Github user NicoK commented on a diff in the pull request:
https://github.com/apache/flink/pull/4939#discussion_r148756933
--- Diff: flink-filesystems/flink-s3-fs-hadoop/pom.xml ---
@@ -182,6 +182,21 @@ under the License.
<version>${project.version}</version>
<scope>test</scope>
</dependency>
+ <!-- for HadoopS3FileSystemITCase.testRecursiveUploadForYarn -->
--- End diff --
Yes this is about the recursive upload which needs to be tested once with
hdfs and once more with s3.
Sure we could flip the dependency and let the tests in the `yarn`
sub-project depend on `flink-s3-fs-hadoop` (and I don't mind which depends on
which, actually) but wouldn't this be just the same but in reverse?
---
