Closed skrater closed 8 years ago
@gfronza
@skrater Well, setting autoAck
to false won't behave as you said. You are just telling (or not) to RabbitMQ to handle the ack operation instead of the client code. IMHO, we must have a much stronger reason to open this possibility to developers, because when something goes wrong, we'll have two possibilities to cover up.
As for the logger thing, log.Success is just a little bit unusual (outside of the well known [info, debug, warn, error] world). So I would change to log.Debug, but that's just my opinion. @alexandrevicenzi what do you think?
@gfronza AFAIK, autoAck just telling to ack the message when he is delivered to the server (porthos), not when he is processed. This will flush the queue much faster and improve performance, but it need to be a developer choice, he needs to know that some messages will be lost when a panic goes up. Thats not a cruticial feature, i just did not want to leave ackMessages hardcoded.
About the log, I think it would look better with Debug, i will change it.
Ps.: This discussion is being better for my English than for Porthos. kkkk
I still believe that we should only have 1 option. autoAck seems to be more appropriate for rpc calls (where we have really aggressive timeouts thresholds) and we avoid then ack call.
I'll merge it, but we might consider having a default value for the autoAck option (true
).
@skrater just some thoughts (something to discuss):