Github user HeartSaVioR commented on the issue:
https://github.com/apache/storm/pull/2449
We (as a one of maintainer of project) may be annoying, but someone would
not want to manage their own dependencies and want to have Hadoop dependency
build-in distribution. Spark and Flink provides the option to select the Hadoop
version. That may be another option, if majority is using Storm with HDFS.---
