Changelog

0.6.71Published on November 15, 2026
Download

New Features

  • Settings adds toolbar customization, allowing configuration of the visibility and order of buttons such as New Connection, New Query, and Backup
  • Global Quick Open upgraded to a unified search entry, enabling structured search across connections, databases, tables, fields, SQL files, and plugin commands, with pinyin matching and scope filtering
  • Top menu adds a searchable recently closed tabs selector, allowing quick recovery even after all tabs are closed
  • Sidebar connection and database nodes add a first-level entry for New Database, with Shift+F4 shortcut support
  • When a database or table group is selected in the sidebar, a shortcut can directly open the new table editor
  • Sidebar column name double-click behavior can be configured to generate a conditional query statement
  • Execute SQL File dialog adds target Schema selection, allowing batch imports to land in a specified Schema
  • Data grid adds WHERE/ORDER BY width ratio settings for the condition panel and an ORDER BY show/hide toggle
  • Database export supports limiting the number of exported rows per table, and can take the earliest or latest N rows by a specified field
  • SQL editor supports customizing shortcuts for previous/next completion item
  • Field completion adds a setting: unambiguous fields can omit the table name prefix
  • Double-clicking brackets in the SQL editor selects the content inside the brackets
  • Oracle table right-click adds table grants view, displaying table-level and column-level privileges by grantee
  • Oracle stored procedures, functions, sequences, packages, and synonyms support exporting object DDL as SQL files
  • Oracle object browser supports directly compiling invalid stored procedures, functions, and views from the right-click menu
  • PostgreSQL sidebar adds an entry for viewing row-level security policies
  • PostgreSQL mode adds an entry for the list of user-defined aggregate functions
  • PostgreSQL sequence view displays options such as step, range, and cache, and provides a new sequence template
  • Redis key list supports filtering by type, and Stream entries also display the corresponding local time

Improvements

  • MySQL table list adds an AUTO_INCREMENT value column
  • MySQL completion supports IGNORE INDEX and USE INDEX index hints
  • MySQL provides field = '' completion when entering a string field at the start of a WHERE condition
  • MySQL table fields can be completed in the INSERT IGNORE column list
  • PostgreSQL table completion can search across Schemas when no Schema is selected, and the Schema selector supports clearing
  • PostgreSQL indicates the specific failed item when batch querying table DDL metadata fails, no longer only showing db error
  • Oracle completion for VALUES and expressions such as SYSDATE in INSERT statements is more accurate
  • Oracle short RAW values (such as SYS_GUID() results) are displayed directly in hexadecimal in the grid
  • Oracle object list metadata queries are trimmed by the requested object type, reducing query load on large Schemas
  • DM field completion handles case according to unquoted identifier rules, so lowercase usernames can also match fields
  • DM editor highlighting adds native data types, auto-increment keywords, and date functions
  • SQL Server batch execution prompts to upgrade the driver when encountering an old driver version, no longer reporting a deserialization error
  • SQL Server schema comparison lazily loads DDL for tables with the same name on both sides, halving the number of metadata queries
  • Kafka topic page adds an auto-load topics toggle saved per connection
  • RabbitMQ subscription list adds a consumer client address column
  • StarRocks view DDL formatting adds line breaks to the query part when displayed
  • TiDB object browser explicitly labels the row count column as estimated row count
  • MongoDB read-only document preview displays types such as ObjectId and BSON dates in Mongo Shell style
  • Column completion inside a CTE body references only the current CTE body and data sources declared before it
  • CSV import supports multi-character delimiters (such as |@|)
  • ClickHouse connection test errors display the full transport-layer cause chain (DNS/TCP/TLS, etc.)
  • AI preset list update: removed Atlas Cloud integration
  • AI preset list update: retired Jalapeno Cloud, prioritizing AICodeMirror
  • AI assistant recognizes SQL conversion requests for the XuguDB dialect
  • Update center displays real download progress and batch progress for Update All
  • Update center dialog height adapts to content, keeping the title and action buttons visible while the body scrolls
  • SQL parameter dialog adds one-click clearing of parameter value history
  • Schema comparison results and the deployment confirmation step display source/target connection information at the top
  • SQL editor database selector options allow directly copying the database name
  • New queries saved to the SQL library default to a database folder of the same name, with icons distinguished by connection color
  • Table details panel adds a New Query entry
  • After hiding columns on the table data page, copying the query SQL includes only visible columns
  • The data grid search overlay can be dragged to other positions within the table
  • WHERE field dropdown is ordered according to the grid's custom column order
  • Candidate fields in the quick filter and sort builder follow the table column layout order
  • Left-clicking the header sort icon directly toggles ascending/descending, while right-click still opens the full menu
  • Result set filter visualization: filtered column headers show a funnel indicator, and filtering for an individual field can be turned off separately
  • After dragging table headers to adjust column order, confirm before saving to avoid accidental layout changes
  • The object list column settings button is fixed to the right side of the header and no longer covers column titles
  • Common object list columns (row count, comments, etc.) are moved forward, and creation and update times are also displayed
  • Object list adds a reset order action to restore the built-in column order with one click
  • Oracle object browser default view search matches all loaded object types, no longer only searching table names
  • Quick Open supports searching by MySQL field name, field comment, and table comment
  • The duration column in the multi-statement execution summary adapts its unit display to the magnitude
  • Entries such as New Query use unified icons and display shortcut hints
  • Desktop main window minimum width reduced to 760 logical pixels, adapting to narrower windows
  • Sidebar drag handle is clearly visible against light backgrounds
  • Dragging the horizontal scrollbar of the result table no longer pops up the SQL hover preview and obscures content
  • Filter input box displays more fully in narrow columns on the Nacos configuration page
  • SQL editor theme styles are reused across tab lifecycles, so restoring cached tabs no longer accumulates duplicate styles
  • Spanish interface fills in 224 missing entries, returning to alignment with English

Fixes

  • When the MySQL server defaults to autocommit=0, writes in the autocommit tab are no longer silently rolled back
  • MySQL query tabs retain session variables, temporary tables, and transaction state after switching databases
  • MySQL large table export immediately stops the server-side query after interruption, no longer stalling the export queue due to result set draining timeout
  • MySQL diagnostics no longer misreport TIMESTAMPDIFF/TIMESTAMPADD time units as unknown fields
  • PostgreSQL foreign table deletion generates DROP FOREIGN TABLE, and deleting foreign tables from the sidebar no longer errors
  • CockroachDB index and column metadata queries use concat(), no longer reporting unknown function: textcat
  • GaussDB column queries use the ::int shorthand, and the sidebar can normally list table fields
  • PostgreSQL view and column comments are fully preserved in single-object DDL and database export
  • openGauss regular account backup filters Schemas by USAGE privilege, no longer failing entirely due to Schemas without privileges
  • Greenplum, Cloudberry, and Hologres disable tuple identifiers, and editing tables without primary keys no longer errors
  • When data grid sorting is rejected by the database, automatically remove row identifiers and retry
  • Kingbase query toolbar no longer mistakenly clears the saved default database
  • DROP INDEX CONCURRENTLY statements are no longer mistakenly marked as syntax errors by the editor
  • Oracle sessions follow the client date language, and Chinese NLS date literals no longer report ORA-01843
  • Oracle semantic diagnostics resolve PUBLIC synonyms, and DUAL etc. are no longer misreported as unknown tables
  • Oracle object browser restores display of comments for tables, views, and materialized views
  • Oracle queries starting with multiple layers of parentheses or leading comments can normally return result rows
  • Oracle and SQL Server synonym DDL and column information resolve to the underlying objects
  • PL/SQL debugging auxiliary identifiers are shortened to within 30 characters under Oracle 11g, restoring debugging functionality
  • SQL Server temporary table completion no longer blocks all metadata requests for the connection due to tempdb directory locks
  • DM instances fall back to JDBC metadata when the ALL_OBJECTS dictionary is missing, and the table list displays normally
  • DM falls back to USER when CURRENT_USER is rejected, and the job page opens normally
  • DM dexp/dimp quotes passwords containing @ or /, and export/import no longer fails to connect
  • ClickHouse queries follow the switch to the new database after executing USE, no longer returning data from the old database
  • InfluxDB reports errors truthfully when statement errors are returned with HTTP 200, no longer showing success with 0 rows
  • Teradata table preview and bounded reads use TOP/ROW_NUMBER pagination, no longer generating LIMIT that reports error 3706
  • Doris and StarRocks tables display table size normally in sidebar size mode
  • In narrow columns of the Redis value panel, the word wrap toggle no longer wraps vertically and covers the controls below
  • MongoDB collection filters automatically recognize the BSON date type, and date range conditions compare by date
  • Nacos service instance filtering changes to filtering loaded instances, and unknown clusters no longer retain the old list
  • After configuring an SSH transport layer for Consul, HTTP requests are correctly routed through the transport layer and no longer fail with direct connections
  • Navicat-compatible HTTP tunnel accepts script versions other than 2.04
  • Manually specified JDBC driver dependencies are isolated from the plugin classpath, no longer triggering NoSuchMethodError
  • MCP database whitelist validation for MySQL static CTE queries no longer misreports cross-database access
  • When high-risk MCP SQL is pending approval, the desktop requests system attention so the confirmation dialog is visible even when the main window is in the background
  • AI features are restored in detached SQL windows, with an AI toggle provided in the window title bar
  • Concurrent creation reusing the same actual connection pool no longer misreports a connection pool publish credential error
  • Saving connection configuration no longer tears down online plugin connection pools and resets connected terminals
  • Creating a new query from a quick connection correctly applies the connection's configured default Schema
  • The preference for encrypting sensitive information in snippet sync is no longer lost after saving and reopening
  • Background export tasks and their logs are restored after an application crash or restart
  • After a manual transaction expires while idle, export gives a clear prompt and cleans up the invalid session
  • Importing a multi-table SQL dump file automatically hands off to SQL file execution, no longer getting stuck with an error in the single-table wizard
  • The footer statistics state ends normally after loading all rows completes, no longer spinning indefinitely
  • Custom column layout is preserved when re-executing after only modifying the outer WHERE condition
  • Batch editing conditions with an empty WHERE condition prompts for input, no longer calculating affected rows across the entire table
  • The export menu displays each export scope such as selection and selected columns in a separate column
  • After searching and filtering in multi-database execution, Select All selects only visible targets, and the selected list is no longer clipped at the bottom
  • After pasting SQL with trailing blank lines, the current statement shortcut can still locate and execute the statement
  • File content search preview restores selection highlighting and supports Ctrl+C copying
  • Very large SQL tab content is stored in external files, no longer causing parsing to freeze on every startup
  • The settings search box no longer triggers browser history and credential autofill
  • After manually installing a stable version, beta version numbers promoted to stable are correctly recognized
  • Windows update cache commit automatically retries when briefly locked by antivirus software, and in-app updates no longer fail
  • Structure view overlays and modal dialogs no longer compete for z-index and make the panel unusable
  • Custom sidebar refresh shortcuts take effect correctly, and refreshing a table node no longer forces expansion
  • Unloaded tables matched by global sidebar search can normally expand column and index groups
  • After a table is rebuilt with a new column order, the old local column order is cleared and no longer stacks incorrectly
  • Restoring an empty draft in the table structure editor no longer shows a blank screen, and metadata read errors remain displayed
  • Drop table operations in schema comparison are unchecked by default and require manual selection to generate DROP TABLE
  • The Transwarp Inceptor dialect is correctly registered, and connections can normally resolve dialect capabilities
  • When plugin capabilities are empty, the encoding is an empty array, and plugin initialization no longer fails