Snowflake Release Notes

Snowflake can now use primary keys in dynamic tables to track row-level changes and enable incremental refresh downstream of full-refresh dynamic tables. This release includes the following capabilities:

• Base table-defined primary keys: When a base table has a primary key with the RELY property, Snowflake uses it for change tracking in downstream dynamic tables. This is especially useful when the base table is periodically rewritten through INSERT OVERWRITE, which normally prevents change tracking across table versions.

• Query-derived primary keys: Snowflake automatically derives primary keys from the query definition of a dynamic table. Queries with GROUP BY clauses or QUALIFY ROW_NUMBER() = 1 filters produce unique constraints that Snowflake relies on for change tracking.

• Incremental refresh on full-refresh dynamic tables: Dynamic tables in incremental refresh mode can now read from upstream dynamic tables that use full refresh mode, as long as the upstream table has a system-derived primary key. To use this capability, set REFRESH_MODE = INCREMENTAL explicitly on the downstream dynamic table.

New features and improvements

• The driver now includes SPCS_TOKEN in login requests when running inside a Snowpark Container Services (SPCS) container (detected via the SNOWFLAKE_RUNNING_INSIDE_SPCS environment variable).

• Extended login-request telemetry with cloud platform and environment detection (AWS Lambda, EC2, Azure VM/Functions, GCE/Cloud Run, GitHub Actions). Detection runs once at startup in the background within a 200ms timeout. You can disable this feature by setting the SNOWFLAKE_DISABLE_PLATFORM_DETECTION environment variable.

...

Bug fixes

• Fixed an issue where idle sessions were not evicted from the connection pool when closing them fails.

• Fixed an issue where sessions that receive HTTP 401 during query execution were returned to the connection pool.

...

Dynamic Apache Iceberg™ tables now support the following table properties:

• PARTITION BY: Partition the table using Iceberg partition expressions such as identity, bucket, truncate, year, month, day, and hour transforms.
• TARGET_FILE_SIZE: Control the target Parquet file size for table writes. Defaults to AUTO, which lets Snowflake choose the optimal file size.
• PATH_LAYOUT: Choose between a flat or hierarchical (Hive-style) directory layout for data files. Use HIERARCHICAL together with PARTITION BY to write data to partition-aware paths.

Clean Rooms API Version: 14.3

The following new features and enhancements are now available in Snowflake Data Clean Rooms:

• Collaboration API - Generally Available: The new Collaboration API is now generally available. The new architecture supports fully symmetric, multi-party collaboration with flexible roles and fine-grained data access controls for any number of participants. Read the overview and try out building a new collaboration yourself.
• Collaboration Configuration APIs: Two new stored procedures are now available for managing collaboration configuration flags:
  • GET_CONFIGURATION returns a table of configuration key-value pairs for the collaboration. Currently exposes the TEMPLATE_AUTO_APPROVAL setting per collaborator.
  • SET_CONFIGURATION sets a configuration value for the collaboration.
• Case-Insensitive Code Spec IDs: Code spec IDs are now case-insensitive, extending the case-insensitive identifier support introduced in the previous release. You no longer need to match the exact casing used during registration when referencing a code spec by ID.