The following deprecated features and behavior changes were introduced this month. If you have any questions about these changes, please contact Snowflake Support.
For more details about the new features, enhancements, and fixes introduced this month, see Juli 2021.
Unless otherwise noted, these changes were enabled through the 2021_06 bundle in Release 5.28.
In this Topic:
DESCRIBE SECURITY INTEGRATION Command: New Properties for Snowflake OAuth¶
With this release, the output of the DESCRIBE INTEGRATION command for Snowflake OAuth security integrations includes two new properties:
For Snowflake OAuth, a list of all supported endpoints for a client application to receive an authorization code from Snowflake.
For Snowflake OAuth, a list of all supported endpoints for a client application to exchange an authorization code for an access token or to obtain a refresh token.
SQL Changes — General¶
Cloned Tables: Change to How Sequences are Referenced¶
With this release, when a database or schema is cloned and it contains a table that references a sequence in the same database/schema, the cloned table references the cloned sequence in the cloned object.
When a database or schema was cloned that contained a table that referenced a sequence in the same database/schema object, the cloned table continued to reference the original sequence in the source object.
The cloned table references the cloned sequence in the cloned database/schema.
SQL Changes — Commands & Functions¶
DESCRIBE USER / SHOW USERS Commands: New DEFAULT_SECONDARY_ROLES Value in Output¶
With this release, the output of the DESCRIBE USER and SHOW USERS commands includes a new DEFAULT_SECONDARY_ROLES value. The value is displayed as a row in the DESCRIBE USER output and as a column in the SHOW USERS output.
This value is added to support future functionality.
The SHOW USERS command displays the DEFAULT_SECONDARY_ROLES column in the middle of the output. Verify whether your scripts rely on the location of columns in the command output and adjust them to be position independent.
DESCRIBE INTEGRATION / SHOW INTEGRATIONS Commands: Comment Included in Output¶
With this release, the output of the DESCRIBE INTEGRATION and SHOW INTEGRATIONS commands change as follows:
The CREATE INTEGRATION and ALTER INTEGRATION commands supported specifying a comment; however, the comment was not always displayed in the output of certain commands:
DESCRIBE INTEGRATION output did not include a COMMENT row.
SHOW INTEGRATIONS output did not include a COMMENT column.
The commands display comments in the appropriate output row/column.
This current behavior applies to all types of integrations: API, notification, security, and storage.
SHOW MANAGED ACCOUNTS Command: New Account Name Field in Output¶
With this release, the output of the SHOW MANAGED ACCOUNTS command changes as follows:
urlcolumn displayed the account locator URL in the following format:
urlcolumn displays the account name URL, the new URL format introduced in the Organizations feature. The new URL has the following format:
In addition, the output includes a new column,
account_locator_url, to display the account locator URL.
Depending on the region and cloud platform where your account is hosted, the account locator URL may have additional segments as follows:
CURRENT ACCOUNT Function: Change to Account Name Returned in Output¶
This change has been postponed and removed from the 2021_06 bundle. The new date has not been determined yet.
With this release, the output of the CURRENT_ACCOUNT function changes as follows:
The function returned the account locator (formerly referred to as the account name) for the user’s current session.
The account locator is a read-only identifier that is assigned to the account when it is provisioned and cannot be changed.
The function returns the new version of the account name, introduced in the Organizations feature.
Account names uniquely identify an account within an organization and can be changed by users with the ORGADMIN role.
Data Loading / Unloading Changes¶
COPY INTO <location> Command: NUMBER Column Mapping to Parquet Data Type¶
When NUMBER columns in Snowflake tables are unloaded to Parquet files, the corresponding columns in the files have the equivalent DECIMAL logical data type.
With this release, the criteria for determining the precision and scale of the columns in the unloaded Parquet files has changed as follows:
The DECIMAL column in the Parquet files had the lowest precision that could represent the range of values in the column. This behavior could lead to inconsistent data types among the unloaded files, making the files unusable with external tools.
The DECIMAL column in the Parquet files has the same precision and scale as the source column in the Snowflake table.
For example, if the table column has the default precision and scale
(NUMBER(38, 0)), then the same column is unloaded to a
DECIMAL(38, 0)column in the Parquet files.
Data Sharing Changes¶
Other Implemented Changes¶
Snowflake Connector for Python: Changes to the write_pandas() Function¶
Version 2.5 of the Snowflake Connector for Python changes the way in which the write_pandas() function quotes column names:
In version prior to 2.5:
If the column name does not meet the requirements for an unquoted identifier, the
In versions 2.5 and later:
Because the column names are quoted, the case of the column names in the DataFrame will be expected to match the case of the column names in the database table.