10.9 Release Notes: Mar 17, 2026-Mar 20, 2026¶

Snowflake supports directory, root stage, and SnowGit imports¶

Snowflake now supports the following import types for Python, Java, and Scala UDFs, UDTFs, UDAFs, and stored procedures:

• Directory imports: Import an entire directory from a stage.

• Root stage imports: Import from the root of a stage.

• SnowGit imports: Import folders from Git-backed refs such as @repo/branches/main/....

For more information, see Python UDF handler examples, Scala UDF handler examples, and Java UDF handler examples.

TSS history account usage view (General availability)¶

Customers can now view information about the registration and activation of customer-managed keys (CMKs) with Tri-Secret Secure by using the TRI_SECRET_SECURE_HISTORY account usage view.

For more information, see TRI_SECRET_SECURE_HISTORY view.

DML error logging for tables¶

This release adds support for DML error logging for tables. When this feature is turned on for a table, data manipulation language (DML) statements can continue when they perform an operation on the table and a supported error occurs. Errors are logged in an error table that is associated with the base table.

For more information, see DML error logging.

Additional date and time formats¶

Date and time formats are standardized ways of representing dates and times. This release adds the following additional date and time formats: Y, MO, D, H24, H12, HH, H, ME, S, and P.

For more information, see date and time formats.

Additional fixed-position numeric format models¶

SQL format models are used to specify how numeric values are converted to text strings and vice versa. This release adds the following additional fixed-position numeric format models: % and parameterized TM9.

For more information, see fixed-position numeric format models.

CKE document access history¶

Providers can now track which documents are being accessed in their Cortex Knowledge Extensions (CKEs) using access history data in the LISTING_ACCESS_HISTORY view.

Two new system functions help providers map hashed document IDs back to their original primary key columns:

• SYSTEM$ENCODE_CKE_PRIMARY_KEY: Transform and anonymize the primary key from the set of selected columns.

  For more information, see SYSTEM$ENCODE_CKE_PRIMARY_KEY.

• SYSTEM$CKE_HASH_FUNCTION: Hash the primary key.

  For more information, see SYSTEM$CKE_HASH_FUNCTION.

For more information, see CKE document access history.