Jim Hughes created FLINK-39677:
----------------------------------
Summary: ARRAY_SORT fails at runtime with "Comparison method
violates its general contract!" on arrays with duplicate elements
Key: FLINK-39677
URL: https://issues.apache.org/jira/browse/FLINK-39677
Project: Flink
Issue Type: Bug
Affects Versions: 2.2.1, 2.0.2, 2.1.2, 1.20.4, 2.3.0
Reporter: Jim Hughes
*Title:* ARRAY_SORT fails at runtime with "Comparison method violates its
general contract!" on arrays with duplicate elements
*Affects Versions:* 2.0 and earlier (wherever {{ArraySortFunction}} exists)
h3. Description
{{ArraySortFunction.ArraySortComparator.compare}} is built from a single SQL
{{>}} evaluator and returns {{+1}} or {{-1}} based on the boolean result — it
never returns {{0}}. For two equal elements {{a}} and {{b}}, both {{a > b}} and
{{b > a}} are {{FALSE}}, so {{compare(a, b) == compare(b, a) == -1}}. This
violates {{java.util.Comparator}}'s antisymmetry requirement.
TimSort detects the contradiction once the array is large enough to take the
merge path (≥32 elements with duplicates), throwing:
{noformat}
java.lang.IllegalArgumentException: Comparison method violates its general
contract!
at java.util.TimSort.mergeHi(TimSort.java:903)
at java.util.Arrays.sort(Arrays.java:1234)
at
org.apache.flink.table.runtime.functions.scalar.ArraySortFunction.eval(ArraySortFunction.java:91)
{noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
