Enrico Minack created SPARK-40601:
-------------------------------------

             Summary: Improve error when cogrouping groups with mismatching key 
sizes
                 Key: SPARK-40601
                 URL: https://issues.apache.org/jira/browse/SPARK-40601
             Project: Spark
          Issue Type: Improvement
          Components: PySpark, SQL
    Affects Versions: 3.4.0
            Reporter: Enrico Minack


Cogrouping two grouped DataFrames in PySpark that have different group key 
cardinalities raises an error that is not very descriptive:
{code:python}
left.groupby("id", "k")
    .cogroup(right.groupby("id"))
{code}
{code:java}
Traceback (most recent call last):
py4j.protocol.Py4JJavaError: An error occurred while calling 
o726.collectToPython.
: java.lang.IndexOutOfBoundsException: 1
        at 
scala.collection.mutable.ResizableArray.apply(ResizableArray.scala:46)
        at 
scala.collection.mutable.ResizableArray.apply$(ResizableArray.scala:45)
        at scala.collection.mutable.ArrayBuffer.apply(ArrayBuffer.scala:49)
        at 
org.apache.spark.sql.catalyst.plans.physical.HashShuffleSpec.$anonfun$createPartitioning$5(partitioning.scala:650)
{code}
*Note:* This is Python-specific as cogrouping with differing group key sizes is 
not possible in Scala. The respective Scala API is fully typed on the key.

The problem is that {{EnsureRequirements.ensureDistributionAndOrdering}} calls 
into {{HashShuffleSpec.createPartitioning(clustering)}} where length of 
{{clustering}} is smaller than largest bits ({{{}v.head{}}}) in 
{{{}hashKeyPositions{}}} (EnsureRequirements.scala:159):
{code:java}
hashKeyPositions.map(v => clustering(v.head))
{code}

Possible fixes:
 # Assert identical size for group keys, and provide meaningful 
cogroup-specific error message.
 # {{EnsureRequirements}} identifies this situation and provide a meaningful 
distribution-requirements-specific error message.
 # Ideally both.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org

Reply via email to