[SPARK-22868] 64KB JVM bytecode limit problem with aggregation

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Incomplete
• Affects Version/s: 2.2.1, 2.3.0
• Component/s: SQL
• Labels: bulk-closed
• Assignee: Unassigned
• Reporter: Kazuaki Ishizaki
• Created: 22/Dec/17 04:11
• Updated: 08/Oct/19 05:44
• Resolved: 08/Oct/19 05:44

Description

The following programs can throw an exception due to the 64KB JVM bytecode limit

    val df = spark.sparkContext.parallelize(
      Seq((1.1, 2.2, 3.3, 4.4, 5.5, 6.6, 7.7, 8.8, 9.9, 10.0, 11.1, 12.2, 13.3, 14.4, 15.5, 16.6, 17.7, 18.8, 19.9, 20.0, 21.1, 22.2)),
      1).toDF()

    df.agg(
      kurtosis('_1), kurtosis('_2), kurtosis('_3), kurtosis('_4), kurtosis('_5),
      kurtosis('_6), kurtosis('_7), kurtosis('_8), kurtosis('_9), kurtosis('_10),
      kurtosis('_11), kurtosis('_12), kurtosis('_13), kurtosis('_14), kurtosis('_15)
    ).collect

    df.groupBy('_22)
      .agg(
        kurtosis('_1), kurtosis('_2), kurtosis('_3), kurtosis('_4), kurtosis('_5),
        kurtosis('_6), kurtosis('_7), kurtosis('_8), kurtosis('_9), kurtosis('_10),
        kurtosis('_11), kurtosis('_12), kurtosis('_13), kurtosis('_14), kurtosis('_15)
    ).collect

    df.groupBy(
        round('_1, 0), round('_2, 0), round('_3, 0), round('_4, 0), round('_5, 0),
        round('_6, 0), round('_7, 0), round('_8, 0), round('_9, 0), round('_10, 0))
      .agg(
        kurtosis('_1), kurtosis('_2), kurtosis('_3), kurtosis('_4), kurtosis('_5),
        kurtosis('_6), kurtosis('_7)
    ).collect

*/