Daft 0.0.18 Release Notes#

The Daft 0.0.18 adds bugfixes and new functionality - highlights:

  • Ray Datasets integration - Daft Dataframes can now feed into ML training on Ray easily using the .to_ray_dataset() method

  • Big performance improvements from fixes in join algorithm

  • .apply now infers the return type from the function’s type annotation if available

New Features#

Ray Datasets Integration#

Daft Dataframes can now be converted easily into a Ray Datasets, which makes it really easy to go from preprocessing/analyzing your data in Daft to ML training using the Ray ecosystem of tools, all on the same Ray cluster.

See: #316

String Concatenation#

Expression.str.concat now allows for concatenation of strings in a Daft Dataframe.

See: #313

Enhancements#

  • Infer return type of function from type annotations in .apply #311

  • Import lit from daft top-level module #315

  • Fix placement group for benchmarking script #312

  • Run 10-min quickstart in documentation #305

  • Cleanup multicolumn joins in tpch #304

  • Change min(left, right) to max(left, right) for join partitioning #303

Bug Fixes#

  • Fix .url.download() for s3 URLs #310

Closed Issues#

  • .apply should infer the return_type from type annotations if available #309

  • Add .str.concat expression operator #306