An example for PySpark based project.

• JDK 7 (as the local test downloads Spark 1.6.0, which dropped JDK 6 support)

• Python PEP 8 (for checking code-style in local test)

  pip install pep8

• Python Py4J == 0.9.0 (for initiating Spark in local test)

  pip install -r requirements.txt

• For test in your local, just run below:

  ./dev/run_tests
  

  This contains some codes to download Spark 1.6.0 to run tests. So, it is okay to run the script without some specific modifications.

• As currently PySpark cannot be installed via pip install due to this issue, Spark has to be installed and SPARK_HOME="your Spark path" and PYTHONPATH=$SPARK_HOME/python/:$PYTHONPATH should be set in order to use this library outside after building.