Ala Luszczak created SPARK-26878:
------------------------------------
Summary: QueryTest.compare does not handle maps with array keys
correctly
Key: SPARK-26878
URL: https://issues.apache.org/jira/browse/SPARK-26878
Project: Spark
Issue Type: Bug
Components: SQL, Tests
Affects Versions: 2.4.0
Reporter: Ala Luszczak
The current strategy for comparing Maps is sorting the (key, value) tuples by
_.toString, zipping tuples from both maps together, and then comparing tuples
within each of the pairs separately.
See:
https://github.com/apache/spark/blob/ac9c0536bc518f173f2ff53bee42b7a89d28ee20/sql/core/src/test/scala/org/apache/spark/sql/QueryTest.scala#L344-L346
This is not ideal for byte arrays. The string representations of byte arrays
looks like “[B@7d263ddc” and has nothing to do with values actually contained
within the array.
Hence, if a map has byte array keys, then random values get compared with each
other, which can result in false negatives.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
