- Rewrite FETCH FIRST to ROWNUM for Oracle 11g (no 12c+ syntax)
- Filter APEX_*, FLOWS_*, $-prefixed system schemas from sidebar
- Set schema prefix on table nodes for schema-aware databases
- Use all_objects instead of all_tables+all_views (4x faster)
- Disable statement cache to avoid stale cursor issues
- Update rust-oracle with streaming TTC, drain fixes
When executing SQL from the AI panel, pass the SQL string directly to
tryExecute() instead of relying on the executableSql computed property,
which could use stale selectedSql or cursorPos values.
Serialize JSON/JSONB as strings in Postgres and MySQL drivers to avoid
repeated JSON.stringify in the render loop. Truncate cell display to
256 chars so the browser doesn't lay out huge text nodes on scroll.
Move refresh/commit/rollback buttons from DataGrid's internal toolbar
to ContentArea's table info header. Remove the redundant second row
and the column count display.
The DANGER_RE regex matched keywords anywhere in the SQL text, causing
SELECT queries using REPLACE() or TRUNCATE() functions to incorrectly
trigger the destructive-SQL warning dialog. Check only the first keyword
of each semicolon-separated statement instead.
Also fix dialog content overflow by adding min-w-0 to the <pre> element,
and move macOS frameworks config from tauri.conf.json to CI-only
TAURI_CONFIG so local `tauri dev` no longer requires libltdl/libodbc
to be globally installed.
Closes#135
- Bundle libodbc.2.dylib into .app/Contents/Frameworks/ via Tauri's
macOS frameworks config to fix "app can't be opened" on macOS
- Replace relaunch() with exit(0) to avoid Tauri restart bug (#11392)
- Lock update dialog during download and after install
- Render release notes as markdown
Execute GET/POST/PUT/DELETE queries like Kibana Dev Tools.
Search results are displayed as table rows with _id + _source fields.
Also hide database selector for Elasticsearch connections.
- Use try_get with i16 fallback to prevent panic on column metadata
- Add N'' prefix for string literals in SQL Server INSERT/UPDATE to fix Chinese character corruption
Extract all inline database type checks (schema-aware, pagination style,
feature support sets, etc.) into a single source of truth. Adding a new
database type now requires updating only one file instead of 15+.
- Add GaussDB/openGauss support via ODBC (resolves#106)
- Use pg_catalog queries instead of information_schema for compatibility
- Share ODBC environment across drivers with auto ODBCSYSINI detection
- Fix ODBC error message false positive on "Driver" keyword match
- Merge connection dialog categories into a single flat list
- Remove duplicate DDL button from data grid footer
Connect to DM8 databases using the odbc-api crate with unixODBC.
Includes schema browsing, query execution, DDL generation,
and full frontend integration with download link to DM ODBC driver.
Use `marked` library to parse LLM output so bold, headings, lists and
other markdown syntax render properly instead of showing raw symbols.
Also fix missing oxfmt darwin-arm64 native binding.
Add create, drop, empty, truncate, and duplicate structure operations
to the sidebar table right-click menu. Also extends the table structure
editor to support create mode from the database/schema context menu.
Add a third transfer mode that inserts new rows and updates existing
rows based on primary key, supporting PostgreSQL, MySQL, SQLite,
SQL Server, Oracle, and DuckDB with database-specific upsert syntax.
Also fixes DuckDB primary key detection in transfer operations.
Move the column type annotation from inline display (#type) to a hover
tooltip on the header cell. The tooltip shows the column type with color
and comment (if present), matching the tab tooltip style.
Add a browse button next to the file path input for SQLite and DuckDB
connections, allowing users to select database files via system dialog
instead of manually typing paths. Clear the default host value when
switching to file-based database types.
Add schema selector to the editor toolbar for schema-aware databases
(PostgreSQL, SQL Server, Oracle). When a schema is selected, the backend
sets the appropriate search_path before executing queries so users no
longer need to write schema-qualified table names.
Switch oracle-rs to t8y2/oracle-rs fork with SYSDBA support.
Add sysdba field to ConnectionConfig and a checkbox in the
Oracle connection dialog to enable connecting as SYS with
SYSDBA privileges.
t8y2