2025)¶

🎉 Highlights¶

In this release, we're excited to support reading and writing GCS with FileSystemCatalog in Bodo and upgrading MPI4Py to version 4.

✨ New Features¶

Support reading/writing to GCS with FileSystemCatalog.
Add API get_gpu_ranks to return list of ranks pinned to GPU.

⚙️ Dependency Upgrades¶

Upgrade to MPI4Py 4.

🐛 Bug Fixes¶

Fix caching in BodoSQLContext.sql
Fix BodoDataFrame.to_csv()/to_json() string output in spawn mode.
Fix error in writing parquet files with Snowflake.

2025.4.1¶

Mini release adding experimental support for Iceberg time travel in Python reader

2025.4.2¶

🎉 Highlights¶

In this release, we are exited to add hooks for Bodo to accelerate User Defined Functions (UDFs) in Pandas. We also fixed several bugs related to our Pandas support in the compiler.

✨ New Features¶

Added hooks for accelerating Pandas UDFs with Bodo. Starting in Pandas 3.0, you will be able to use Bodo to automatically jit compile your UDFs and execute them over a DataFrame in parallel by passing engine=bodo.jit to DataFrame.apply().

🐛 Bug Fixes¶

Use Arrow for boxing/unboxing Datetime arrays.
Support passing an empty tuple as an argument to a function in Spawn-mode.
Support passing non-constant UDF args in DataFrame.apply().
Support handling Pandas classes imported directly in the compiler.