Michael Edgar

> Do you have a particular issue with jib being there at compile-time? No technical issue. It's mostly dealing with limiting dependencies on our side as well. Before adding an...

@maoling, would you mind doing a review of this change? It resolves a lock contention situation where a slow connection allows the client's SendThread to enter `cleanUp` in `ClientCnxnSocketNetty` just...

Thanks @maoling, @anmolnar. I would be grateful for a sanity check on the approach taken in this PR.

Hi @maoling, @anmolnar - I'm curious if you've had an opportunity to review these changes.

> do we have a way to test this fix with an unit test ? (even with Mockito) I am open to suggestions on an approach... Since this is dealing...

@eolivelli , @maoling - I've finally added some tests with mocked dependencies. Please take a look. Also, this is for the 3.5 branch - should I instead replace the PR...

@maoling - please let me know if this PR is OK as-is or if it should target `master` (or another release version) instead.

The issue is that for RESTEasy Classic the smallrye-openapi extension uses the RESTEasy extension's path that already includes the http.root-path, whereas for reactive it's using the raw configuration property. I'll...

> You think it would not be good enough to simply enable `delegateToKafkaAcls` in order to cover all the listed cases (like @scholzj suggests)? I do think it's sufficient, yes....

Thanks. Given that `UnidirectionalTopicOperator` just became beta in 0.39, I'm guessing this would need to be implemented in both versions of the topic operator. > You would need to get...