v4.1.0

Breaking changes

• The asyncio SQLAlchemy adapter classes in pyathena.aio.sqlalchemy.base are renamed, with no aliases for the old names: AsyncAdapt_pyathena_cursor → AthenaAioAdaptedCursor, AsyncAdapt_pyathena_connection → AthenaAioAdaptedConnection, AsyncAdapt_pyathena_dbapi → AthenaAioAdaptedDBAPI (#1151). Code that imports the old names must use the new ones. Engines created with the awsathena+aio* URLs are unaffected.

Behavior changes

• PandasCursor parses CSV results with json columns with pandas' default low_memory=True and converts the json columns after parsing (#1141). pandas now applies these options to the JSON text instead of the converted value:
  • parse_dates naming a json column no longer parses a JSON string such as "2020-01-01", which keeps its quotes; cast in SQL instead.
  • With engine="python" and NA options such as keep_default_na=True, a JSON string that matches an NA token, such as "NA", stays a string; SQL NULL is still None.

Benchmark

The benchmark of this release is in benchmarks/history/2026-10-13-437276e1.md and benchmarks/history/2026-10-13-437276e1-filesystem.md (#1157). Highlights:

• PandasCursor reads CSV results faster than in 4.0.0, for example 10 million flat rows in 74 seconds instead of 110 (#1134, #1144).
• Chunked UNLOAD reads in PandasCursor and PolarsCursor stay at about 600 MiB for 10 million rows (#1135, #1143).

What's Changed

• Run benchmark fleets on Spot hosts that retire when done by @laughingman7743 in #1129
• Record the PyAthena 4.0.0 large-data benchmark run by @laughingman7743 in #1130
• Prefetch CSV result files while pandas parses them by @laughingman7743 in #1134
• Read chunked UNLOAD results in bounded memory by @laughingman7743 in #1135
• Close asyncio cursor result sets in a thread by @laughingman7743 in #1142
• Benchmark read_sql_query and trim the next benchmark selection by @laughingman7743 in #1146
• Start chunked UNLOAD reads sooner and pre-buffer large Parquet files one row group at a time by @laughingman7743 in #1143
• Convert json CSV columns after parsing instead of forcing low_memory=False by @laughingman7743 in #1141
• Prefetch CSV results in S3FSCursor and chunked reads, read Arrow CSV in parallel by @laughingman7743 in #1144
• Resume benchmark fleet jobs after Spot interruptions by @laughingman7743 in #1145
• Rename the asyncio SQLAlchemy adapter classes to CapWords and enforce N801 by @laughingman7743 in #1151
• Document the fsspec cache types that cache_type accepts by @laughingman7743 in #1150
• Add a filesystem suite comparing S3FileSystem and AioS3FileSystem by @laughingman7743 in #1152
• Allow c8gn.2xlarge benchmark hosts for the filesystem suite by @laughingman7743 in #1154
• Generalize the parallel read-ahead and read small UNLOAD files in parallel ranges by @laughingman7743 in #1153

Full Changelog: v4.0.0...v4.1.0