send NACK and just let the message timeout here. This is the existing behaviour.
Mind that PullConsumer is modelled after GenServer and there, raising within callback will crash the process. This would work well with process state, so you could recover from invalid state cases by restarting the consumer with new, clean one.
So I would personally not do anything here. What do you think?
Similar to #39 I think we want to help user code deal with unexpected errors being raised or process exits. This can often come from things like a statement timeout from Ecto, or trying to call a process that has crashed etc. These situations are often things that will self-correct in a well functioning BEAM application. So sending a NACK (or maybe letting the message timeout?) makes sense, rather than crashing the whole consumer.
The text was updated successfully, but these errors were encountered: