Tsurugi 1.12.0 - Release Notes (en)

[akirakw]

1.12.0 Release Notes

Overview

Tsurugi 1.12.0 adds support for the TRUNCATE TABLE statement, which deletes all rows in a table, as a new SQL feature.

TRUNCATE TABLE can delete all rows of a table quickly even when the table holds a large amount of data, which makes it suitable for use cases such as emptying a work table used for temporary bulk data, or replacing a validation dataset to repeat trials.
A DELETE statement without a condition performs the same operation, but its processing time grows with the amount of data, so you can choose between the two depending on your purpose.

TRUNCATE TABLE keeps the table definition and indexes as they are, so it does not require the index re-creation that is needed when a table is initialized by combining DROP TABLE and CREATE TABLE.

Tsurugi 1.12.0 also adds support for the EXTRACT function, which extracts a specific field from datetime types.

The EXTRACT function extracts fields such as year, month, day, hour, minute, and second from values of the DATE, TIME, TIMESTAMP, and TIMESTAMP WITH TIME ZONE types. In addition to the SQL standard notation, Tsurugi supports an extended notation that specifies the range to extract, such as YEAR TO HOUR.

These features make it easier to handle time-series data, for example counting rows per day or per time slot, aggregating sales by month, or narrowing down the data to a specific year and month.

This version also includes several important bug fixes and stability improvements.
We strongly recommend all users of previous versions to upgrade to this release.

New Features and Improvements

New Features and Improvements - SQL

• Added support for the TRUNCATE TABLE statement
  • TRUNCATE TABLE is a statement that deletes all rows of the target table. Compared with a DELETE statement without a condition, it can process tables with a large amount of data quickly. The table definition and indexes are kept.
  • The syntax is as follows. For a table that has an identity column (GENERATED ... AS IDENTITY), you can specify how the sequence is handled with RESTART IDENTITY / CONTINUE IDENTITY.

    -- Deletes all rows of the table
    TRUNCATE TABLE t;
    
    -- Resets the sequence associated with the table to its initial value
    TRUNCATE TABLE t RESTART IDENTITY;
    
    -- Keeps the current value of the sequence (the default when omitted)
    TRUNCATE TABLE t CONTINUE IDENTITY;

Important

TRUNCATE TABLE is treated as a DDL statement, in the same way as CREATE TABLE and DROP TABLE.
For this reason, the execution of TRUNCATE TABLE cannot be undone by a rollback operation. There is no way to restore the deleted table data, so execute it only after checking carefully.

Important

In this version, there is a restriction where TRUNCATE TABLE ... RESTART IDENTITY cannot be used under certain conditions on tables that have an identity column. For details, see the following document:

• Tsurugi 1.12.0 - Known Issues (en) - TSURUGI-IS-812

• Added support for the EXTRACT function, which extracts a field from datetime types

  • The EXTRACT function extracts a specific field from a value of the DATE, TIME, TIMESTAMP, or TIMESTAMP WITH TIME ZONE type.

  • In the SQL standard notation, you can specify YEAR, MONTH, DAY, HOUR, MINUTE, SECOND, TIMEZONE_HOUR, and TIMEZONE_MINUTE. The return type is DECIMAL for SECOND only, and INTEGER for the others.

    SELECT EXTRACT(YEAR FROM TIMESTAMP '2026-07-01 12:34:56.789');      --> 2026
    SELECT EXTRACT(MONTH FROM TIMESTAMP '2026-07-01 12:34:56.789');     --> 7
    SELECT EXTRACT(SECOND FROM TIMESTAMP '2026-07-01 12:34:56.789');    --> 56.789000000
    SELECT EXTRACT(SECOND(1) FROM TIMESTAMP '2026-07-01 12:34:56.789'); --> 56.7 (one fractional digit)

  • As an extended notation, Tsurugi also supports specifying the range to extract. You can specify DATE, YEAR TO MONTH, YEAR TO DAY, YEAR TO HOUR, YEAR TO MINUTE, and YEAR TO SECOND, and the returned value has the fields below the specified range truncated. For YEAR TO HOUR and the finer ranges, only TIMESTAMP-family types can be specified as the operand.

    SELECT EXTRACT(DATE FROM TIMESTAMP '2026-07-01 12:34:56.789');          --> 2026-07-01
    SELECT EXTRACT(YEAR TO MONTH FROM TIMESTAMP '2026-03-21 12:34:56.789'); --> 2026-03-01
    SELECT EXTRACT(YEAR TO HOUR FROM TIMESTAMP '2026-07-01 12:34:56.789');  --> 2026-07-01 12:00:00
    SELECT EXTRACT(YEAR TO MONTH FROM DATE '2026-03-21');                   --> 2026-03-01

  • By using the EXTRACT function in a WHERE clause condition, you can narrow down the data to a specific year or month.

    SELECT id FROM events WHERE EXTRACT(YEAR FROM ts) = 2026 AND EXTRACT(MONTH FROM ts) = 8;

  • To group by the result of the EXTRACT function, retrieve the result of the EXTRACT function as a column in a WITH clause or a subquery, and use that column in the GROUP BY clause.

    WITH s AS (SELECT EXTRACT(YEAR TO MONTH FROM ts) AS ym, amount FROM sales)
    SELECT ym, SUM(amount) FROM s GROUP BY ym ORDER BY ym;

Note

In this version, the EXTRACT function cannot be specified directly in a GROUP BY clause. Support in a future version is under consideration.

For other detailed specifications and current restrictions on SQL features, see the following document in tsurugidb:

• Available SQL features in Tsurugi

New Features and Improvements - UDF

• Added support for specifying multiple endpoints per UDF plugin and connecting to them in round-robin fashion
  • To distribute the load of UDF processing across multiple peer nodes, the endpoint parameter in the [udf] section of the UDF plugin configuration file (<plugin-name>.ini) can now take multiple endpoints separated by a pipe character |. When multiple endpoints are specified, an endpoint is selected in round-robin fashion for each UDF invocation.
    • In this version, if the UDF server at the selected endpoint has stopped, an error is returned once. After that, the stopped server is excluded and endpoints are selected in round-robin fashion from the remaining ones.
    • When the stopped UDF server is restarted, dispatching to it resumes automatically on subsequent invocations. The interval until it resumes is determined by a backoff scheme based on how long the UDF server was down.
  • The --grpc-endpoint and --grpc-server-endpoint options of udf-plugin-builder can now also take multiple endpoints separated by a pipe character |.
  • The corresponding parameters in the configuration file tsurugi.ini can now also take multiple endpoints separated by a pipe character |.

Note

For detailed specifications and current restrictions on UDF features, see the following document in tsurugi-udf:

• tsurugi-udf - udf-plugin

For details on tsurugi.ini, see the following document:

• tsurugidb - Configuration file parameters

New Features and Improvements - Client

• Tsurugi MCP Server
  • Updated the dependent MCP Java SDK to version 2.0.1

Bug Fixes

Bug Fixes - SQL

• Fixed an issue where a number of 2^64 (18446744073709551616) or greater specified in certain places in SQL that are not treated as numeric literals was handled as 0
• Improved the display of the candidate list shown when a SQL syntax error occurs
• Fixed an issue where specifying an unsupported type such as CAST(NULL AS BIT(...)) in an expression caused the database to terminate abnormally under certain conditions
• Fixed an issue where including a column that constitutes the primary key in a secondary index with DESC caused INSERT into that table to fail
• Fixed an issue where executing SELECT COUNT(DISTINCT ...) on a large number of records resulted in an error or made the database unstable
  • This fixes the issue described in Tsurugi 1.11.0 - Known Issues (en) - TSURUGI-IS-945.

Bug Fixes - Authentication & Authorization

• Fixed an issue where the validity period of a credentials file generated by tgctl credentials was shorter than the default of 90 days, and the --expiration option was not applied either

Bug Fixes - UDF

• Fixed an issue where connecting from the UDF BLOB client to the BLOB relay service caused the database to terminate abnormally when enabled in the [grpc_server] section of tsurugi.ini was false
• Fixed an issue where installing udf-plugin failed after the dependent scikit-build-core was updated to version 1.0.0

Upgrade Notes

Client Compatibility List

Tsurugi 1.12.0 supports the following clients.
Please upgrade each client as needed when upgrading Tsurugi.

Important

For Tsurugi UDF, you must regenerate and redeploy UDF plugins using udf-plugin-builder.

Client	Version
Tsubakuro/Java	1.16.0+
Iceaxe	1.16.0+
Tanzawa	1.16.0+
Tsubakuro/Rust	0.10.0+
Tsurugi Python DB-API	0.10.0+
Tsurugi MCP Server	0.8.0+
Tsurugi JDBC	0.5.0+
Tsurugi UDF	0.5.0+

Upgrade Procedure

For instructions on upgrading from previous versions, see:

• Tsurugi Upgrade Guide

Other Information

For a full list of changes in this version, see the Changelog:

• Tsurugi 1.12.0 - Changelog

For known issues and additional information about fixes in this version, see:

• Tsurugi 1.12.0 - Known Issues(en)

---

This discussion was created from the release 1.12.0.