IGNITE-28899 Calcite. Fix litmus error during join by zstan · Pull Request #13389 · apache/ignite

zstan · 2026-07-21T14:29:49Z

https://issues.apache.org/jira/browse/IGNITE-28899

zstan · 2026-07-24T10:09:21Z

+                continue;
+
+            // Keys can relate to affinity not pk, just assumption here.
+            ImmutableIntList distKeys = table.distribution().getKeys();


of course we need to have 'real pk' here, but fix it right now - seems too huge.

zstan · 2026-07-24T10:27:36Z

            inputRows += mq.getRowCount(input);

-        double mem = 0.5 * inputRows * aggregateFieldsCount() * IgniteCost.AVERAGE_FIELD_SIZE;
+        double mem = 0.7 * inputRows * aggregateFieldsCount() * IgniteCost.AVERAGE_FIELD_SIZE;


empiric constants need to be fixed cause row count estimation is changed for now and some tests (2-3) are fails, all empirics need to be reevaluated during further (i hope) statistics activity changes.

zstan · 2026-07-24T10:38:57Z

        prepareTables();

-        String sql = "SELECT sum(i.price * i.amount)" +
+        String sql = "SELECT /*+ MERGE_JOIN */ sum(i.price * i.amount)" +


Sometimes i observe hashJoin here, i really don`t know what algo is more preferably here and i think it need to be discovered more accurate in further activities.

zstan · 2026-07-24T10:50:44Z

+    private static final int DATE_DIM_SIZE = 73_049;
+
+    /** */
+    private static final String SELECT = "SELECT /*+ DISABLE_RULE(" +


Of course it work perfectly well without this disablings, i just afraid about some king of flaky...

Copilot

Pull request overview

Note

Copilot couldn't run its full agentic review because it didn't start before the timeout. Make sure your repository has a runner available, or add a copilot-code-review.yml file specifying one with the runs-on attribute. See the docs for more details.

Updates Calcite planner metadata/costing and merge-join execution behavior to address a Calcite litmus failure seen during join planning/execution (IGNITE-28899), and adds regression tests to cover the scenario.

Changes:

Reworked join row-count estimation and adjusted join cost formulas/constants to influence plan selection.
Modified merge-join runtime node control flow around buffering and early termination.
Added new planner/integration tests and updated several existing assertions to match the new behavior.

Reviewed changes

Copilot reviewed 23 out of 23 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
modules/calcite/src/test/java/org/apache/ignite/testsuites/PlannerTestSuite.java	Registers the new join row-count estimation planner test in the suite.
modules/calcite/src/test/java/org/apache/ignite/testsuites/IntegrationTestSuite.java	Adds the new litmus regression integration test to the suite.
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/planner/hints/JoinTypeHintPlannerTest.java	Relaxes/adjusts hint-related plan assertions (notably around wrong-table hints).
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/planner/UncollectPlannerTest.java	Updates plan shape expectations for UNNEST + join planning.
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/planner/PlanSplitterTest.java	Refactors colocation assignments setup to reuse a single assignments instance.
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/planner/JoinRowCountEstimationTest.java	New planner tests validating join row-count estimation logic.
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/planner/CorrelatedSubqueryPlannerTest.java	Updates plan assertion for table scan condition presence.
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/planner/AggregatePlannerTest.java	Removes an early return that skipped part of a test scenario.
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/integration/QueryBlockingTaskExecutorIntegrationTest.java	Adds hint to force merge join and simplifies plan matching into a single regex check.
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/integration/LitmusCheckIntegrationTest.java	New integration test meant to catch Calcite litmus exceptions/regressions.
modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite/integration/DistributedJoinIntegrationTest.java	Forces MERGE_JOIN via hint for deterministic join plan coverage.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/util/Commons.java	Adds a helper to build a target mapping from a projected bitset.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/rel/set/IgniteSetOp.java	Adjusts memory estimation coefficient used in set-op costing.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/rel/IgniteNestedLoopJoin.java	Revises NLJ cost computation to use metadata row-count and applies an additional penalty factor.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/rel/IgniteMergeJoin.java	Revises merge join cost computation to use metadata row-count.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/rel/IgniteHashJoin.java	Revises hash join cost computation to use metadata row-count.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/rel/IgniteCorrelatedNestedLoopJoin.java	Revises CNLJ cost computation to use metadata row-count.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/metadata/cost/IgniteCost.java	Changes the hash lookup cost constant affecting planner decisions.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/metadata/IgniteMdSelectivity.java	Updates metadata provider wiring and exposes selectivity constants for tests/consumers.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/metadata/IgniteMdRowCount.java	Implements custom join row-count estimation with PK/FK heuristics and new metadata provider wiring.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/metadata/IgniteMdColumnOrigins.java	Updates metadata provider wiring and minor indentation cleanup.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/exec/rel/MergeJoinNode.java	Reworks buffering/join scheduling and replaces “finishing” cleanup with immediate downstream end conditions.
modules/calcite/src/main/java/org/apache/ignite/internal/processors/query/calcite/CalciteQueryProcessor.java	Removes a previously-added workaround disabling Correlate assertions.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

zstan · 2026-08-20T06:14:58Z

-        assertPlan(sql, schema, hasFragmentsCount(3).and(hasChildThat(isInstanceOf(IgniteTrimExchange.class))));
+        // First table is replicated and planned with Exchange to colocate data, but set of nodes for partitioned
+        // table is differ, so exchange is added after fragments split for colocation.
+        assertPlan(sql, schema, hasFragmentsCount(3)


Current plan:

IgniteMergeJoin(condition=[=($0, $3)], joinType=[inner]): rowcount = 5.625E9, cumulative cost = IgniteCost [rowCount=5.627075E9, cpu=2.2525784224829422E10, memory=4900000.0, io=0.0, network=4000000.0] IgniteSort(sort0=[$0], dir0=[ASC]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=575000.0, cpu=4600679.76331665, memory=900000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, DEVELOPER]], filters=[=(+($t2, 1), ?0)]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=2000000.0, memory=0.0, io=0.0, network=0.0] IgniteExchange(distribution=[single]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=1500000.0, cpu=2.118354506610649E7, memory=4000000.0, io=0.0, network=4000000.0] IgniteSort(sort0=[$0], dir0=[ASC]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=1000000.0, cpu=2.068354506610649E7, memory=4000000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, PROJECT]], requiredColumns=[{0, 2}]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=500000.0, memory=0.0, io=0.0, network=0.0]

master branch plan:

IgniteProject(ID=[$2], NAME=[$3], PROJECTID=[$4], ID0=[$0], VER0=[$1]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1672500.0, cpu=8572500.0, memory=1170000.0, io=0.0, network=225000.0] IgniteExchange(distribution=[single]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1661250.0, cpu=8561250.0, memory=1170000.0, io=0.0, network=225000.0] IgniteHashJoin(condition=[=($2, $0)], joinType=[inner]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1650000.0, cpu=8550000.0, memory=1170000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, PROJECT]], requiredColumns=[{0, 2}]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=500000.0, memory=0.0, io=0.0, network=0.0] IgniteTrimExchange(distribution=[affinity[identity=hash, cacheId=1355342585][0]]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=575000.0, cpu=2300000.0, memory=0.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, DEVELOPER]], filters=[=(+($t2, 1), ?0)]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=2000000.0, memory=0.0, io=0.0, network=0.0]

zstan · 2026-08-20T07:16:32Z

-        // fragments split to send data to initiator node.
-        assertPlan(sql, schema, hasFragmentsCount(3).and(hasChildThat(isInstanceOf(IgniteTrimExchange.class))));
+        // table is differ, so exchange is added after fragments split for colocation.
+        assertPlan(sql, schema, hasFragmentsCount(2)


Current plan:

IgniteMergeJoin(condition=[=($2, $3)], joinType=[inner], variablesSet=[[]], leftCollation=[[2]], rightCollation=[[0]]): rowcount = 5.625E9, cumulative cost = IgniteCost [rowCount=5.627075E9, cpu=2.2525784224829422E10, memory=4900000.0, io=0.0, network=4000000.0], id = 3041 IgniteSort(sort0=[$2], dir0=[ASC]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=575000.0, cpu=4600679.76331665, memory=900000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, DEVELOPER]], filters=[=(+($t2, 1), ?0)]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=2000000.0, memory=0.0, io=0.0, network=0.0] IgniteExchange(distribution=[single]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=1500000.0, cpu=2.118354506610649E7, memory=4000000.0, io=0.0, network=4000000.0] IgniteSort(sort0=[$0], dir0=[ASC]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=1000000.0, cpu=2.068354506610649E7, memory=4000000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, PROJECT]], requiredColumns=[{0, 2}]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=500000.0, memory=0.0, io=0.0, network=0.0]

master branch plan:

IgniteProject(ID=[$2], NAME=[$3], PROJECTID=[$4], ID0=[$0], VER0=[$1]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1672500.0, cpu=8572500.0, memory=1170000.0, io=0.0, network=225000.0] IgniteExchange(distribution=[single]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1661250.0, cpu=8561250.0, memory=1170000.0, io=0.0, network=225000.0] IgniteHashJoin(condition=[=($4, $0)], joinType=[inner], variablesSet=[[]]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1650000.0, cpu=8550000.0, memory=1170000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, PROJECT]], requiredColumns=[{0, 2}]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=500000.0, memory=0.0, io=0.0, network=0.0] IgniteTrimExchange(distribution=[affinity[identity=hash, cacheId=1355342585][2]]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=575000.0, cpu=2300000.0, memory=0.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, DEVELOPER]], filters=[=(+($t2, 1), ?0)]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=2000000.0, memory=0.0, io=0.0, network=0.0]

zstan · 2026-08-20T07:30:37Z

-        assertPlan(sql, schema, hasFragmentsCount(3).and(hasChildThat(isInstanceOf(IgniteTrimExchange.class))));
+        // First table is replicated and planned with Exchange to colocate data, but set of nodes for partitioned
+        // table is differ, so exchange is added after fragments split for colocation.
+        assertPlan(sql, schema, hasFragmentsCount(2)


Current plan:

IgniteMergeJoin(condition=[=($0, $3)], joinType=[inner], variablesSet=[[]], leftCollation=[[0]], rightCollation=[[0]]): rowcount = 5.625E9, cumulative cost = IgniteCost [rowCount=5.627075E9, cpu=2.2525784224829422E10, memory=4900000.0, io=0.0, network=4000000.0] IgniteSort(sort0=[$0], dir0=[ASC]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=575000.0, cpu=4600679.76331665, memory=900000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, DEVELOPER]], filters=[=(+($t2, 1), ?0)]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=2000000.0, memory=0.0, io=0.0, network=0.0] IgniteExchange(distribution=[single]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=1500000.0, cpu=2.118354506610649E7, memory=4000000.0, io=0.0, network=4000000.0] IgniteSort(sort0=[$0], dir0=[ASC]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=1000000.0, cpu=2.068354506610649E7, memory=4000000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, PROJECT]], requiredColumns=[{0, 2}]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=500000.0, memory=0.0, io=0.0, network=0.0]

master branch plan:

IgniteProject(ID=[$2], NAME=[$3], PROJECTID=[$4], ID0=[$0], VER0=[$1]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1672500.0, cpu=8572500.0, memory=1170000.0, io=0.0, network=225000.0] IgniteExchange(distribution=[single]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1661250.0, cpu=8561250.0, memory=1170000.0, io=0.0, network=225000.0] IgniteHashJoin(condition=[=($2, $0)], joinType=[inner], variablesSet=[[]]): rowcount = 11250.0, cumulative cost = IgniteCost [rowCount=1650000.0, cpu=8550000.0, memory=1170000.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, PROJECT]], requiredColumns=[{0, 2}]): rowcount = 500000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=500000.0, memory=0.0, io=0.0, network=0.0] IgniteTrimExchange(distribution=[affinity[identity=hash, cacheId=1355342585][0]]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=575000.0, cpu=2300000.0, memory=0.0, io=0.0, network=0.0] IgniteTableScan(table=[[PUBLIC, DEVELOPER]], filters=[=(+($t2, 1), ?0)]): rowcount = 75000.0, cumulative cost = IgniteCost [rowCount=500000.0, cpu=2000000.0, memory=0.0, io=0.0, network=0.0]

Also want to show that previously expected plan still possible but for lower table size, if we try to change
protected static final int DEFAULT_TBL_SIZE = 10;//500_000;
below plan will be raized:

IgniteProject(ID=[$2], NAME=[$3], PROJECTID=[$4], ID0=[$0], VER0=[$1]): rowcount = 2.25, cumulative cost = IgniteCost [rowCount=28.25, cpu=83.0, memory=23.4, io=0.0, network=45.0], id = 4147 IgniteExchange(distribution=[single]): rowcount = 2.25, cumulative cost = IgniteCost [rowCount=26.0, cpu=80.75, memory=23.4, io=0.0, network=45.0], id = 4146 IgniteHashJoin(condition=[=($2, $0)], joinType=[inner], variablesSet=[[]]): rowcount = 2.25, cumulative cost = IgniteCost [rowCount=23.75, cpu=78.5, memory=23.4, io=0.0, network=0.0], id = 4145 IgniteTableScan(table=[[PUBLIC, PROJECT]], requiredColumns=[{0, 2}]): rowcount = 10.0, cumulative cost = IgniteCost [rowCount=10.0, cpu=10.0, memory=0.0, io=0.0, network=0.0], id = 1784 IgniteTrimExchange(distribution=[affinity[identity=hash, cacheId=1355342585][0]]): rowcount = 1.5, cumulative cost = IgniteCost [rowCount=11.5, cpu=46.0, memory=0.0, io=0.0, network=0.0], id = 4144 IgniteTableScan(table=[[PUBLIC, DEVELOPER]], filters=[=(+($t2, 1), ?0)]): rowcount = 1.5, cumulative cost = IgniteCost [rowCount=10.0, cpu=40.0, memory=0.0, io=0.0, network=0.0], id = 1793

Thus i think - if we want a fine tuning here - it need to be done under different issue.

tkalkirill · 2026-08-24T09:42:10Z

            .and(hasDistribution(IgniteDistributions.affinity(0, null, "hash")))),
            algo.rulesToDisable);

-        // TODO: https://issues.apache.org/jira/browse/IGNITE-16334 Eventually planner skips optimal join plan.


Why did you delete the link? Does that mean the problem with the deleted link has been fixed?

yes, i run huge number of runs here - all ok, thus i delete it.

tkalkirill · 2026-08-24T10:49:57Z

-            rows * (IgniteCost.ROW_COMPARISON_COST + IgniteCost.ROW_PASS_THROUGH_COST), 0, rightSize, 0);
+        double rowCnt = mq.getRowCount(this);
+
+        RelOptCost cost = costFactory.makeCost(rowCnt,


Could we keep the CPU cost proportional to leftCnt * rightCnt? A nested-loop join still compares every input pair, while the output row count may be orders of magnitude smaller. Multiplying it by 10 does not preserve the actual complexity and may make NLJ incorrectly cheaper than a hash join.

All such playing with coeffs is some kind of hack and i agree that multiplying complexity looks strange but all these hacks need to be eliminated when we will implement a real statistics for calcite now - it`s just a kind of assumptions. The main goal of this fix - is to store test plans and expectations as more as possible. If you want to play with it - try to comment multiplication and run PlanSplitterTest tests for example.

I agree that the current cost model is heuristic and preserving existing plans is useful. However, output cardinality and the amount of work are different estimates. Even with exact statistics, an NLJ over two 1M-row inputs may perform up to 1e12 comparisons while producing a single row, so 10 * rowCnt still severely underestimates its CPU cost.
Could we keep rowCnt as the output cardinality, but calculate CPU from leftCnt * rightCnt, using rowCnt only for the pass-through part? Also, multiplyBy(10) scales memory cost as well as CPU, which seems unintended.
If this changes PlanSplitterTest, I suggest reviewing the affected plans and updating justified expectations rather than tuning a global coefficient specifically to preserve snapshots.

Change it, if I correctly understand your idea.

tkalkirill · 2026-08-24T10:51:30Z

    /** {@inheritDoc} */
    @Override public Double getRowCount(Join rel, RelMetadataQuery mq) {
-        return rel.estimateRowCount(mq);
+        return joinRowCount(mq, rel);


This bypasses IgniteCorrelatedNestedLoopJoin.estimateRowCount(), which removes the join-condition selectivity because it is already applied by an external filter. The generic estimator may therefore count the condition twice. Could we preserve the specialized CNL handling here?

Missunderstand plz explain this ? Also seems this behind of scope of this issue but if this relevant (i still don`t know why) - seems it need to be investigated.

Before this change, getRowCount(Join, mq) called rel.estimateRowCount(mq). Due to virtual dispatch, an IgniteCorrelatedNestedLoopJoin invoked its specialized estimateRowCount() override.
CorrelatedNestedLoopJoinRule pushes the join condition into an external filter on the right input, while the resulting CNL still keeps the same join condition. Therefore, the right input row count already includes the condition selectivity. The generic joinRowCount() applies the join-condition selectivity again.
IgniteCorrelatedNestedLoopJoin.estimateRowCount() explicitly compensates for this by dividing the generic estimate by mq.getSelectivity(this, getCondition()).
Calling joinRowCount(mq, rel) directly bypasses that override and reintroduces the double application. This is relevant to this change because the previous dispatch preserved the specialized behavior, while the new dispatch does not.
Could we add a specific metadata overload for IgniteCorrelatedNestedLoopJoin, delegating to rel.estimateRowCount(mq), and cover it with a regression test?

ok, I changed it.

zstan · 2026-08-26T05:58:02Z

Current cut shows:

TpchBenchmark, scale = 0.01, q = 5. (H2 just for comparison here, not related to an issue of course)

PR:
Benchmark             (engine)  (queryId)  (scale)  Mode  Cnt     Score     Error  Units
TpchBenchmark.cached   CALCITE          5     0.01  avgt    3    25.519 ±   4.548  ms/op
TpchBenchmark.cached        H2          5     0.01  avgt    3   193.700 ± 168.796  ms/op
TpchBenchmark.cold     CALCITE          5     0.01    ss       1093.271            ms/op
TpchBenchmark.cold          H2          5     0.01    ss        816.027            ms/op

master 
Benchmark             (engine)  (queryId)  (scale)  Mode  Cnt     Score    Error  Units
TpchBenchmark.cached   CALCITE          5     0.01  avgt    3   566.541 ± 49.195  ms/op
TpchBenchmark.cold     CALCITE          5     0.01    ss       1564.145           ms/op

tkalkirill · 2026-08-26T07:33:52Z

Perhaps it would be more correct to compare before and after your changes?

zstan · 2026-08-26T09:00:43Z

Perhaps it would be more correct to compare before and after your changes?

I just have a fresh master rebase with my PR, but ok re-check it:

git checkout ff25813
with calcite module rebuild

Benchmark             (engine)  (queryId)  (scale)  Mode  Cnt     Score     Error  Units
TpchBenchmark.cached   CALCITE          5     0.01  avgt    3   603.527 ± 168.897  ms/op
TpchBenchmark.cold     CALCITE          5     0.01    ss       1530.101            ms/op

zstan · 2026-09-16T12:47:45Z

-                    .and(hasDistribution(broadcast()))
-                ))
-            )
+            .and(nodeOrAnyChild(isInstanceOf(IgniteUncollect.class)))


current plan:

IgniteHashJoin(condition=[=($0, $2)], joinType=[inner], variablesSet=[[]]): rowcount = 1.0, cumulative cost = IgniteCost [rowCount=1000007.0, cpu=1000016.0, memory=30.4, io=0.0, network=4000000.0] IgniteExchange(distribution=[single]): rowcount = 500000.0 IgniteTableScan(table=[[PUBLIC, HASH_TBL]]): rowcount = 500000.0 IgniteUncollect IgniteProject(EXPR$0=[ARRAY(1, 2, 3)]) IgniteValues(tuples=[[{ 0 }]])

master branch plan:

IgniteProject(ID=[$1], NAME=[$2], EXPR$0=[$0]): rowcount = 4.0, cumulative cost = IgniteCost [rowCount=2300018.0, cpu=9200030.0, memory=0.0, io=0.0, network=48.0], id = 2654 IgniteExchange(distribution=[single]): IgniteCorrelatedNestedLoopJoin(condition=[=($1, $0)], joinType=[inner], variablesSet=[[$cor0]], variablesSet=[[0]], correlationVariables=[[$cor0]]): rowcount = 4.0 IgniteTrimExchange(distribution=[hash[0]]) IgniteUncollect IgniteProject(EXPR$0=[ARRAY(1, 2, 3)]) IgniteValues(tuples=[[{ 0 }]]) IgniteTableScan(table=[[PUBLIC, HASH_TBL]], filters=[=($t0, $cor0.EXPR$0)]): rowcount = 75000.0

zstan · 2026-09-24T08:46:40Z

@tkalkirill @alex-plekhanov make a review plz ?

zstan force-pushed the ignite-28899 branch from bd1871e to f68fd75 Compare July 23, 2026 06:25

zstan commented Jul 23, 2026

View reviewed changes

Comment thread ...rc/main/java/org/apache/ignite/internal/processors/query/calcite/exec/rel/MergeJoinNode.java Outdated

zstan commented Jul 23, 2026

View reviewed changes

Comment thread ...main/java/org/apache/ignite/internal/processors/query/calcite/metadata/IgniteMdRowCount.java Outdated

zstan commented Jul 24, 2026

View reviewed changes

Comment thread .../main/java/org/apache/ignite/internal/processors/query/calcite/metadata/cost/IgniteCost.java Outdated

zstan commented Jul 24, 2026

View reviewed changes

Comment thread ...t/java/org/apache/ignite/internal/processors/query/calcite/planner/UncollectPlannerTest.java

zstan force-pushed the ignite-28899 branch from 5c99502 to 62aa8ed Compare July 24, 2026 12:29

zstan marked this pull request as ready for review July 24, 2026 12:30

zstan requested a review from Copilot July 24, 2026 12:30

Copilot AI reviewed Jul 24, 2026

View reviewed changes

Copilot started reviewing on behalf of zstan July 24, 2026 13:00

View session

zstan force-pushed the ignite-28899 branch from 62aa8ed to 13417a9 Compare August 11, 2026 14:23

zstan force-pushed the ignite-28899 branch 3 times, most recently from a6dbda5 to eea65ee Compare August 19, 2026 13:54

zstan commented Aug 20, 2026

View reviewed changes

zstan force-pushed the ignite-28899 branch 3 times, most recently from 809cdcb to 063753c Compare August 20, 2026 11:08

tkalkirill reviewed Aug 24, 2026

View reviewed changes

zstan added 2 commits August 25, 2026 17:52

IGNITE-28899 Calcite. Fix litmus error during join

4b959cc

fix after review

e6b318b

zstan force-pushed the ignite-28899 branch from 7743097 to e6b318b Compare August 25, 2026 14:52

fix after review changes from IGNITE-25129

1322ea1

zstan added 4 commits August 26, 2026 14:10

change IgniteNestedLoopJoin cost after review

721c2f4

concreete getRowCount for IgniteCorrelatedNestedLoopJoin

5e7216b

add test

0e00b1b

CNLJ test ok

9455947

zstan force-pushed the ignite-28899 branch from 7535ee5 to 9455947 Compare August 27, 2026 14:14

zstan and others added 13 commits August 27, 2026 20:23

assertionOnDeserialization fix

ba1fcea

add litmus comment

5dd9336

Merge branch 'master' into ignite-28899

b3999a5

Merge branch 'master' into ignite-28899

cff9d91

fix import

1e78f11

cl fix

6025901

interm

2c5c8cb

self review

f821da4

self review cl fix

f844134

mdcount fix

4850afd

style check

1cf3a54

fix test

faecb58

revert test

14780e1

zstan commented Sep 16, 2026

View reviewed changes

unused import

cffdb81

Merge branch 'master' into ignite-28899

c914121

Conversation

zstan commented Jul 21, 2026 • edited by tkalkirill Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zstan Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zstan Aug 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

zstan commented Aug 26, 2026

Uh oh!

tkalkirill commented Aug 26, 2026

Uh oh!

zstan commented Aug 26, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zstan commented Sep 24, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

zstan commented Jul 21, 2026 •

edited by tkalkirill

Loading

zstan Jul 24, 2026 •

edited

Loading

zstan Aug 27, 2026 •

edited

Loading