fix(build): Fix greedy bin-packing to respect CMake positional grouping

The previous rebalancing produced groups with unequal sizes (8 files in
group0), but CMake's velox_add_grouped_tests takes files positionally in
fixed batches of VELOX_TESTS_PER_GROUP=10. This caused IndexLookupJoinTest
(378s) and NestedLoopJoinTest (46s) to spill into group0 alongside
HashJoinTest (807s), resulting in a 1800s timeout.

Fix by running greedy bin-packing with fixed capacity constraints: each
group must have exactly 10 files (8 for the last group with 78 total).
Now group0 has HashJoinTest + 9 lightest tests (~816s), while groups 1-7
are all ~419-424s each.

Author: kgpai

velox/exec/tests/CMakeLists.txt

@@ -37,99 +37,101 @@ target_link_libraries(
 )
 
 # Sources are ordered so that heavy tests are spread across groups via greedy
-# bin-packing by estimated execution time. With VELOX_TESTS_PER_GROUP=10 this
-# produces 8 groups of roughly equal wall-clock time (~8.6 min each), except
-# group0 which is dominated by HashJoinTest (~13.7 min). The previous
-# alphabetical order concentrated slow tests (HashJoinTest, IndexLookupJoinTest,
-# HashTableTest) into the same groups, creating a worst-case of ~18.8 min.
+# bin-packing by estimated execution time. With VELOX_TESTS_PER_GROUP=10,
+# CMake takes files positionally in batches of 10 (files 1-10 = group0,
+# 11-20 = group1, etc.), so each batch of 10 must be independently balanced.
+# Group0 is dominated by HashJoinTest (~13.6 min); groups 1-7 are ~7 min each.
+# The previous alphabetical order concentrated slow tests (HashJoinTest,
+# IndexLookupJoinTest, HashTableTest) into the same groups, creating a
+# worst-case of ~18.8 min.
 set(
   VELOX_EXEC_TEST_SOURCES
-  # group0 (~822s): HashJoinTest is the single heaviest test; pad with light tests.
+  # group0 (~816s): HashJoinTest is the single heaviest test; pad with light tests.
   HashJoinTest.cpp
-  ParallelProjectTest.cpp
-  AsyncConnectorTest.cpp
-  ProbeOperatorStateTest.cpp
-  ValuesTest.cpp
-  PlanNodeStatsTest.cpp
+  PrintPlanWithStatsTest.cpp
+  ThreadDebugInfoTest.cpp
   WriterFuzzerUtilTest.cpp
   UnorderedStreamReaderTest.cpp
-  # group1 (~523s)
+  ProbeOperatorStateTest.cpp
+  PlanNodeStatsTest.cpp
+  ValuesTest.cpp
+  ParallelProjectTest.cpp
+  AsyncConnectorTest.cpp
+  # group1 (~424s)
   IndexLookupJoinTest.cpp
-  NestedLoopJoinTest.cpp
-  SpatialIndexTest.cpp
-  MergerTest.cpp
-  WindowTest.cpp
-  ContainerRowSerdeTest.cpp
+  PrefixSortTest.cpp
+  HashPartitionFunctionTest.cpp
+  ExpandTest.cpp
+  SplitListenerTest.cpp
+  AssignUniqueIdTest.cpp
+  HashBitRangeTest.cpp
+  SplitTest.cpp
+  ArrowStreamTest.cpp
   SqlTest.cpp
-  PrintPlanWithStatsTest.cpp
-  WindowFunctionRegistryTest.cpp
-  TopNTest.cpp
-  # group2 (~519s)
+  # group2 (~423s)
   TableScanTest.cpp
+  AggregationTest.cpp
   MergeTest.cpp
+  LimitTest.cpp
+  TopNTest.cpp
+  SpatialIndexTest.cpp
+  PlanNodeSerdeTest.cpp
+  SpillStatsTest.cpp
+  WindowFunctionRegistryTest.cpp
+  PlanNodeToSummaryStringTest.cpp
+  # group3 (~422s)
+  MergeJoinTest.cpp
+  OperatorTraceTest.cpp
+  LocalPartitionTest.cpp
   SortBufferTest.cpp
-  MarkSortedTest.cpp
-  PartitionedOutputTest.cpp
-  SplitListenerTest.cpp
-  ThreadDebugInfoTest.cpp
+  ScaleWriterLocalPartitionTest.cpp
+  ContainerRowSerdeTest.cpp
+  EnforceSingleRowTest.cpp
+  MixedUnionWithTableScanTest.cpp
+  ColumnStatsCollectorTest.cpp
   AggregateFunctionRegistryTest.cpp
-  RoundRobinPartitionFunctionTest.cpp
-  ArrowStreamTest.cpp
-  # group3 (~516s)
-  MergeJoinTest.cpp
+  # group4 (~422s)
+  MultiFragmentTest.cpp
   OutputBufferManagerTest.cpp
-  OperatorTraceTest.cpp
-  MemoryReclaimerTest.cpp
-  FunctionResolutionTest.cpp
-  TraceUtilTest.cpp
-  HilbertIndexTest.cpp
-  PlanNodeSerdeTest.cpp
+  StreamingAggregationTest.cpp
+  MixedUnionTest.cpp
+  VectorHasherTest.cpp
+  CountingJoinTest.cpp
+  MergerTest.cpp
   ConcatFilesSpillMergeStreamTest.cpp
-  AssignUniqueIdTest.cpp
-  # group4 (~516s)
-  MultiFragmentTest.cpp
-  ExchangeClientTest.cpp
-  AggregationTest.cpp
-  MarkDistinctTest.cpp
-  MixedUnionWithTableScanTest.cpp
-  FilterToExpressionTest.cpp
-  ExpandTest.cpp
-  StreamingEnforceDistinctTest.cpp
-  CustomJoinTest.cpp
-  PlanNodeToSummaryStringTest.cpp
-  # group5 (~515s)
+  AddressableNonNullValueListTest.cpp
+  RoundRobinPartitionFunctionTest.cpp
+  # group5 (~421s)
   RowContainerTest.cpp
-  SpillTest.cpp
-  StreamingAggregationTest.cpp
-  HashJoinBridgeTest.cpp
-  PrefixSortTest.cpp
-  HashPartitionFunctionTest.cpp
+  HashTableTest.cpp
+  NestedLoopJoinTest.cpp
+  MemoryReclaimerTest.cpp
+  RowNumberTest.cpp
   EnforceDistinctTest.cpp
-  TaskListenerTest.cpp
-  EnforceSingleRowTest.cpp
-  HashBitRangeTest.cpp
-  # group6 (~514s)
+  MarkSortedTest.cpp
+  TraceUtilTest.cpp
+  FilterToExpressionTest.cpp
+  ScaledScanControllerTest.cpp
+  # group6 (~421s)
   TableWriterTest.cpp
-  LocalPartitionTest.cpp
+  ExchangeClientTest.cpp
   OrderByTest.cpp
-  LimitTest.cpp
-  RowNumberTest.cpp
-  ScaledScanControllerTest.cpp
-  VectorHasherTest.cpp
-  SpillStatsTest.cpp
-  ColumnStatsCollectorTest.cpp
-  AddressableNonNullValueListTest.cpp
-  # group7 (~513s)
-  SpillerTest.cpp
-  HashTableTest.cpp
-  MixedUnionTest.cpp
-  ScaleWriterLocalPartitionTest.cpp
+  HashJoinBridgeTest.cpp
   FilterProjectTest.cpp
-  TopNRowNumberTest.cpp
-  CountingJoinTest.cpp
   UnnestTest.cpp
+  StreamingEnforceDistinctTest.cpp
+  FunctionResolutionTest.cpp
+  TaskListenerTest.cpp
   PlanNodeToStringTest.cpp
-  SplitTest.cpp
+  # group7 (~419s)
+  SpillerTest.cpp
+  SpillTest.cpp
+  WindowTest.cpp
+  PartitionedOutputTest.cpp
+  TopNRowNumberTest.cpp
+  MarkDistinctTest.cpp
+  CustomJoinTest.cpp
+  HilbertIndexTest.cpp
 )
 
 set(