You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by StephanEwen <gi...@git.apache.org> on 2017/10/05 08:55:56 UTC

[GitHub] flink issue #4776: [FLINK-7643] [core] Rework FileSystem loading to use fact...

Github user StephanEwen commented on the issue:

    https://github.com/apache/flink/pull/4776
  
    @bowenli86 We need to fail lazily, because Flink should be able to always work without MapR FS or HDFS being in the classpath.
    
    With the change currently, you can start Flink without any Hadoop dependencies and it works fine. It only fails then if you try to use HDFS. Failing everything eager means it always fails when MapR or Hadoop classes are not in the classpath.
    
    BTW: That behavior is the same as right now - I did not change it, just 


---