@rajinisivaram @lindong28 I agree it would be unwise to include this in the 2.1.0 release. Thank you both for your willingness to entertain the possibility over the past days. The point that convinces me is this one: > we have to ensure that there are no paths where existing authentication has > changed in any way. Since we dont have tests that cover every possible > scenario, I am not sure how we can verify that.
I am actively working on the AuthenticationException issue, and while it is difficult to run experiments where the turnaround time is on the order of hours, I am hopeful that I can figure out what is going on there. Let's shoot for merging this to trunk when the above two issues are resolved. I don't expect to be able to write exhaustive tests to guarantee that authentication is not impacted by re-authentication, but I will try to re-organize the code so that the diff makes it much easier to confirm by inspection that there is no impact. I did not keep this in mind as I worked, so I think I can accomplish the same things with more clearly fenced-in changes. I think that -- at a minimum -- will be helpful. Thanks again for your willingness to give the PR a chance for 2.1.0, and Rajini thank you especially for the technical advice on the implementation and the reviews. Hopefully I can get this into a trunk-mergeable state soon. [ Full content available at: https://github.com/apache/kafka/pull/5582 ] This message was relayed via gitbox.apache.org for [email protected]
