[
https://issues.apache.org/jira/browse/KAFKA-16084?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17804460#comment-17804460
]
Greg Harris commented on KAFKA-16084:
-------------------------------------
[~cegerton] I agree.
In this case, none of those conditions apply:
[https://github.com/apache/kafka/blob/b2bfd5d11014c65a191a41912a38cd88925fd9bc/connect/runtime/src/test/java/org/apache/kafka/connect/runtime/standalone/StandaloneHerderTest.java#L140]
The worker, plugins, pluginLoader, and loaderSwap are all created both with
the annotation and the invocation by accident, and I think one of each should
be removed.
The StandaloneHerder partial mock on the next line however should not be
removed, I think it makes sense.
> Simplify and deduplicate StandaloneHerderTest mocking
> -----------------------------------------------------
>
> Key: KAFKA-16084
> URL: https://issues.apache.org/jira/browse/KAFKA-16084
> Project: Kafka
> Issue Type: Test
> Components: connect
> Reporter: Greg Harris
> Priority: Minor
> Labels: newbie++
>
> The StandaloneHerderTest has some cruft that can be cleaned up. What i've
> found:
> * The `connector` field is written in nearly every test, but only read by one
> test, and looks to be nearly irrelevant.
> * `expectConfigValidation` has two ways of specifying consecutive
> validations. 1. The boolean shouldCreateConnector which is true in the first
> invocation and false in subsequent invocations. 2. by passing multiple
> configurations via varargs.
> * The class uses a mix of Mock annotations and mock(Class) invocations
> * The test doesn't stop the thread pool created inside the herder and might
> leak threads
> * Mocking for Worker#startConnector is 6 lines which are duplicated 8 times
> throughout the test
> * Some waits are 1000 ms and others are 1000 s, and could be pulled out to
> constants or a util method
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
