Eventual-Inc/Daft - CodSpeed

Daft

Blog Docs Changelog

Performance History

Latest Results

Merge branch 'main' into feature/python-io-extensions

rohankumardubey:feature/python-io-extensions

3 hours ago

Merge branch 'main' into everettVT/slurm-parquet-read-crashes

everettVT/slurm-parquet-read-crashes

13 hours ago

test: Cover null rows in try_decompress, decode, and try_decode Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

dependabot/cargo/minor-682c387934

13 hours ago

test: Move PNG encoding regression guard to a dedicated test Extract encode_image_array_as_png into daft.ai.utils, use it from the OpenAI and Google prompters, and test it directly in tests/ai/test_image_encoding.py so the guard does not depend on a provider extra. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

everettVT/pillow-13

14 hours ago

fix(benches): port parquet_read bench to tango-bench 0.9 API Same change as the minhash benches: drop the removed MeasurementSettings and the deprecated tango_main! call. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

dependabot/cargo/tango-bench-0.9.0

14 hours ago

chore(deps): exclude pyiceberg, daft-lance, typesafe-sdk, pymdown-extensions Each breaks CI on this group and needs its own PR: - pymdown-extensions 12.3: Highlight.__init__ now requires `md`; mkdocstrings still calls the old signature, so docgen/readthedocs fail. - pyiceberg 0.12.0: from_bytes now decodes promoted types (int->long, float->double); tests/io/iceberg/test_iceberg_stats.py expects None. - daft-lance 0.6.0: breaks tests/io/lancedb (writes, factory, scalar index). - typesafe-sdk 0.7.5: moved to httpx2; test_jev_packs_rows_per_request's fake server disconnects. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

dependabot/pip/all-b744bbb960

14 hours ago

chore(deps): bump arrow and parquet suite to 60.0.0 arrow-avro 60 pulls in arrow-schema 60, so the rest of the arrow-rs crates, parquet, and serde_arrow (0.15.1, arrow-60 feature) move with it to keep a single arrow version in the graph. API changes handled: - OffsetBuffer::first/last now return the value instead of Option<&T>. - Field::with_metadata takes impl Into<Metadata>; collect into HashMap. - ParquetMetaData::offset_index replaced by page_index_for_row_group. - Parquet array readers take a padding_threshold/parent_threshold. The Daft field reader now threads it the same way parquet's builder does: None at the root, Some(list def_level) for children of lists and maps, passed through unchanged for struct children. - get_column_writers removed; use ArrowRowGroupWriterFactory. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

dependabot/cargo/arrow-avro-60.0.0

14 hours ago

fix: Skip null rows in binary encode/decode kernels flate2 1.1.10 rejects an empty zlib stream with "incomplete deflate stream". The kernels passed the empty slice behind each null row to the codec, so decompress("zlib") on a column with nulls now errors. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

dependabot/cargo/minor-682c387934

14 hours ago

Latest Branches

0%

feat(io): add async Python IO extensions#7564

3 hours ago

a8be7c1

rohankumardubey:feature/python-io-extensions

0%

fix(io): prevent deadlocks in parallel Python scans#7586

13 hours ago

d2b5a73

everettVT/slurm-parquet-read-crashes

+10%

chore(deps): bump the minor group with 48 updates#7637

14 hours ago

f765de6

dependabot/cargo/minor-682c387934

© 2026 CodSpeed Technology

Home Terms Privacy Docs