v0.7.0

What's Changed 🚀

💥 Breaking Changes

• chore!: remove spark connect @universalmind303 (#5743)

✨ Features

• feat(lance): row-level schema evolution support @Jay-ju (#5749)
• feat: Improve model api typing @colin-ho (#5809)
• feat(deltalake): allow users to ignore deletion vectors on read @kevinzwang (#5758)
• feat: Capture UDF argument names and values in structured logging. @rohitkulshreshtha (#5771)
• feat: support split and merge jsonl/ndjson files @caican00 (#5695)
• feat(tools): add markdown-to-notebook converter for documentation @ykdojo (#5691)
• feat: support label_selector specification in ray actor/task creation @Jay-ju (#5042)
• feat: support native csv writer @stayrascal (#5706)
• feat: Update optional dependencies for daft[postgres] @desmondcheongzx (#5586)
• feat(lance): add distributed compaction for Lance @huleilei (#5699)
• feat: Use JSON Serialization for Plans in Subscribers @srilman (#5709)
• feat: extended hash function to take and hash multiple inputs @rahulkodali (#5692)
• feat: Add Apache Gravitino catalog in catalog module @shaofengshi (#5694)
• feat: Better errors when lazy imports fail @samstokes (#5753)
• feat: Added structured logging for UDF errors. @rohitkulshreshtha (#5688)
• feat: Add statistics info to ScanTask when reading Lance dataset @plotor (#5727)
• feat: dynamic batching per operator @universalmind303 (#5676)
• feat: Allow users to disable the suffix range request @TheR1sing3un (#5188)
• feat: Streaming sample by size @colin-ho (#5663)
• feat: Allow dashboard to show query canceled/failed/dead information when query exited abnormally @VOID001 (#5576)
• feat: Add Google AI provider with prompt @everettVT (#5640)
• feat: Add pow expression @kliwongan (#5237)
• feat: No truncate in .collect preview @colin-ho (#5632)
• feat: sample api supports precise sampling by size params @caican00 (#5600)
• feat: audio file subtype @universalmind303 (#5602)
• feat(tos): enhance the retry logic to aware response @stayrascal (#5569)
• feat: emit selectivity metric to OTel in swordfish filter op @samstokes (#5584)
• feat: Adding otel logger collector for collecting UDF errors. @rohitkulshreshtha (#5624)

🐛 Bug Fixes

• fix: handle Windows paths and query params in local_path_from_uri @ykdojo (#5819)
• fix: use pytest.importorskip for lance in test_limit_offset @ykdojo (#5818)
• fix: support skip empty json/jsonl files @caican00 (#5660)
• fix: minor doc fix @yuchaoran2011 (#5814)
• fix: CountRows with Limit returns unexpected result when reading Lance dataset @plotor (#5550)
• fix: Check for missing dependencies in OpenAI provider @everettVT (#5747)
• fix: fix btree index invalid issue when reading lance for point lookup @caican00 (#5673)
• fix: Combine deltalake with unity extra @everettVT (#5785)
• fix: enhance unit tests @caican00 (#5787)
• fix: patch CVE-2025-66478 update next dependencies to 16.0.7 @everettVT (#5786)
• fix: use single consolidated progress bar in Jupyter notebooks @ykdojo (#5774)
• fix: CuPy → NumPy needs explicit conversion @Jay-ju (#5680)
• fix: Fix Pydantic cloudpickle serialization in Google Colab @ykdojo (#5705)
• fix: update AI integration tests for new Subscriber interface @ykdojo (#5763)
• fix(optimizer): Prevent limits from being pushed below explodes in non-top-level projections @desmondcheongzx (#5292)
• fix(io): load all splits in read_huggingface fallback path @ykdojo (#5757)
• fix(test): use read_huggingface instead of read_parquet for HF test @ykdojo (#5755)
• fix: add disk cleanup to nightly integration-test-io job @ykdojo (#5711)
• fix: Postgres overwrite table should enable RLS and set up pgvector automatically @desmondcheongzx (#5657)
• fix: make it easier to enable different logging levels @Abyss-lord (#5661)
• fix: Dashboard logo animation. @j3nkii (#5672)
• fix(ci): add disk cleanup to integration-test-ai job @ykdojo (#5733)
• fix: update hypothesis test to use new expression API @ykdojo (#5723)
• fix: Fix type annotation check on Python 3.14 @srilman (#5721)
• fix: add fallback mechanism for HuggingFace datasets without parquet files @ykdojo (#5650)
• fix: Import or skip lance @colin-ho (#5662)
• fix: Add missing trailing slashes to S3-compatible endpoint urls @desmondcheongzx (#5575)
• fix: Add outer try-finally block in executor generator @colin-ho (#5633)
• fix: test_explain @universalmind303 (#5656)
• fix: Unify the naming and type of URI parameter for Lance-related APIs @plotor (#5634)
• fix: Fix blocked and oom issues for scan lance @caican00 (#5592)
• fix: Executing explain will panic when ScanTask is empty @plotor (#5582)
• fix: Embed text dropping texts @colin-ho (#5641)
• fix: limit(n) return n rows directly @caican00 (#5597)
• fix: Upgrade to deltalake 1.2.1 @colin-ho (#5580)
• fix: add disk cleanup to integration-test-io-credentialed job @ykdojo (#5610)
• fix: add disk cleanup to doctests job @ykdojo (#5609)
• fix: Hashable identifier @colin-ho (#5598)

🚀 Performance

• perf: Lazy udf worker @colin-ho (#5542)
• perf: Use growable for build side @colin-ho (#5613)
• perf: optimize setting lance schema @Jay-ju (#5704)

♻️ Refactor

• refactor(arrow2): values_iter removals for primitive array @universalmind303 (#5802)
• refactor(arrow2): remove arrow2 from daft-functions-binary @universalmind303 (#5799)
• refactor(arrow2): remove deprecated usages from daft-functions-utf8 @universalmind303 (#5800)
• refactor(arrow2): remove deprecated methods from daft-functions-uri crate @universalmind303 (#5798)
• refactor(arrow2): remove arrow2 from daft-functions-tokenize @universalmind303 (#5797)
• refactor(arrow2): rename and deprecate to_arrow and as_arrow functions @universalmind303 (#5796)
• refactor: write empty dataframe to parquet/json files via native IO @stayrascal (#5682)
• refactor(arrow-rs): Move temporal conversions from arrow2 to arrow-rs @srilman (#5782)
• refactor(arrow-rs): Remove arrow2 Index generics usages @srilman (#5761)
• refactor(arrow2): rename and deprecate .to_arrow methods @universalmind303 (#5789)
• refactor(arrow): use arrow for ffi instead of arrow2 @universalmind303 (#5775)
• refactor(arrow): remove daft-arrow from daft-sql & daft-context crates @universalmind303 (#5773)
• refactor(arrow-rs): Move all validity daft_arrow::bitmap::Bitmaps to daft_arrow::buffer::NullBuffer @srilman (#5750)
• refactor: abstract MultipartWriter to write data to object store @stayrascal (#5702)
• refactor: Remove Unloaded MicroPartitions @srilman (#5710)
• refactor(arrow-rs): Add daft-arrow middleman crate for Rust & Arrow usage @srilman (#5730)

📖 Documentation

• docs: Update slack invite @everettVT (#5813)
• docs: add logging settings @Jay-ju (#5671)
• docs: fix broken Bodo benchmark link @ykdojo (#5762)
• docs: add voice-analytics-example and update index @everettVT (#5737)
• docs: fix broken Lance documentation link @ykdojo (#5724)
• docs: remove redundant About Daft section from README @ykdojo (#5689)
• docs: remove redundant Table of Contents from README @ykdojo (#5684)
• docs: add Daft Cloud mentions to distributed execution docs @ykdojo (#5686)
• docs: fix quickstart connector links formatting @ykdojo (#5687)
• docs: update README to reflect AI/multimodal positioning @ykdojo (#5677)
• docs: Improve mkdocstrings template for Python examples rendering @ykdojo (#5642)
• docs: changed dev url to a live link to prevent 404 @j3nkii (#5669)
• docs: add Python version requirement to README @ykdojo (#5655)
• docs: update index overview page @ykdojo (#5627)
• docs: remove Python tabs from quickstart @ykdojo (#5626)
• docs: update contributor policy, add contributing section, remove old… @madvart (#5251)
• docs: add tip to find your dylib @universalmind303 (#5625)
• docs: add data persistence section to quickstart @ykdojo (#5607)
• docs: revamp quickstart with Amazon product dataset example @ykdojo (#5585)

✅ Tests

• test: fix flaky OpenAI test by using pattern constraint for hex color format @ykdojo (#5808)
• test(io): remove flaky test_read_huggingface_http_urls test @ykdojo (#5795)

👷 CI

• ci: increase unit-test timeout to 75 minutes for macOS @ykdojo (#5731)
• ci: exclude Kaggle from link checker @ykdojo (#5725)

🔧 Maintenance

• chore(deps): bump the minor group across 1 directory with 45 updates @dependabot[bot] (#5734)
• chore: Provide query end state to RuntimeStatsManager on query end @colin-ho (#5791)
• chore: update uvlock to remove tensorflow @kevinzwang (#5780)
• chore: Codeowners @colin-ho (#5502)
• chore: Don't install pytorch in iceberg test docker compose @colin-ho (#5781)
• chore: remove arrow dep from common-image @universalmind303 (#5772)
• chore: Pin dependencies @colin-ho (#5667)
• chore!: remove spark connect @universalmind303 (#5743)
• chore: remove ir and proto crates @universalmind303 (#5742)
• chore(deps): bump actions/checkout from 5 to 6 in the all group @dependabot[bot] (#5713)
• chore(deps): bump ctor from 0.5.0 to 0.6.1 @dependabot[bot] (#5717)
• chore: Cleanup additional Ray runner artifacts @srilman (#5714)
• chore: Remove the old Ray Runner @srilman (#5375)
• chore: Remove expression namespaces @colin-ho (#5619)
• chore: Remove runner from context @colin-ho (#5628)
• chore: add deprecation to daft.udf @kevinzwang (#5665)
• chore(deps): bump the all group with 13 updates @dependabot[bot] (#5480)
• chore: update bug report template @universalmind303 (#5652)
• chore: remove checklist from pr template @universalmind303 (#5653)
• chore: Remove deprecated agg methods and series split @colin-ho (#5630)
• chore: remove broken docpublish job from build-docs workflow @ykdojo (#5631)
• chore: Hint users to use .collect when printing empty dataframes @colin-ho (#5616)
• chore: reduce binary size by feature flagging derive(Debug) @universalmind303 (#5622)
• chore: deprecate llm_generate function @universalmind303 (#5603)
• chore: reduce sdist size @universalmind303 (#5606)

⬆️ Dependencies

4 changes

• chore(deps): bump the minor group across 1 directory with 45 updates @dependabot[bot] (#5734)
• chore(deps): bump actions/checkout from 5 to 6 in the all group @dependabot[bot] (#5713)
• chore(deps): bump ctor from 0.5.0 to 0.6.1 @dependabot[bot] (#5717)
• chore(deps): bump the all group with 13 updates @dependabot[bot] (#5480)

Full Changelog: v0.6.14...v0.7.0