Graceful shutdown and message processing

[ypicard]

Summary

Description

When a SIGINT signal is sent to the process running the consumer while it is processing a message, the consumer behaves as follows:

1. Waits for AWS requests to stop (primarily long polling).
2. Emits the stopped event.

This leads to the following situation:

1. There is no way for consumers of the library to ensure that an in-flight message finishes processing before the stopped event is emitted.
2. It appears that both the heartbeat and visibility timeout extensions stop as well. Even when adding custom logic to emit a separate stopped event after message processing completes, the acknowledgment and deletion steps fail with an error like:

SQSError: Error changing visibility timeout: Value ... for parameter ReceiptHandle is invalid. 
Reason: Message does not exist or is not available for visibility timeout change

Question

Before assuming this is a bug, I wanted to ask whether this behavior is by design.
Should consumers expect this shutdown behavior when handling in-flight messages?

Thank you for the hard work and for providing such a solid and reliable wrapper! 🙏

Example

No response