Hello,
I just had a commit to Spark accepted that fixes the management of
Parquet logging:
https://github.com/apache/spark/pull/3271
The Spark project committer (Michael Armbrust) suggested I mention the
issue to you guys. If you notice the code in Spark you need to bend over
backward to control/integrate Parquet logging into any other project
that uses Parquet. This is primarily for 2 reasons:
1) Direct use of java.util.logging
2) The static initializer in parquet.Log
I would like to suggest that slf4j be used. This will decouple the
choice of specific logger and allow the users of Parquet to specify it.
It will also make the use of the static initializer unnecessary.
Just a suggestion.
Jim
