Github user elbamos commented on the pull request:
https://github.com/apache/incubator-zeppelin/pull/463#issuecomment-162747222
@jongyoul
Thank you for considering my concern.
I do *not* agree that we should support using Zeppelin with Spark without a
manually-installed Spark.
If Spark is not installed, the Zeppelin installer should not try to install
it.
Installing Spark should not be the responsibility of the Zeppelin
installer.
(The exception would be if Zeppelin has Spark as a dependency. But if
Spark is a dependency, the install should look for Spark, and fail to install
if it is missing. But I don't think we intend for Spark to be a dependency.)
(If Spark *is* a Zeppelin dependency, then why do we have a concept that
there can be more than one InterpreterGroup? If Spark is a dependency, then
every interpreter must be able to move data to and from Spark. That is not our
functionality. Therefore, I do not think Spark should be a dependency.)
The correct behavior as Zeppelin is currently built, is that if SPARK_HOME
is not set, %spark.pyspark and %spark should start without a SparkContext or
fail to start entirely.
The concept is "fail early with a helpful error message."
I'm not sure I understand your idea? I have written code to make %spark
fail if SPARK_HOME is not set (or if spark.home IS set).
But again, I have been told that support for manual-Spark is already going
to be removed in a pending PR.
Can you clarify your proposal?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
