What dbeaver/dbeaver shipped
Written by FoxPlug from the repository's own releases; not affiliated with Dbeaver. Paste your own repository in the box below and FoxPlug writes the same weekly update for your product. This page is an automatic summary of the releases, pull requests and commits in github.com/dbeaver/dbeaver; Dbeaver did not write it and does not use or endorse FoxPlug. Every line links to the change it describes.
Get a weekly update like this for your product, free
Week of November 2, 2026
What shipped
- Added support for PostgreSQL 14+ publications and subscriptions in the navigator and metadata editors, with a three-step wizard for new subscriptions. Pull request #42538
- Added a named Apache Arrow Flight SQL entry to the connection wizard using the official Apache JDBC driver, so a custom generic driver is no longer needed. Pull request #42645
- Made the connection name editable in connection settings and kept it synchronized with General settings and generated names. Pull request #42600
- Enabled multirow selection in the MySQL dump and export object chooser using native SWT behavior. Pull request #42766
- Fixed endless segmented data transfers from SQL Server when Use SQL to limit fetch size is enabled by applying the JDBC offset after TOP-based query transformation. Pull request #42764
- Added independent CSV header quoting options: always, when necessary, and never, while data-cell quoting stays separate. Pull request #42728
- Added nested ClickHouse Map, Tuple, and Array editing with typed SQL expressions for updates. Pull request #42724
- Updated PostgreSQL JDBC from 42.7.13 to 42.7.14 in PostgreSQL, Greenplum, and generic driver descriptors. Pull request #42738
- Updated the MariaDB Connector/J driver download minimum to the patched 3.5.11 release. Pull request #42755
- Disabled MySQL native script execution in the SQL editor when no database is selected, preventing a cast failure. Pull request #42747
Why it matters
This week brings new database support for PostgreSQL publications and subscriptions and a named Apache Arrow Flight SQL connection entry, so more work can be done without custom driver setup. It also fixes several connection and data transfer problems, including SQL Server segmented transfers and MySQL native execution without a database, and adds finer CSV header quoting control. A release 26.2.0.1 is also listed.
Changelog entry
- Added PostgreSQL 14+ publications and subscriptions to the navigator and metadata editors. Pull request #42538
- Added an Apache Arrow Flight SQL driver entry to the connection wizard. Pull request #42645
- Made the connection name editable in connection settings. Pull request #42600
- Enabled multirow selection in the MySQL dump and export object chooser. Pull request #42766
- Fixed SQL Server segmented data transfers when Use SQL to limit fetch size is enabled. Pull request #42764
- Added CSV header quoting options: always, when necessary, and never. Pull request #42728
- Added nested ClickHouse Map, Tuple, and Array editing. Pull request #42724
- Disabled MySQL native execution when no database is selected. Pull request #42747
- Updated PostgreSQL JDBC to 42.7.14. Pull request #42738
- Updated MariaDB Connector/J minimum to 3.5.11. Pull request #42755
- Fixed duplicate Oracle package search results. Pull request #42716
- Released 26.2.0.1. Release
This week: PostgreSQL publications and subscriptions in the navigator, a named Apache Arrow Flight SQL driver, fixed SQL Server segmented transfers, and independent CSV header quoting options. Also released 26.2.0.1.
This week the project added PostgreSQL 14+ publications and subscriptions to the navigator and metadata editors, including a three-step wizard for new subscriptions. A named Apache Arrow Flight SQL entry now appears in the connection wizard using the official Apache JDBC driver, so a custom generic driver is no longer required. Fixes this week cover endless segmented data transfers from SQL Server with SQL fetch-size limiting, MySQL native script execution when no database is selected, and duplicate Oracle package search results. CSV export gained independent header quoting options, and PostgreSQL JDBC moved to 42.7.14. Release 26.2.0.1 is also available.
Week of October 26, 2026
What shipped
- Release 26.2.4 shipped with AI provider configuration through environment variables, fixes for AI scope selection in nested object folders, and improved AI Chat image previews. Release
- The new connection dialog now always shows commercial driver categories such as CData as individual drivers regardless of the Show commercial drivers preference. Pull request #42646
- A MotherDuck driver was added that keeps the token out of the connection URL. Pull request #42436
- Updates from a ZIP archive on Windows now use a PowerShell helper that extracts and validates the archive instead of launching it as an installer. Pull request #42521
- JDBC URL parameters can now be replaced in place using the existing template parser, including host and port, without rebuilding the rest of the URL. Pull request #42595
- POM-only Maven dependencies are now skipped when assembling driver libraries, so requests for nonexistent JARs such as apache-curator-4.2.0.jar no longer happen. Pull request #42601
- The KEEP keyword was added to the Oracle SQL dialect, restoring syntax highlighting and autocompletion for the KEEP clause inside CTEs. Pull request #42625
- Firebird triggers are now recognized with DML actions in any order rather than a fixed order. Pull request #42577
- Oracle TIMESTAMP literal formatting was fixed for the offset and fraction when values are converted to SQL. Pull request #42592
- The Collect diagnostic info dialog now remembers the last used output folder, including across application restarts. Pull request #42491
Why it matters
This week brought a maintenance release alongside fixes across drivers, dialect parsing and the connection wizard. Commercial driver categories are now always visible, a token-safe MotherDuck driver arrived, and Windows ZIP updates and driver library downloads were corrected. Oracle, Firebird and JDBC URL handling also saw targeted fixes.
Changelog entry
- Release 26.2.4: AI provider configuration via environment variables, AI scope selection fixes for nested object folders, improved AI Chat image previews. Release
- Connection dialog: commercial driver categories always displayed as individual drivers. Pull request #42646
- Added MotherDuck driver that keeps the token out of the connection URL. Pull request #42436
- Windows ZIP-distribution updates extract and validate the archive via a PowerShell helper. Pull request #42521
- JDBC URL parameters replaced at their original positions, including host and port. Pull request #42595
- POM-only Maven dependencies skipped when assembling driver libraries. Pull request #42601
- Oracle SQL dialect: KEEP keyword added for highlighting and autocompletion in CTEs. Pull request #42625
- Firebird triggers recognized with DML actions in any order. Pull request #42577
- Oracle TIMESTAMP literal offset and fraction format fixed. Pull request #42592
- Collect diagnostic info dialog remembers the last used output folder across restarts. Pull request #42491
- JDBC URL field stays editable only when URL mode is selected. Pull request #42602
- Show commercial drivers preference added to the driver selection page and synced between wizard pages. Pull request #42607
DBeaver 26.2.4 is out. This week: commercial driver categories always shown in the connection wizard, a MotherDuck driver that keeps tokens out of the URL, fixed Windows ZIP updates, POM-only Maven artifacts skipped in driver libraries, and Oracle KEEP plus Firebird trigger fixes
DBeaver 26.2.4 shipped this week. The new connection dialog now always shows commercial driver categories such as CData as individual drivers, independent of the Show commercial drivers preference. A MotherDuck driver was added that keeps the token out of the connection URL. Windows ZIP-distribution updates now extract and validate the archive through a PowerShell helper instead of launching it as an installer. Driver library assembly skips POM-only Maven dependencies. On the SQL side, the KEEP keyword returned to the Oracle dialect for highlighting and autocompletion, Firebird triggers are recognized with DML actions in any order, and Oracle TIMESTAMP literals format offsets and fractions correctly.
Week of October 19, 2026
What shipped
- ROLLBACK TO SAVEPOINT in the SQL editor no longer rolls back the whole manual-commit transaction. Pull request #42492
- PostgreSQL reads schema constraints once when loading all indexes instead of running a query per table. Pull request #42460
- SQL Server database descriptions are now read and saved instead of being silently dropped. Pull request #41921
- Lock Manager now works on Redshift, showing conflicting table locks and waiting sessions. Pull request #42539
- Metadata can reload automatically after DDL, with Ask, Always and Never options. Pull request #42496
- Temporary PostgreSQL SSL private key files are now restricted to owner-only permissions. Pull request #42441
- The idle connection auto-close setting is no longer lost when the configuration is reloaded. Pull request #42430
- PostgreSQL temporary view values are preserved when the metadata connection cannot resolve their row type. Pull request #42534
- CSV import date conversion no longer throws an exception for each value. Pull request #42503
- The tnsnames.ora reader is now closed after parsing TNS names. Pull request #42561
Why it matters
This week fixes several long-standing correctness problems: savepoint rollback no longer discards a whole manual-commit transaction, PostgreSQL schema index loading stops querying once per table, and SQL Server database descriptions finally persist. It also adds Lock Manager support for Redshift, automatic metadata reload after DDL, and tighter permissions on temporary PostgreSQL SSL key files, plus a set of startup and UI fixes.
Changelog entry
- ROLLBACK TO SAVEPOINT no longer rolls back the whole transaction in manual commit mode. Pull request #42492
- PostgreSQL reads schema constraints once when loading all indexes. Pull request #42460
- SQL Server database descriptions are read and persisted. Pull request #41921
- Lock Manager supported on Redshift using SVV_TRANSACTIONS. Pull request #42539
- Metadata can reload automatically on DDL execution with Ask, Always and Never options. Pull request #42496
- Temporary PostgreSQL SSL private key files are restricted to owner-only permissions. Pull request #42441
- Idle connection auto-close setting is no longer lost on reload. Pull request #42430
- PostgreSQL temporary view row values are preserved when the row type is unresolved. Pull request #42534
- CSV import date conversion no longer raises an exception per value. Pull request #42503
- Tnsnames.ora reader is closed after parsing TNS names. Pull request #42561
- JDBCResultSetCallable uses a zero-based index for attribute values. Pull request #42525
- CData tab restored in the connection dialog, with commercial drivers hidden by default. Pull request #42529
This week: ROLLBACK TO SAVEPOINT no longer rolls back the whole transaction, PostgreSQL reads schema constraints once when loading indexes, SQL Server database descriptions are saved, and Lock Manager works on Redshift.
This week's changes touch several long-standing annoyances. In manual commit mode, ROLLBACK TO SAVEPOINT no longer rolls back the entire transaction. Loading all indexes for a PostgreSQL schema now reads constraints once instead of running a query per table with a unique index. SQL Server database descriptions are now read and saved instead of being silently dropped. Lock Manager is enabled on Redshift, metadata can reload automatically after DDL with Ask, Always and Never options, and temporary PostgreSQL SSL key files are restricted to owner-only permissions. Also included: CSV import date conversion without an exception per value, a closed tnsnames.ora reader, and a fix for the idle connection auto-close setting being lost on reload.
Week of October 12, 2026
What shipped
- Released version 26.2.3, which includes AI Assistant changes: drag-and-drop of Navigator objects and files into AI Chat, attaching images from files, drag-and-drop and the clipboard, and a fix for closing multiple SQL Editor tabs clearing the selected AI Chat connection and conversation. Release
- Added a native Mimer SQL plugin, an official contribution from Mimer Information Technology, replacing the previous generic-JDBC-driver-based support with schemas, tables, columns, indexes and more. Pull request #42175
- Added query execution plan support for Apache Hive 2 and Hive 4 drivers, parsing MapReduce and Tez plans including operators, dependencies, estimates, statistics and conditions. Pull request #42421
- Fixed GBase 8s identifier quoting when a connection uses SQLMODE=mysql, so backticks are used for identifiers instead of double quotes. Pull request #41721
- Fixed SQLite metadata loading when a temporary table named
pragma_table_listshadows the built-in function, by qualifyingpragma_table_listwith the main schema. Pull request #42376 - Fixed the YashanDB user password change action, which failed with YAS-04225, by using a dedicated password manager that runs ALTER USER... IDENTIFIED BY without REPLACE. Pull request #42411
- Updated the preferred Snowflake JDBC driver from 4.1.0 to 4.3.4 to pick up fixes for two CVEs. Pull request #42398
- Canceling a data export now reports an ERROR notification reading Data transfer was canceled instead of a completion message, and the wizard closes without a duplicate cancellation dialog. Pull request #42410
- Fixed USING clause generation when changing the data type of a PostgreSQL foreign table column. Pull request #42328
- Fixed a PostgreSQL role with connection limit 0 being shown as unlimited in the Source tab; zero is now printed as zero. Pull request #42379
Why it matters
This week brings a new release plus a native Mimer SQL plugin and Hive query plan support, so users of those databases get proper tooling instead of generic driver behavior. Many of the fixes address correctness in daily work: identifier quoting, metadata loading, password changes and export cancellation. A Snowflake driver update also picks up security fixes.
Changelog entry
- Release 26.2.3 with AI Chat drag-and-drop, image attachments, and a fix for tab-close clearing the selected connection. Release
- Added native Mimer SQL support as a full plugin. Pull request #42175
- Added Hive query execution plan support for Hive 2 and Hive 4 drivers. Pull request #42421
- Fixed GBase 8s identifier quoting in MySQL mode. Pull request #41721
- Fixed SQLite metadata query shadowing by qualifying
pragma_table_listwith the main schema. Pull request #42376 - Fixed YashanDB user password change failing with YAS-04225. Pull request #42411
- Updated Snowflake JDBC driver from 4.1.0 to 4.3.4 for CVE fixes. Pull request #42398
- Data export cancellation now reports an ERROR notification and skips the duplicate dialog. Pull request #42410
- Fixed USING clause generation for PostgreSQL foreign table column type changes. Pull request #42328
- Fixed PostgreSQL role connection limit 0 being displayed as -1. Pull request #42379
- Made PostgreSQL dependency types readable in the Dependencies panel. Pull request #42011
- Avoided nested Firebird metadata sessions during export by preloading column domain types. Pull request #42422
This week: 26.2.3 released, native Mimer SQL support added, Hive query plans now supported. Fixes for GBase 8s quoting, SQLite metadata, YashanDB password change, export cancellation, and a Snowflake JDBC driver update with security fixes.
This week's digest covers release 26.2.3 and a range of fixes across the databases people work with daily. A native Mimer SQL plugin replaces the previous generic JDBC support, and query execution plans are now supported for Apache Hive 2 and Hive 4 drivers. Fixes include GBase 8s identifier quoting in MySQL mode, SQLite metadata loading when a temporary table shadows pragma_table_list, YashanDB password changes, and PostgreSQL foreign table column type changes. Canceling a data export now reports cancellation clearly, and the preferred Snowflake JDBC driver moves from 4.1.0 to 4.3.4 with fixes for two CVEs.
Week of October 5, 2026
What shipped
- Date and timestamp value handlers no longer share a single SimpleDateFormat instance, so parallel data exports stop silently corrupting date values. Pull request #42213
- Vertica JDBC connections work again on Java 25, where an empty default keystore path was being treated as an existing file. Pull request #42340
- Closing the last SQL editor no longer clears the selected connection and conversation in the chat. Pull request #42311
- Metadata for newly created objects is refreshed after the transaction commits in manual commit mode. Pull request #42101
- Saving data editor changes now warns when the database reports zero affected rows instead of reporting success. Pull request #42331
- Online and offline SQL Server databases now use different icons in the navigator so they can be told apart without opening each one. Pull request #41919
- Backspace and Delete in the SQL editor delete text with an empty string instead of null, avoiding the resulting null document event. Pull request #42305
- Firebird EXECUTE BLOCK statements are prepared and their parameter markers bound rather than run through the usual script path. Pull request #42292
- Non-admin users can open S3 files while the project-resource-view permission check is kept for project resources. Pull request #42317
- Athena connections pass the driver's ApplicationName parameter so its API calls carry it in the User-Agent. Pull request #42316
Why it matters
This week's changes address several long-standing correctness and stability problems: concurrent exports no longer corrupt dates, Vertica works on Java 25, and data editor saves surface zero affected rows instead of hiding them. Connection and editor behavior is also steadier, with SQL Server database state visible in the navigator and the chat keeping its connection when the last editor closes.
Changelog entry
- Fixed silent date and timestamp corruption in parallel data exports caused by shared SimpleDateFormat instances. Pull request #42213
- Fixed Vertica TLS failure on Java 25 caused by an empty default keystore path. Pull request #42340
- Fixed the chat losing its selected connection and conversation when the last SQL editor is closed. Pull request #42311
- Refreshed metadata for newly created objects after transaction commit in manual commit mode. Pull request #42101
- Added a warning when saving data editor changes reports zero affected rows. Pull request #42331
- Marked online and offline SQL Server databases with different icons in the navigator. Pull request #41919
- Fixed text deletion in the SQL editor producing a null document event. Pull request #42305
- Fixed Firebird EXECUTE BLOCK execution by preparing the statement and binding parameter markers. Pull request #42292
- Passed the Athena JDBC driver ApplicationName connection parameter. Pull request #42316
- Allowed non-admin users to open S3 files. Pull request #42317
- Added confirmation before running a shell command to obtain a connection password. Pull request #42215
- Preserved custom colors across light and dark theme changes and restarts in standalone products. Pull request #42273
This week: concurrent data exports no longer corrupt dates, Vertica works on Java 25, zero affected rows now warn on save, SQL Server shows online/offline in the navigator, Firebird EXECUTE BLOCK runs correctly, and the chat keeps its connection when the last editor closes.
This week's changes: date value handlers no longer share a SimpleDateFormat instance, fixing silent corruption during parallel exports. Vertica JDBC connections work on Java 25. Saving data editor changes warns when the database reports zero affected rows. SQL Server databases show online and offline differently in the navigator. Firebird EXECUTE BLOCK statements are prepared and bound instead of run as scripts. The chat keeps its selected connection when the last SQL editor closes. Athena passes its application name to the driver. Non-admins can open S3 files. Metadata for newly created objects refreshes after commit.
Week of September 28, 2026
What shipped
- Export target wizard page redesigned with a compact source table, vertical source toolbar, and separate Format and Migration choices for better organization. Pull request #42280
- ClickHouse now supports table rename operations using the RENAME TABLE SQL statement with fully qualified table names. Pull request #42278
- Fixed PostgreSQL bulk imports that were failing or corrupting values containing literal backslashes when using custom CSV escape characters. Pull request #42251
- Added support for YashanDB, a database system widely adopted in China across finance, government, and enterprise sectors. Pull request #40195
- Fixed database user password changes by classifying expired-password errors and supporting non-interactive updates for Oracle, MySQL, MariaDB, DB2, and HANA. Pull request #42242
- Fixed Tree row highlighting in Windows dark theme by limiting workarounds to specific controls and rendering native TreeView states across all columns. Pull request #42284
- Fixed widget colors when switching between light and dark themes by refreshing Windows native control styling for text, buttons, combo boxes, trees, and tables. Pull request #42257
- Added a classic SWT theme selector to the User Interface preference page with preview, cancel rollback, persistence, and restart prompts. Pull request #42233
- Fixed function search in the Create Trigger dialog so typing a function name now properly filters the visible list. Pull request #42147
Why it matters
This week brings stability improvements across multiple database systems and UI themes, including new YashanDB support for users in Asia-Pacific regions. Critical fixes for password changes, bulk imports, and dark theme handling address real pain points. The redesigned export wizard and AI engine selector improve workflow efficiency.
Changelog entry
- Export wizard: Redesigned target page with compact source table and separate Format and Migration choices Pull request #42280
- ClickHouse: Added table rename support with RENAME TABLE SQL generation Pull request #42278
- PostgreSQL: Fixed bulk imports failing with literal backslashes and custom CSV escape characters Pull request #42251
- Added support for YashanDB database system Pull request #40195
- Database passwords: Fixed expired-password handling for Oracle, MySQL, MariaDB, DB2, and HANA Pull request #42242
- Windows dark theme: Fixed Tree row highlighting and native TreeView state rendering Pull request #42284
- Theme switching: Fixed widget color refresh for Text, buttons, combo boxes, trees, and tables Pull request #42257
- Preferences: Added classic SWT theme selector to User Interface page with preview and rollback Pull request #42233
- SQL editor: Fixed Tab navigation between template fields in linked editing mode Pull request #42227
- Triggers: Fixed function search in Create Trigger dialog Pull request #42147
26.2.2 ships with an AI engine selector redesign, YashanDB support, fixed password changes for Oracle/MySQL/MariaDB/DB2/HANA, and PostgreSQL bulk import fixes for backslash handling.
DBeaver 26.2.2 is out with significant improvements: redesigned AI engine selector with better accessibility, new YashanDB database support, fixed password change operations across Oracle, MySQL, MariaDB, DB2, and HANA, and resolved PostgreSQL bulk import issues with escape character handling. Plus dark theme stability fixes and a new theme switcher in preferences.
Week of September 21, 2026
What shipped
- DBeaver 26.2.1 released with fixes for disabled tool visibility, AI Chat dark theme flash, and improved prompt caching. Release
- ClickHouse read-only sessions no longer fail when DBeaver attempts to set JSON output formatting. Pull request #42211
- GBase 8s now properly handles JSON type values in Copy As SQL operations. Pull request #41839
- PostgreSQL PUBLIC pseudo-role now appears in Permissions tabs for all database objects. Pull request #42091
- Autocomplete activation can now be configured to trigger on Enter, Tab, Both, or None. Pull request #42161
- Fixed project property startup deadlock that could freeze the application with a detached Database Navigator. Pull request #42224
- Column comments containing -- no longer cause misplaced commas in generated CREATE TABLE statements. Pull request #42127
- Databricks driver updated from 3.4.1 to 3.4.3 with fix for incomplete JDBC URLs. Pull request #42208
- XLSX export target restored by packaging required Apache Commons bundles for POI 5.4.3. Pull request #42198
- Doris CSV import now works on single-backend installations by generating tables with one replica. Pull request #42094
Why it matters
Version 26.2.1 addresses critical stability issues like startup deadlocks and improves database compatibility across multiple systems. Bug fixes for PostgreSQL permissions, ClickHouse sessions, and CSV imports ensure users can work with diverse database configurations without unexpected failures.
Changelog entry
- Fixed toolbars for disabled tools remaining visible Release
- Added profile and model selectors to AI Chat Release
- Fixed white flash when opening AI Chat in dark theme Release
- Improved AI Chat prompt caching to reduce repeated input-token processing Release
- ClickHouse read-only sessions skip JSON-as-string setting changes Pull request #42211
- GBase 8s Copy As SQL handles JSON type values correctly Pull request #41839
- PostgreSQL PUBLIC pseudo-role appears in Permissions tabs Pull request #42091
- Fixed project property startup deadlock with detached Database Navigator Pull request #42224
- Autocomplete can trigger on Enter, Tab, Both, or None Pull request #42161
- Column comments with -- no longer cause misplaced commas in CREATE TABLE Pull request #42127
- Databricks driver updated to 3.4.3 Pull request #42208
- XLSX export restored with Apache Commons bundles for POI 5.4.3 Pull request #42198
- Doris CSV import works on single-backend installations Pull request #42094
- GBase handles missing
read_defaultstrroutine gracefully Pull request #41622 - Oracle package nodes no longer duplicate in Database Navigator Pull request #42206
DBeaver 26.2.1 ships with fixes for startup deadlocks, PostgreSQL permissions visibility, ClickHouse read-only sessions, and improved autocomplete controls.
DBeaver 26.2.1 is out with important stability and compatibility improvements. Fixed a startup deadlock affecting Database Navigator, restored PostgreSQL PUBLIC role visibility in permissions, resolved ClickHouse read-only connection issues, and improved autocomplete configuration. XLSX exports work again with updated POI dependencies. Doris CSV imports now support single-backend setups.
Week of September 14, 2026
What shipped
- Undo and redo operations are now available for result set cell edits, with Escape resetting a cell to its original value while keeping the reset undoable. Pull request #42040
- Mass row insertion performance is improved by reserving rows in one operation and initializing values with a single database session. Pull request #42102
- A new DBeaver-specific General preferences page consolidates update and regional settings with reorganized Eclipse preferences. Pull request #42131
- Data Compare has been redesigned. Pull request #42133
- A persistent preference now allows disabling the large script execution warning dialog for scripts with more than 100 statements. Pull request #42034
- MariaDB driver is updated to Connector/J 3.5.10. Pull request #42140
- Numeric type mapping for CSV imports to Timeplus is fixed to prevent silent truncation of integer values. Pull request #42047
- H2 class loading is restricted with a configurable allowlist exposed in H2 preferences. Pull request #42038
- GBase 8s driver now supports manual and URL connection configuration modes. Pull request #42106
- Preference pages across the application now include searchable keywords for easier discovery. Pull request #42136
Why it matters
This week brings usability improvements across editors and preferences, performance optimizations for data operations, and fixes for driver compatibility and data type handling. Users benefit from better undo/redo support, faster bulk inserts, and more discoverable settings.
Changelog entry
- Add undo/redo history for result set cell edits with Escape to reset cells to original values Pull request #42040
- Optimize mass row insertion by reserving rows in one operation and using single database session initialization Pull request #42102
- Add DBeaver-specific General preferences page for update and regional settings with reorganized Eclipse preferences Pull request #42131
- Redesign Data Compare feature Pull request #42133
- Add preference to disable large script execution warning dialog Pull request #42034
- Update MariaDB driver to Connector/J 3.5.10 Pull request #42140
- Fix numeric type mapping for CSV imports to Timeplus and mark native JDBC driver as non-thread-safe for Timeplus Pull request #42047
- Restrict H2 class loading with configurable allowlist in H2 preferences Pull request #42038
- Add URL connection mode support for GBase 8s driver Pull request #42106
- Add searchable keywords to preference pages across core, database integrations, AI, SQL, connection, navigator, statistics, and DataDam UI Pull request #42136
- Fix SQL editor toolbar background colors on macOS to synchronize with connection color Pull request #42160
- Improve CSmartCombo with compact popup shell and theme-derived colors Pull request #42164
- Replace network handler removal confirmation with standard Yes/No dialog Pull request #42166
- Make GitHub issue-reporting link clickable in unsupported date/time value message Pull request #42153
- Handle temporary files properly under single.dbeaver-temp root with session isolation Pull request #42073
- Fix SQL script connection associations after restart by persisting project ID instead of project name Pull request #42126
- Improve empty AI custom scope handling with actionable guidance Pull request #42130
- Fix diagnostic log collection by centralizing archive filename matching Pull request #42143
- Restore Database Navigator context menus on status-line and editor breadcrumbs Pull request #42135
This week: undo/redo for result sets, faster mass row insertion, redesigned Data Compare, searchable preferences, and fixes for MariaDB, H2 security, and CSV imports.
This week's DBeaver update includes several improvements for developers: undo/redo operations for result set edits, optimized mass row insertion performance, a redesigned Data Compare tool, and searchable preference pages. We've also updated the MariaDB driver to 3.5.10, fixed numeric type mapping for CSV imports, added configurable H2 class restrictions, and improved GBase 8s connection support. These changes enhance productivity and data integrity across the platform.