pedrumj2 opened a new pull request, #13584: URL: https://github.com/apache/gluten/pull/13584
## What changes are proposed in this pull request? Native [arrays_overlap](https://spark.apache.org/docs/3.5.6/api/sql/index.html#arrays_overlap) now treats `-0.0` and `0.0` as different values, as Spark does, so it returns `false` for `array(-0.0D)` and `array(0.0D)` instead of `true`. [facebookincubator/velox#19805](https://github.com/facebookincubator/velox/pull/19805) added Spark's `arrays_overlap` to Velox, and this repo's Velox has included it since apache/gluten#13540. [registerPrestoVectorFunctions](https://github.com/apache/gluten/blob/01e71ec06b814c7b63af0306c24d9974bd2a30e8/cpp/velox/operators/functions/RegistrationAllFunctions.cc#L41-L44) still registers Velox's Presto `arrays_overlap` under the same name, and native queries run that version, which treats `-0.0` and `0.0` as equal. That registration has been the only one in `registerPrestoVectorFunctions` since apache/gluten#13511, so this PR removes the function and its call. Native queries that compare arrays of floating point values now return Spark's answer on Spark 4.1 and earlier. [NormalizeFloatingNumbers](https://github.com/apache/spark/blob/v4.2.0/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/NormalizeFloatingNumbers.scala#L112) normalizes `arrays_overlap`'s inputs on Spark 4.2, so Spark 4.2 returns `true` here, but Gluten passes that normalization through without applying it and returns `false`. Until apache/gluten#13565 maps the normalization to Velox, this PR excludes Spark 4.2's [SPARK-54918: arrays_overlap normalizes -0.0 to +0.0](https://github.com/apache/spark/blob/v4.2.0/sql/core/src/test/scala/org/apache/spark/sql/DataFrameFunctionsSuite.scala#L6534-L6539) test in [VeloxTestSettings](https://github.com/apache/gluten/blob/01e71ec06b814c7b63af0306c24d9974bd2a30e8/gluten-ut/spark42/src/test/scala/org/apache/gluten/utils/velox/VeloxTestSettings.scala#L1025), and runs the `arrays_overlap with signed zeros` test it adds to [ScalarFunctionsValidateSuite](https://github.com/apache/gluten/blob/01e71ec06b814c7b63af0306c24d9974bd2a30e 8/backends-velox/src/test/scala/org/apache/gluten/functions/ScalarFunctionsValidateSuite.scala#L1225) on Spark 4.1 and earlier. TODO comments mark both for removal once apache/gluten#13565 lands. ## How was this patch tested? ### Reproduction 1. Clone the repo and check out the base commit: ```bash git clone https://github.com/apache/gluten.git cd gluten git checkout 01e71ec06b814c7b63af0306c24d9974bd2a30e8 ``` 2. Build the native libraries and the bundle jar: ```bash docker run --rm -v "$PWD":/work/gluten:z apache/gluten:centos-9-jdk8 bash -c ' source /opt/rh/gcc-toolset-12/enable cd /work/gluten ./dev/builddeps-veloxbe.sh --run_setup_script=OFF ./build/mvn clean install -Pbackends-velox -Pspark-3.5 -DskipTests ' ``` 3. Fetch Spark 3.5.5: ```bash curl -sSL https://archive.apache.org/dist/spark/spark-3.5.5/spark-3.5.5-bin-hadoop3.tgz | tar -xz ``` 4. Write the input table and the query: ```bash cat > write_input.sql <<'EOF' CREATE TABLE arrays USING parquet AS SELECT array(-0.0D) AS a, array(0.0D) AS b; EOF cat > query.sql <<'EOF' SELECT arrays_overlap(a, b) FROM arrays; EXPLAIN SELECT arrays_overlap(a, b) FROM arrays; EOF ``` 5. Write the table with plain Spark, then query it with Gluten: ```bash docker run --rm -v "$PWD":/work/gluten:z apache/gluten:centos-9-jdk8 bash -c ' cd /work/gluten rm -rf metastore_db spark-warehouse spark-3.5.5-bin-hadoop3/bin/spark-sql --master "local[2]" -f write_input.sql 2>/dev/null BUNDLE=$(ls package/target/gluten-velox-bundle-spark3.5_2.12-linux_amd64-*.jar) spark-3.5.5-bin-hadoop3/bin/spark-sql --master "local[2]" \ --jars "$BUNDLE" --driver-class-path "$BUNDLE" \ --conf spark.plugins=org.apache.gluten.GlutenPlugin \ --conf spark.memory.offHeap.enabled=true \ --conf spark.memory.offHeap.size=4g \ --conf spark.shuffle.manager=org.apache.spark.shuffle.sort.ColumnarShuffleManager \ -f query.sql 2>/dev/null ' ``` On the base commit, the native plan returns `true`: ``` true == Physical Plan == VeloxColumnarToRow +- ^(1) ProjectExecTransformer [arrays_overlap(a#0, b#1) AS arrays_overlap(a, b)#10] +- ^(1) FileFileSourceScanExecTransformer parquet spark_catalog.default.arrays[a#0,b#1] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex(1 paths)[file:/work/gluten/spark-warehouse/arrays], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<a:array<double>,b:array<double>> NativeFilters: [] ``` On this pull request's branch, rerunning steps 2 and 5 returns `false`, still in the native plan: ``` false == Physical Plan == VeloxColumnarToRow +- ^(1) ProjectExecTransformer [arrays_overlap(a#0, b#1) AS arrays_overlap(a, b)#10] +- ^(1) FileFileSourceScanExecTransformer parquet spark_catalog.default.arrays[a#0,b#1] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex(1 paths)[file:/work/gluten/spark-warehouse/arrays], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<a:array<double>,b:array<double>> NativeFilters: [] ``` Spark without Gluten, which is step 5's second command without the `--jars`, `--driver-class-path` and `--conf` options, returns the same `false`: ``` false == Physical Plan == *(1) Project [arrays_overlap(a#0, b#1) AS arrays_overlap(a, b)#10] +- *(1) ColumnarToRow +- FileScan parquet spark_catalog.default.arrays[a#0,b#1] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex(1 paths)[file:/work/gluten/spark-warehouse/arrays], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<a:array<double>,b:array<double>> ``` ### Automated tests `arrays_overlap with signed zeros` compares Gluten with Spark on arrays that hold `-0.0` and `0.0`. On the base commit's native libraries, the new test fails: ```bash ./build/mvn -o -B test -Pbackends-velox -Pspark-3.5 -pl backends-velox -Dspotless.check.skip=true -Dscalastyle.skip=true -Dcheckstyle.skip=true -Dtest=none -DfailIfNoTests=false -Dsuites="org.apache.gluten.functions.ScalarFunctionsValidateSuite arrays_overlap" ``` ``` - arrays_overlap - arrays_overlap with signed zeros *** FAILED *** … !== Correct Answer - 2 == == Gluten Answer - 2 == struct<> struct<> ![false] [true] [true] [true] (GlutenQueryTest.scala:460) … Tests: succeeded 1, failed 1, canceled 0, ignored 0, pending 0 ``` With this change: ```bash ./build/mvn -o -B test -Pbackends-velox -Pspark-3.5 -pl backends-velox -Dspotless.check.skip=true -Dscalastyle.skip=true -Dcheckstyle.skip=true -Dtest=none -DfailIfNoTests=false -DwildcardSuites=org.apache.gluten.functions.ScalarFunctionsValidateSuite ``` ``` ScalarFunctionsValidateSuite: … - arrays_overlap … - arrays_overlap with signed zeros … Tests: succeeded 91, failed 0, canceled 0, ignored 0, pending 0 All tests passed. ``` ```bash ./build/mvn -o -B test -Pspark-ut -Pbackends-velox -Pspark-3.5 -pl gluten-ut/spark35 -Dspotless.check.skip=true -Dscalastyle.skip=true -Dcheckstyle.skip=true -Dtest=none -DfailIfNoTests=false -DwildcardSuites=org.apache.spark.sql.catalyst.expressions.GlutenCollectionExpressionsSuite,org.apache.spark.sql.GlutenDataFrameFunctionsSuite ``` ``` GlutenCollectionExpressionsSuite: … - ArraysOverlap … GlutenDataFrameFunctionsSuite: … - arrays_overlap function … Tests: succeeded 194, failed 0, canceled 0, ignored 7, pending 0 All tests passed. ``` ## Was this patch authored or co-authored using generative AI tooling? Generated-by: Co-authored with Claude Code (claude-opus-5-5) <!-- Thank you for submitting a pull request! Here are some tips: 1. For first-time contributors, please read our contributing guide: https://github.com/apache/gluten/blob/main/CONTRIBUTING.md 2. If necessary, create a GitHub issue for discussion beforehand to avoid duplicate work. 3. If the PR is specific to a single backend, include [VL] or [CH] in the PR title to indicate the Velox or ClickHouse backend, respectively. 4. If the PR is not ready for review, please mark it as a draft. --> -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
