{"id":120215,"date":"2024-10-17T19:09:08","date_gmt":"2024-10-17T17:09:08","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/vypusk-subd-mysql-9-1-0"},"modified":"2024-10-17T19:09:08","modified_gmt":"2024-10-17T17:09:08","slug":"vypusk-subd-mysql-9-1-0","status":"publish","type":"post","link":"https:\/\/prohoster.info\/en\/blog\/news\/vypusk-subd-mysql-9-1-0","title":{"rendered":"Release of DBMS MySQL 9.1.0","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
Oracle has formed a new branch of the MySQL DBMS, version 9.1.0. MySQL Community Server 9.1.0 builds have been prepared for all major Linux distributions, FreeBSD, macOS, and Windows. Under the new release model implemented last year, MySQL 9.1 is categorized into the 'Innovation' branches, which will also include the next major release, MySQL 9.2. Innovation branches are recommended for those who want early access to new features, and they are published every three months, supported only until the next major release (for example, support for branch 9.0 ended after the release of 9.1). An LTS release is planned for summer next year, which is recommended for implementations that require predictability and long-term stability. Following the LTS branch, a new Innovation branch will be created: MySQL 10.0. <\/p>\n
Key changes in MySQL 9.1: <\/p>\n
\n
The 'CREATE VIEW' operation has been enhanced with support for the 'IF NOT EXISTS' expression, allowing the creation of a view only if a view with the specified name does not already exist, thereby preventing an error if the view has already been created.\n
Loading triggers during read-only operations has been eliminated. Previously, triggers were loaded on any access to a table, which led to memory consumption and CPU load in situations where triggers were not used (for example, during SELECT queries). In MySQL 9.1, the processing and loading of triggers are separated into two stages: reading metadata about the trigger and parsing + executing the trigger. Data obtained in the first stage is stored once and then shared across different instances of the trigger. Parsing and executing the trigger occurs only for expressions that result in data changes. Additionally, to reduce memory consumption, dynamic memory allocation for error handling in triggers has replaced static buffer allocation.\n
The output of the 'EXPLAIN' operation has been expanded to include information about multi-range reading and the use of the semijoin strategy, where in JOIN operations, only rows from one table are processed.\n
Atomic execution of 'CREATE DATABASE' and 'DROP DATABASE' operations has been ensured to protect against failures in case of a crash or file system errors occurring during these operations. For example, previously, a crash during 'CREATE DATABASE' could lead to an incomplete and non-functional database structure, requiring manual intervention for cleanup.\n
Stored procedures written in JavaScript now support the VECTOR type. Data of this type can now be passed as input and return parameters.\n
Added the ability to authenticate when connecting to the DBMS using the OpenID Connect protocol.\n
27 vulnerabilities have been fixed, with the most serious one (CVE-2024-5535) assigned a critical severity level (9.1 out of 10). The vulnerability can be exploited remotely without authentication. The issue is caused by reading data from outside the allocated buffer in the OpenSSL library, leading to a crash or memory content leakage in response after connecting to proxy server<\/a> with incorrect protocol parameters. <\/ul>\n