Eventual-Inc/Daft - CodSpeed

Daft

Blog Docs Changelog

Performance History

Latest Results

Merge branch 'main' into feat/symbolize-string-groupby-only

BABTUNA:feat/symbolize-string-groupby-only

7 hours ago

test(pycapsule): use PyArrow 15.0.0-compatible APIs CI runs unit tests against pyarrow==15.0.0 (matrix lower bound). Earlier tests used APIs only available in newer pyarrow: - pa.chunked_array(obj) as PyCapsule consumer (added post-15) - pa.record_batch(dict, schema=...) dict-form constructor Replaced with pa.RecordBatch.from_pydict / from_arrays and exercised requested_schema cast via pa.RecordBatchReader.from_stream(obj, schema=) which works across all supported pyarrow versions.

aaron-ang:arrow-pycapsule

11 hours ago

address comments

YuangGao:fix/iceberg-respect-target-file-size-3823

12 hours ago

feat(iceberg): honor table-level write.target-file-size-bytes property

YuangGao:fix/iceberg-respect-target-file-size-3823

15 hours ago

re-enable Iceberg map-type round-trip test

YuangGao:test/unskip-iceberg-roundtrip-2459

16 hours ago

feat: add conv function for PySpark parity

YuangGao:feat/add-conv-function

18 hours ago

feat: add Python UDAF support (@daft.udaf) (#6790) ## Changes Made Adds `@daft.udaf` decorator that lets users define custom aggregate functions in Python with a three-stage `aggregate`/`combine`/`finalize` pipeline. Supports single-state, multi-state, parameterized, and multi-input UDAFs, working with both `groupby().agg()` and global `agg()`. Rust side adds `PyAggFn` implementing the `AggFn` trait, bridging to Python via GIL. Python side adds the decorator API (`udaf.py`) and execution bridge (`agg_execution.py`). Includes 19 tests and user docs. --- During Python UDAF development, two supporting changes were made to daft-core. - `AggFn::name()` return type relaxed from `&'static str` to `&str`, removing the `Box::leak` workaround needed by runtime-named implementations. - Introduced `AggExpr::AggFnCombine`, a internal planner node that enables proper distributed two-phase aggregation staging, also replacing the ad-hoc `try_eager_combine` mechanism. ## Related Issues #6698

main

1 day ago

refactor(distributed): decouple shuffle cleanup from Ray (#6809) ## Changes Made Flight shuffle writes data to local disk on each worker, and cleanup currently hardcodes a `crate::python::ray::clear_shuffle_dirs_on_all_nodes` call in `runner.rs` that uses `ray.remote` to fan out deletion. This PR decouples that by moving cleanup into a `cleanup_shuffle_dirs` method on the `WorkerManager` trait, so the runner no longer knows which communication channel is used. `RayWorkerManager` delegates to the same bridge function; future non-Ray backends (e.g. K8s) can implement their own. ## Related Issues This lays groundwork for Daft Native Kubernetes Support (#6639).

main

1 day ago

Latest Branches

-1%

feat(inline-agg): symbolize string group keys in multi-column grouped aggregation#6748

8 hours ago

971192b

BABTUNA:feat/symbolize-string-groupby-only

-1%

feat: Arrow PyCapsule Interface#6745

20 days ago

ec4b272

aaron-ang:arrow-pycapsule

0%

feat(iceberg): honor table-level write.target-file-size-bytes property#6912

13 hours ago

5c477ed

YuangGao:fix/iceberg-respect-target-file-size-3823

© 2026 CodSpeed Technology

Home Terms Privacy Docs