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

Posted to dev@zeppelin.apache.org by "Luca Rosellini (JIRA)" <ji...@apache.org> on 2015/06/18 16:57:00 UTC

[jira] [Created] (ZEPPELIN-119) SparkILoop loadFiles is called before binder initialization

Luca Rosellini created ZEPPELIN-119:
---------------------------------------

             Summary: SparkILoop loadFiles is called before binder initialization
                 Key: ZEPPELIN-119
                 URL: https://issues.apache.org/jira/browse/ZEPPELIN-119
             Project: Zeppelin
          Issue Type: Bug
          Components: Interpreters
    Affects Versions: 0.5.0
            Reporter: Luca Rosellini


In SparkInterpreter, method open(), the call to SparkILoop loadFiles method is performed before putting 'sqlc', 'sc', 'z' and 'out' into the binder object.

This means that if you have any scala init script you want to pass to Spark through the 'args' interpreter option (e.g.: args="-i init.scala"), in those init script you won't have access to any Spark's context.

Having access to 'sc' or 'sqlc' is useful to customize context initialization, like for example, adding custom UDFs to sql context.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)