[SPARK-32478] Error message to show the schema mismatch in gapply with Arrow vectorization

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.0.0
• Fix Version/s: 3.0.1, 3.1.0
• Component/s: SparkR
• Assignee: Hyukjin Kwon
• Reporter: Hyukjin Kwon
• Created: 29/Jul/20 07:39
• Updated: 12/Dec/22 18:11
• Resolved: 30/Jul/20 06:16

Description

Currently, the error message is confusing when the output schema type is not matched with the actual R DataFrame in gapply:

./bin/sparkR --conf spark.sql.execution.arrow.sparkr.enabled=true

df <- createDataFrame(list(list(a=1L, b="2")))
count(gapply(df, "a", function(key, group) { group }, structType("a int, b int")))

  org.apache.spark.SparkException: Job aborted due to stage failure: Task 43 in stage 2.0 failed 1 times, most recent failure: Lost task 43.0 in stage 2.0 (TID 2, 192.168.35.193, executor driver): java.lang.UnsupportedOperationException
	at org.apache.spark.sql.vectorized.ArrowColumnVector$ArrowVectorAccessor.getInt(ArrowColumnVector.java:212)
	...

We should probably also document that the type should be matched always.

Issue Links

• links to: [Github] Pull Request #29283 (HyukjinKwon)