[ZEPPELIN-630] Introduce new way of dependency loading to intepreter

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.6.0
• Assignee: Mina Lee
• Reporter: Mina Lee
• Created: 25/Jan/16 17:50
• Updated: 25/Feb/16 17:07
• Resolved: 01/Feb/16 02:08

Description

Currently dependency loading feature(%dep) is dedicated to spark interpreter dependencies are loaded during Zeppelin runtime.
When user wants to include external libraries to other interpreter than spark, user needs to download the library, add this to ZEPPELIN_CLASSPATH by editting `bin/zeppelin-daemon.sh` or `bin/interpreter.sh`.

I would like to suggest new way of dependency loading by enabling:
1. Download library via GUI 'interpreter' menu
2. Add downloaded library to interpreter's classpath at process initiation time.

Every interpreter can leverage this feature to load external libraries. For exampe, JDBC interpreter can get benefit from it by adding proper jdbc-connector in 'interpreter' menu without editing shell script.

Issue Links

• relates to: ZEPPELIN-259 Generalize the Dependency implementation
• supercedes: ZEPPELIN-194 Exception on %sql over CSV dataframe
• supercedes: ZEPPELIN-381 sqlContext doesn't work with tables created by external jars (e.g. com.databricks.spark.csv); spark <1.5.1
• supercedes: ZEPPELIN-609 Add dependencies with property zeppelin.dep.additionalDependencies