Release

1.38.0 (2025-09-04)

Snowpark Python API Updates

New Features

• Added support for the following AI-powered functions in functions.py:
  • ai_extract
  • ai_parse_document
  • ai_transcribe
• Added time travel support for querying historical data:
  • Session.table() now supports time travel parameters: time_travel_mode, statement, offset, timestamp, timestamp_type, and stream.
  • DataFrameReader.table() supports the same time travel parameters as direct arguments.
  • DataFrameReader supports time travel via option chaining (e.g., session.read.option("time_travel_mode", "at").option("offset", -60).table("my_table")).
• Added support for specifying the following parameters to DataFrameWriter.copy_into_location for validation and writing data to external locations:
  • validation_mode
  • storage_integration
  • credentials
  • encryption
• Added support for Session.directory and Session.read.directory to retrieve the list of all files on a stage with metadata.
• Added support for DataFrameReader.jdbc(PrPr) that allows ingesting external data source with jdbc driver.
• Added support for FileOperation.copy_files to copy files from a source location to an output stage.
• Added support for the following scalar functions in functions.py:
  • all_user_names
  • bitand
  • bitand_agg
  • bitor
  • bitor_agg
  • bitxor
  • bitxor_agg
  • current_account_name
  • current_client
  • current_ip_address
  • current_role_type
  • current_organization_name
  • current_organization_user
  • current_secondary_roles
  • current_transaction
  • getbit

Bug Fixes

• Fixed the repr of TimestampType to match the actual subtype it represents.
• Fixed a bug in DataFrameReader.dbapi that udtf ingestion does not work in stored procedure.
• Fixed a bug in schema inference that caused incorrect stage prefixes to be used.

Improvements

• Enhanced error handling in DataFrameReader.dbapi thread-based ingestion to prevent unnecessary operations, which improves resource efficiency.
• Bumped cloudpickle dependency to also support cloudpickle==3.1.1 in addition to previous versions.
• Improved DataFrameReader.dbapi (PuPr) ingestion performance for PostgreSQL and MySQL by using server side cursor to fetch data.

Snowpark pandas API Updates

New Features

• Completed support for pd.read_snowflake(), pd.to_iceberg(),
  pd.to_pandas(), pd.to_snowpark(), pd.to_snowflake(),
  DataFrame.to_iceberg(), DataFrame.to_pandas(), DataFrame.to_snowpark(),
  DataFrame.to_snowflake(), Series.to_iceberg(), Series.to_pandas(),
  Series.to_snowpark(), and Series.to_snowflake() on the "Pandas" and "Ray"
  backends. Previously, only some of these functions and methods were supported
  on the Pandas backend.
• Added support for Index.get_level_values().

Improvements

• Set the default transfer limit in hybrid execution for data leaving Snowflake to 100k, which can be overridden with the SnowflakePandasTransferThreshold environment variable. This configuration is appropriate for scenarios with two available engines, "Pandas" and "Snowflake" on relational workloads.
• Improve import error message by adding --upgrade to pip install "snowflake-snowpark-python[modin]" in the error message.
• Reduce the telemetry messages from the modin client by pre-aggregating into 5 second windows and only keeping a narrow band of metrics which are useful for tracking hybrid execution and native pandas performance.
• Set the initial row count only when hybrid execution is enabled. This reduces the number of queries issued for many workloads.
• Add a new test parameter for integration tests to enable hybrid execution.

Bug Fixes

• Raised NotImplementedError instead of AttributeError on attempting to call
  Snowflake extension functions/methods to_dynamic_table(), cache_result(),
  to_view(), create_or_replace_dynamic_table(), and
  create_or_replace_view() on dataframes or series using the pandas or ray
  backends.