What's Changed

• fix: Allow pdb.agg() with solve_mvcc argument to work with all search operators by @paradedb-bot in #4534
• fix: Allow joinscan pushdown for DISTINCT where fast fields have typmod casts by @paradedb-bot in #4580
• fix(docs): correct 0.22.2 changelog release link by @paradedb-bot in #4581
• fix: Aggregate custom scan handles NULL bool values in GROUP BY by @paradedb-bot in #4584
• chore: Upgrade to 0.22.4 by @paradedb-bot in #4587

Full Changelog: v0.22.3...v0.22.4

What's Changed

• feat: Snapshot all indexes via ParallelScanState for consistent parallel JoinScan DocAddresses by @paradedb-bot in #4443
• refactor(benchmarks): scope query connections in helper by @paradedb-bot in #4444
• fix: rebuild custom scan heap state on rescan by @paradedb-bot in #4461
• fix: bypass min_rows_per_worker for sorted/TopK scans by @paradedb-bot in #4462
• chore: Add a sort-merge-join variant of semi_join_filter. by @paradedb-bot in #4467
• chore: Tighten and test min/max bounds for Arrow datetime tests by @paradedb-bot in #4465
• feat(topn): Support ORDER BY indexed expressions in TopN scan path (#4390) by @RuchirRaj in #4466
• ci: auto-fix nix cargoHash when Cargo.lock changes by @paradedb-bot in #4470
• chore: Prepare 0.22.3. (#4469) by @mdashti in #4471

Full Changelog: v0.22.2...v0.22.3

What's Changed

• fix(joinscan): align memory pool allocation with postgres work_mem semantics by @paradedb-bot in #4417
• fix: Expose pdb.agg(.., solve_mvcc=bool) argument in all cases by @paradedb-bot in #4419
• refactor(scan): unify threshold pruning under DynamicFilterPhysicalExpr by @paradedb-bot in #4420
• fix: Bound Tantivy datetime conversions in proptests. by @paradedb-bot in #4423
• fix: Track the rti of the score-providing relation for an ORDER BY by @paradedb-bot in #4429
• fix: Allow the joinscan to run for a semi/anti join, even when partitioning is not optimal. by @paradedb-bot in #4432
• refactor(benchmarks): Port benchmark runner from psql to sqlx by @paradedb-bot in #4438
• chore: Audit pull request events for actions. by @paradedb-bot in #4440
• chore: Prepare 0.22.2. by @paradedb-bot in #4441

Full Changelog: v0.22.1...v0.22.2

What's Changed

• fix: Expose pdb.agg(.., solve_mvcc=bool) argument in all cases by @stuhood in #4421
• chore: Audit pull request events for actions. by @paradedb-bot in #4439
• chore: Prepare 0.21.16. by @paradedb-bot in #4436

Full Changelog: v0.21.15...v0.21.16

What's Changed

• ci: Add other ORM repos to release by @paradedb-bot in #4385
• fix: Handle relabel nodes in more places in join scan planning by @paradedb-bot in #4388
• chore: Feedback from logical late materialization change by @paradedb-bot in #4387
• fix(docker): add setuptools for Python 3.13 compatibility with barman by @paradedb-bot in #4399
• fix: Allow aggregate scan planning warnings to be disabled. by @paradedb-bot in #4402
• fix: Fix casts in aggregate scan target lists. by @paradedb-bot in #4404
• chore: Prepare 0.22.1. by @paradedb-bot in #4408

Full Changelog: v0.22.0...v0.22.1

What's Changed

• fix: Handle wrapped aggregates in aggregate pushdown by @mithuncy in #3804
• chore: Revert "chore: Do not benchmark release branches on push." by @ankitml in #3860
• feat: allow multiple language stopwords filters per field by @mdashti in #3852
• ci: Fix macos deploys on PG18+ by @philippemnoel in #3864
• ci: Adding Antithesis Bug report action by @philippemnoel in #3868
• chore: Default to PG18 by @philippemnoel in #3869
• docs: Update roadmap by @philippemnoel in #3870
• feat: use native rust icu tokenizer by @maazzzzz in #3792
• fix: Extract heap-filter quals for top level function expressions by @stuhood in #3873
• chore: Swap CI to Ubuntu 2404 by @philippemnoel in #3874
• feat: Add composite type support for BM25 indexes (>32 fields) by @mithuncy in #3776
• fix: error if query needs positions but field doesn't have them by @rebasedming in #3877
• chore: Stabilize Cargo.lock. by @stuhood in #3884
• chore: Add a check that the lockfile is stable in CI. by @stuhood in #3886
• chore: Rename pdbscan to basescan by @stuhood in #3885
• fix: race condition in parallel index scans by @mdashti in #3872
• fix: crash in parallel hash joins with generic plans by @mdashti in #3881
• docs: Update Postgres version (18) in installation notes by @maina-m in #3896
• chore: upgrade to 0.21.1 by @rebasedming in #3897
• ci: Run Antithesis for 8 hours by @philippemnoel in #3902
• chore: Remove leftovers ICU feature by @philippemnoel in #3903
• chore: Incorporate Tantivy with index sorting by @stuhood in #3906
• fix: SchemaBot migration validation by @mdashti in #3908
• docs: Update intro page w/ new messaging by @rebasedming in #3905
• fix: Enable columnar-zstd-compression by @rebasedming in #3912
• ci: Move more workflows to Ubicloud by @philippemnoel in #3913
• chore: Upgrade to 0.21.2 by @rebasedming in #3915
• fix: avoid double-panic crashes during query execution by @mdashti in #3895
• fix: SchemaBot migration validation for order-independent comparison by @mdashti in #3917
• chore: Jan 12th Tantivy rebase. by @stuhood in #3924
• fix: pdb.agg works with JSON paths by @rebasedming in #3925
• feat: Push down = ANY() over text arrays by @rebasedming in #3923
• feat: Add TopN Scan Validation by @vrn21 in #3816
• fix: Parameters from InitPlan nodes are uncorrelated by @stuhood in #3921
• feat: add AI context menu for docs by @ankitml in #3937
• feat: Legacy docs removal, banner updated to github url by @ankitml in #3938
• fix: use advisory locks to coordinate background merging by @rebasedming in #3929
• feat: ngram with positions by @rebasedming in #3882
• chore: update contextual menu order and add share on X by @ankitml in #3942
• fix: Aliased JSON fields index correctly by @rebasedming in #3947
• fix: missing ICU tokenizer in upgrade scripts by @rebasedming in #3949
• test: add regression test for text field aggregation bug by @mithuncy in #3952
• chore: Upgrade to 0.21.3 by @rebasedming in #3959
• chore: Add skeleton of join scan by @stuhood in #3889
• chore: retire PostgreSQL 14 support by @mdashti in #3964
• fix: upgrade script for pdb.icu by @rebasedming in #3961
• docs: Fix leftover mentions of PG14 and missing mentions of PG18 by @philippemnoel in #3969
• fix: pdb permission grants by @rebasedming in #3968
• chore: Upgrade to 0.21.4 by @rebasedming in #3973
• docs: add changelogs for 0.20.7 and 0.20.8 by @rebasedming in #3977
• chore: Ankitml/skills mcp docs by @ankitml in #3975
• fix: Always use virtual tuples for fast fields scan by @stuhood in #3957
• perf: Unify visibility checks across aggregates and base scan by @stuhood in #3963
• chore: bring back legacy docs and navbar by @ankitml in #3980
• fix: legacy docs visibility by @ankitml in #3983
• feat: v2 API supports tokenizing uuid by @rebasedming in #3928
• feat: Add support for JSONB exists (?) operator pushdown by @jamessewell in #3982
• feat: custom join scan planning (with simple execution) by @mdashti in #3930
• fix: Fix hang in rare Tantivy intersection case by @stuhood in #3989
• chore: context7 ownership by @ankitml in #3993
• feat: added amcheck-style index verification for BM25 indexes by @mdashti in #3907
• chore: Document on how to cross search on a microservices architecture by @ankitml in #3995
• feat: add cooperative vacuum cancellation for background merge workers by @mithuncy in #3970
• fix: round-trip problem with generic_typmod_out by @rebasedming in #3994
• feat: default literal normalized to fast by @rebasedming in #3979
• chore: Expand join benchmarks by @stuhood in #4002
• chore: Add a DataFusion fast fields scan by @stuhood in #3981
• feat: Add sort_by option to CREATE INDEX syntax by @mithuncy in #3988
• fix: fetch full git graph for ancestry caching by @ankitml in #4006
• chore: Swap join scan execution to an unoptimized DataFusion plan by @stuhood in #4003
• feat: Allow Top N + aggregates over literal normalized fields by @rebasedming in #4007
• ci: Better cherry-pick action by @philippemnoel in #4014
• chore: Upgrade to 0.21.5 by @rebasedming in #4017
• ci: Use pull title when cherry-picking by @philippemnoel in #4020
• chore: Add support for configurable error ignores in stressgres. by @stuhood in #4021
• fix: redundant heap filters for partial index predicates by @mdashti in #4027
• fix: ORDER BY lower() gets Top N if literal normalized by @rebasedming in #4029
• fix: overflow for large document counts and doc ids by @rebasedming in #4036
• chore: Upgrade to 0.20.10 by @rebasedming in #4042
• chore: update pre-commit hooks to latest versions by @philippemnoel in #4047
• chore: Upgrade to 0.21.6 by @rebasedming in #4044
• chore(deps): bump docker/build-push-action from 5 to 6 in the github-actions-dependencies group by @dependabot[bot] in #4051
• chore: moved parallel_build_large test from regression tests to rust tests by @mdashti in #4041
• chore: Use DataFusion's optimizer in the join scan by @stuhood in #4009
• chore: bump new github action for benchmark by @ankitml in #4055
• fix: Unify TopN detection across hooks by @stuhood in #4026
• chore: Deflake the join visibility test. by @stuhood in #4056
• chore: Revert: bump new github action for benchmark" by @ankitml in #4067
• feat: Allow any text/JSON field to be columnar (ie fast) in v2 API by @rebasedming in #4035
• fix: `pdb.score...

What's Changed

• fix: RLS policies that get planned as SubPlans by @paradedb-bot in #4358
• fix: prevent ReadBuffer errors from stale ctids after VACUUM truncation by @paradedb-bot in #4353
• chore: Upgrade to 0.21.15 by @rebasedming in #4366

Full Changelog: v0.21.14...v0.21.15

What's Changed

• ci: Restore --runs 10 to benchmark queries by @paradedb-bot in #4302
• chore: Adjust code ownership by @paradedb-bot in #4317
• chore: use upstream github-action-benchmark instead of paradedb fork by @paradedb-bot in #4330
• fix: Pushdown errors in faceted queries with paradedb.enable_filter_pushdown disabled by @paradedb-bot in #4335
• chore: Upgrade to 0.21.14 by @paradedb-bot in #4341

Full Changelog: v0.21.13...v0.21.14

What's Changed

• perf: 45X speedup on term bucket queries with high bucket counts by @paradedb-bot in #4298
• chore: Upgrade to 0.21.13 by @paradedb-bot in #4300

Full Changelog: v0.21.12...v0.21.13

What's Changed

• ci: Dispatch ORM release job after Docker image is published by @paradedb-bot in #4279
• fix: Eliminate clones in aggregate custom scan, 100x+ speedup in certain cases by @paradedb-bot in #4287
• chore: Upgrade to 0.21.12 by @paradedb-bot in #4290

Full Changelog: v0.21.11...v0.21.12