Gluten 2026 Roadmap

### Description

Time to define 2026 features. Feel free to add more or comment.
For more information about the achievements in 2025 Roadmap, please check https://github.com/apache/gluten/issues/8226
For more information about the achievements in 2024 Roadmap, please check https://github.com/apache/gluten/issues/4709

Apache Gluten graduated to an ASF Top-Level Project in February 2026. This roadmap tracks the community's goals for 2026.

---

### Spark Compatibility

- [ ] Spark 4.0/4.1 GA support — fix all disabled test suites (https://github.com/apache/gluten/issues/11550, https://github.com/apache/gluten/issues/11400)
- [ ] TIMESTAMP_NTZ type support (https://github.com/apache/gluten/issues/11622, https://github.com/apache/gluten/pull/11626)
- [ ] Variant shredding support for Parquet reader and writer (Spark 4.0) (https://github.com/apache/gluten/issues/11371)
- [ ] Parquet type widening support (SPARK-40876) (https://github.com/apache/gluten/issues/11683, https://github.com/apache/gluten/pull/11719)
- [ ] Other Spark Type Support: DayTimeIntervalType, CalendarIntervalType, TimeType, GeographyType, GeometryType
- [ ] ANSI mode support (https://github.com/apache/gluten/issues/10134)
- [ ] Complete remaining unsupported Spark functions (https://github.com/apache/gluten/issues/4039)
- [ ] Deprecate Spark 3.2/3.3 and JDK 8

### Performance Optimization

- [ ] Use runtime stats to choose hash build side (https://github.com/apache/gluten/issues/11774, https://github.com/apache/gluten/pull/11775)
- [ ] Bloom filter optimization — translate might_contain as subfield filter (https://github.com/apache/gluten/issues/11771, https://github.com/apache/gluten/issues/11708, https://github.com/apache/gluten/pull/11711)
- [ ] Partial project UDF optimization (https://github.com/apache/gluten/issues/11783)
- [ ] Execution-aware dynamic join strategy selection after filter execution (https://github.com/apache/gluten/issues/11808)
- [ ] Parquet metadata check limit optimization (https://github.com/apache/gluten/issues/11782)
- [ ] Push dynamic filters to shuffle reader with per-block column statistics (https://github.com/apache/gluten/issues/11605, https://github.com/apache/gluten/pull/11769)
- [ ] Multi-core per task (https://github.com/apache/gluten/issues/7810)
- [ ] Spill enhancement — streaming window functions (https://github.com/apache/gluten/issues/3030)
- [ ] Pick split with most data prefetched (https://github.com/apache/gluten/issues/11821)
- [ ] Complex type Row-to-Columnar optimization

### Native Engine Integration

- [ ] Switch to upstream Velox official release (https://github.com/apache/gluten/issues/8782)
- [ ] Upstream useful Velox PRs not merged from Gluten community (https://github.com/apache/gluten/issues/11585)
- [ ] Bolt backend integration — ByteDance native engine with LLVM JIT (https://github.com/apache/gluten/pull/11261, https://github.com/apache/gluten/discussions/10929)
- [ ] ClickHouse backend upgrade (https://github.com/apache/gluten/pull/11734)
- [ ] Kafka read support for Velox backend (https://github.com/apache/gluten/pull/11801)

### Data Lake & File Formats

- [ ] Full Iceberg support — map write configs with Velox (https://github.com/apache/gluten/issues/11703, https://github.com/apache/gluten/pull/11776)
- [ ] Native Parquet write for complex types (Struct/Array/Map) (https://github.com/apache/gluten/pull/11788)
- [ ] Iceberg equality delete MOR table support (https://github.com/apache/gluten/pull/8056)
- [ ] Hudi MOR table support
- [ ] Delta Lake feature parity
- [ ] JSON file format support
- [ ] ORC writer support

### GPU & Hardware Acceleration

- [ ] GPU BHJ bug fix (https://github.com/apache/gluten/issues/11794)
- [x] Multi-threaded decompression in GPU shuffle reader (https://github.com/apache/gluten/issues/11779, ~https://github.com/apache/gluten/pull/11780~ https://github.com/apache/gluten/pull/12370)
- [ ] GPU code cleanup and stabilization (https://github.com/apache/gluten/pull/11824)
- [ ] ARM SVE optimization
- [ ] FPGA accelerator exploration

### Flink Integration

- [ ] Fix Flink memory leak with RocksDB state backend (https://github.com/apache/gluten/issues/11791)
- [ ] Fix Flink CI build failures (https://github.com/apache/gluten/issues/11793)
- [ ] Stabilize Flink + Velox from experimental to beta
- [ ] Nexmark benchmark support (https://github.com/apache/gluten/issues/11790)

### PySpark & Python Ecosystem

- [ ] PySpark Python UDF support
- [ ] Arrow UDF support
- [ ] Fix Python UDF/UDTF test suites on Spark 4.x (https://github.com/apache/gluten/issues/11550)

### Stability & Quality

- [ ] OOM prevention and memory stability (https://github.com/apache/gluten/issues/11747, https://github.com/apache/gluten/issues/8025)
- [ ] Full fuzzer support and result mismatch resolution (https://github.com/apache/gluten/issues/4652)
- [ ] Timezone edge case fixes (https://github.com/apache/gluten/issues/11597)
- [ ] Complex type validation in native engine (https://github.com/apache/gluten/issues/11746, https://github.com/apache/gluten/issues/11678)
- [ ] Support getting C++ stack traces via GDB from Spark UI (https://github.com/apache/gluten/issues/11677)
- [ ] collect_set ignoreNulls support (https://github.com/apache/gluten/issues/11826)

### Build, CI & Developer Experience

- [ ] VCPKG for macOS build (https://github.com/apache/gluten/pull/11563)
- [ ] IWYU tool for C++ code format checking (https://github.com/apache/gluten/pull/11287)
- [ ] Cache Maven dependencies in CI (https://github.com/apache/gluten/pull/11655)
- [ ] Fix ARM64/aarch64 build issues (https://github.com/apache/gluten/issues/11633, https://github.com/apache/gluten/issues/11639)
- [ ] Fix glog macOS build breakage (https://github.com/apache/gluten/issues/11763)
- [ ] Docker testing improvements (https://github.com/apache/gluten/issues/11501)
- [ ] Remove Arrow-CSV dependency (https://github.com/apache/gluten/issues/11591)

### Community & Governance

- [x] Complete TLP graduation remaining tasks (https://github.com/apache/gluten/issues/11713)
- [ ] Quarterly releases: 1.7 / 1.8 / 1.9 / 2.0
- [ ] GlutenCon 2026
- [ ] Expand ecosystem documentation (Kyuubi, Celeborn, Velox)
- [ ] openEuler OS support

No specific files, tests, or entry points are identified. Start by reviewing the linked issues and pull requests under the relevant 2026 roadmap section, then choose a separately scoped item with clear acceptance criteria; the roadmap is complete when its selected goals are implemented or tracked through their linked work.

Written by the indexing model from the issue text.