Extending recording/publishing infra

[rkeely]

Seems I messed up the previous PR a bit. Had to create a new one and carry the context here.

This PR is to fulfill this #2809 (comment) and address comments from https://github.com/redis/lettuce/pull/2819

Extending individual events requires duplicating a lot of the functionality within the individual events and it makes these mutable.

I was looking more for an approach where we extend the recording/publishing infrastructure. In particular:

• Add Event getSource() to RecordableEvent and make RecordableEvent extends Event
• Capture the original event in JfrRecordableEvent and rework NoOpEventRecorder to create an event instance
• Long-lived events would be published to EventBus (e.g. EventRecorder.RecordableEvent event = …; later: eventBus.publish(event).
• Adopt JfrEventRecorder.publish(…) to check whether the given event is EventRecorder.RecordableEvent. If so, then we call record(…) on the event instead of jdk.jfr.Event jfrEvent = createEvent(…);
• Retrofit DefaultEventBus to unwrap EventRecorder.RecordableEvent before emitting the event

---

This is more about getting the direction correct. Will supply more testing if the infra change looks good.

• Add publish interface in EventRecorder
• Allow DefaultEventBus to publish event through EventRecorder
• Wrap source Event in RecordableEvent

Make sure that:

• You have read the contribution guidelines.
• You have created a feature request first to discuss your contribution intent. Please reference the feature request ticket number in the pull request.
• You use the code formatters provided here and have them applied to your changes. Don’t submit any formatting related changes.
• You submit test cases (unit or integration tests) that back your changes.

[mp911de]

What is the semantic difference between record(Event) and publish(Event)?

Such an arrangement asks for trouble because the methods are ambiguous without clear distinction.

RecordableEvent is part of the API, moving it around introduces a breaking change without good reason. All necessary changes are possible without breaking the public API.

[rkeely]

I can move RecordableEvent back to EventRecorder.

I added publish per #2819 (comment), specifically:

• Adopt JfrEventRecorder.publish(…) to check whether the given event is EventRecorder.RecordableEvent. If so, then we call record(…) on the event instead of jdk.jfr.Event jfrEvent = createEvent(…);

Agreed that they create ambiguity. Should I combine the implementation of JfrEventRecorder.#record and JfrEventRecorder#publish into #record?

[rkeely]

@mp911de posted a new commits to address your last comment. Would you please take a look?
Still its only changing the infra, and if it looks good, I am thinking to make another change that actually publish/record those RecordableEvent through EventBus.