Release

1.28.0 (2025-02-20)

Snowpark Python API Updates

New Features

• Added support for the following functions in functions.py
  • normal
  • randn
• Added support for allow_missing_columns parameter to Dataframe.union_by_name and Dataframe.union_all_by_name.

Improvements

• Improved the random object name generation to avoid collisions.
• Improved query generation for Dataframe.distinct to generate SELECT DISTINCT instead of SELECT with GROUP BY all columns. To disable this feature, set session.conf.set("use_simplified_query_generation", False).

Deprecations

• Deprecated Snowpark Python function snowflake_cortex_summarize. Users can install snowflake-ml-python and use the snowflake.cortex.summarize function instead.
• Deprecated Snowpark Python function snowflake_cortex_sentiment. Users can install snowflake-ml-python and use the snowflake.cortex.sentiment function instead.

Bug Fixes

• Fixed a bug where session-level query tag was overwritten by a stacktrace for dataframes that generate multiple queries. Now, the query tag will only be set to the stacktrace if session.conf.set("collect_stacktrace_in_query_tag", True).
• Fixed a bug in Session._write_pandas where it was erroneously passing use_logical_type parameter to Session._write_modin_pandas_helper when writing a Snowpark pandas object.
• Fixed a bug in options sql generation that could cause multiple values to be formatted incorrectly.
• Fixed a bug in Session.catalog where empty strings for database or schema were not handled correctly and were generating erroneous sql statements.

Experimental Features

• Added support for writing pyarrow Tables to Snowflake tables.

Snowpark pandas API Updates

New Features

• Added support for applying Snowflake Cortex functions Summarize and Sentiment.
• Added support for list values in Series.str.get.

Bug Fixes

• Fixed a bug in apply where kwargs were not being correctly passed into the applied function.

Snowpark Local Testing Updates

New Features

• Added support for the following functions
  • hour
  • minute
• Added support for NULL_IF parameter to csv reader.
• Added support for date_format, datetime_format, and timestamp_format options when loading csvs.

Bug Fixes

• Fixed a bug in Dataframe.join that caused columns to have incorrect typing.
• Fixed a bug in when statements that caused incorrect results in the otherwise clause.