In this release (Release 2025.04.22), we have introduced a series of new features, enhancements, and fixes. These updates will be rolled out in phases to the following regions and are expected to be completed within one to two weeks from the release date, depending on your specific region.

Domestic Regions

Alibaba Cloud (Shanghai)
Tencent Cloud (Shanghai/Beijing/Guangzhou)
AWS (Beijing)

International Regions

Alibaba Cloud (Singapore)
AWS (Singapore)

New Features and Enhancements

Federated Query Enhancement

**【Preview Release】**ORC Format Support: External Catalog/External Schema now supports read and write operations for ORC - formatted tables from HMS.

Data Import and Export Optimization

Import Command COPY INTO <table>

Smart line break recognition: Supports both \r\n and \n formats.
on_error=abort|continue strategy:
- continue mode: Skips compression format errors and continues execution.
- abort mode: Terminates immediately upon encountering an error.
- After execution, it can display the list of imported files.

Export Command COPY INTO <location>

writebom=true parameter: When exporting CSV files, it includes a BOM header, resolving the issue of Chinese character garbling when opened in Excel and improving cross - platform compatibility.
overwrite=true parameter: Empties the target folder (including subdirectories) before importing.

Pipe Function Enhancement

Continuous Pipe file import: Supports ignoring errors (skipping compression format errors) and continuing execution via the on_error=continue parameter.

SQL Features

INTERVAL Type Extension: Allows expressions in INTERVAL expr unit, such as interval 1+2 year.
Metadata Display: The DESC/SHOW commands now display schema information of the share type.
Data Sampling: Added TABLESAMPLE sampling syntax for efficient data sampling analysis.
Vector Search: Supports the ef parameter, which can be configured before executing a query:set cz.vector.index.search.ef=64;
Inverted Index: When creating an inverted index, supports the 'mode' = 'max_word' parameter to enable a more fine - grained tokenization mode:properties ('analyzer' = 'chinese', 'mode' = 'max_word');

Functions

Built - in Functions

Improved Performance of GET_JSON_OBJECT: The implementation of GET_JSON_OBJECT has been optimized to enhance JSON parsing efficiency.

Custom SQL Functions

SQL Function Enhancement: Supports creating user - defined SQL functions of the RETURNS TABLE type.

UDF (User - Defined Functions)

External Function now supports referencing resource files via VOLUME addresses.

User Volume Address Format: volume:user://~/upper.jar
- user indicates the use of the User Volume protocol.
- ~ represents the current user and is a fixed value.
- upper.jar is the name of the target file.
Table Volume Address Format: volume:table://table_name/upper.jar
- table indicates the use of the Table Volume protocol.
- table_name is the name of the table and should be filled in according to the actual situation.
- upper.jar is the name of the target file.
Volume Address Format: volume://volume_name/upper.jar
- volume_name is the name of the created volume.
- upper.jar is the name of the target file.

Volume

【Preview Release】 New internal named Volume object: Named Volume is a user - defined storage location, primarily used for staging data files before importing them into a table. Unlike automatically created user - level (User Volume) and table - level (Table Volume) volumes, a Named Volume must be explicitly created by the user and offers more flexible configuration options. This makes it better suited for team collaboration and complex data loading scenarios. Additionally, the internal Volume is stored in the internal storage space managed by Cloud Lakehouse, eliminating the need for additional cloud storage configuration and providing users with a more convenient and efficient storage solution.

Caching

Preload Cache Status Query: Previously, the cache usage could only be checked via SHOW VCLUSTER vcname PRELOAD CACHED STATUS when the virtual cluster was in a running state. Now, this command also supports checking the cache status and indicating the cluster running state when the virtual cluster is in a paused state.

INFORMATION SCHEMA

The Information Schema has added the OBJECT_PRIVILEGES view, which allows querying all data object privilege grants within the system:
- It can directly query all privileges granted to a specified user (including those indirectly obtained through roles).
- It can directly query which users have been granted privileges for a specified object (such as a table or view, including those indirectly granted through roles).
- It currently does not support querying function authorization situations.
- The authorization data within the view may have a delay of up to 15 minutes compared to real - time data.

SDK

Python SDK

The Python SDK has been enhanced to support the SQLAlchemy 2 interface.

BUG Fixes

Python SDK
- Fixed the issue of the hints parameter being ineffective in the executemany method.
- Resolved the error in the Bulkload SDK when writing to partitioned tables.
- Fixed the error in the Python SDK when executing the optimize syntax.

Behavioral Changes

Simplified External Function Authorization: Changed from requiring both USE FUNCTION and VOLUME usage privileges to only needing the USE FUNCTION privilege.
PIPE Function: Added a new failed state (previously only RUNNING and PAUSED states existed). This facilitates monitoring systems to capture exceptions and allows for setting up alert mechanisms to notify about the FAILED state. It can be viewed via DESC <pipe_name>.