* feat(run): show .NET test results and rerun failed methods
* feat(dotnet): add workspace, NuGet and editor test workflows
* fix(dotnet): discard cancelled queries and scan test targets linearly
Plugin API 1.4: grants for unencrypted HTTP, local sockets, IDE commands and owned database sources; window status items, output channels, content tabs, editor selections, decorations, search, Problems and VCS reads; runs, keyed terminals with exit events, plugin run targets, dynamic run configurations and test parsers; language features, injections, snippets and menu locations; plugin MCP tools on the built-in server; four official API examples. Released 1.3 plugins keep working.
Detected Go run, test and debug configurations ignored the build tags set for the project, so tagged files were left out. They now inherit the project's tags (unless the configuration opts out), including reruns, file-scoped tests and debug launches; maintenance commands keep running without them.
Co-authored-by: shohruh-genius <[email protected]>
Go test reruns now keep full subtest paths (each segment anchored) and run each package with its own selection; tests and benchmarks use their own filters (-run / -bench with -run '^$'), and stale tool or test-binary filters can no longer override reruns or gutter launches. File configurations keep their file target and original run configuration. Boolean test flag aliases are parsed before package targets.
Co-authored-by: shohruh-genius <[email protected]>
Pest test() and it() calls and describe() blocks get gutter Run and Debug,
nested by describe and filtered by the names Pest prints (`outer` → it does
x, plus a data set suffix). Projects with vendor/bin/pest get a Pest test
configuration of the PHPUnit kind, pest commands are recognised, Pest's
pest_qn:// hints are read and test tree reruns filter Pest tests by name.
Quiet Go benchmarks print their measurement without a terminal pass event, so they stayed running and were shown as ignored when the run ended; repeated attempts (-count, -cpu) share a test path and were counted once. The parser now completes measured benchmarks when their package finishes (a failure at a later CPU setting still wins), maps CPU-suffixed and nested benchmark names back to the started test, and counts each new attempt while keeping a failed one failed.
Co-authored-by: shohruh-genius <[email protected]>
Parsing Minitest's console output needs -v, gives no file or line for passing tests and depends on how output interleaves. Ruby test runs (ruby and rake test configurations in the console) now load a small Minitest reporter through RUBYLIB/RUBYOPT that prints one record per test on stdout with its status, time, failure text and real file and line. The console strips the records and the test tree is built from them, so jump to source gets exact locations; the text parser stays as the fallback. It works on Minitest 5 and 6 (6 no longer loads plugins on its own, so a small hook registers the reporter once Minitest is loaded).
Co-authored-by: kjeldahl <[email protected]>
Ruby test runs showed only console text: there was no Ruby output parser and test_*.rb was not a test file. A Minitest parser now builds the test tree (class, test, status, time, failure text) from verbose output (-v / TESTOPTS=-v), test_*.rb counts as a test file, jump to source finds the class and then its def, and the gutter gets a run action per test method (coloured by the last result) and per test class and module that holds test classes; one test, class or module runs with --name=....
The streaming Find in Files helper the jump to source uses moved from the search feature to src/platform, so the run feature doesn't import another feature.
Co-authored-by: kjeldahl <[email protected]>
TestNG's class-level @Test makes the class's public void methods tests; the scanner now recognizes them (excluding @Before*/@After*/@DataProvider/@Factory). Test results locate TestNG failures past org.testng frames and parse TestNG's comparison message. Refs #1368
Gradle prints only failed tests, so the live test tree only filled in
after the run from the XML reports, and stayed empty when the test task
was UP-TO-DATE and wrote no new reports.
Gradle test runs now get an init script that reports each test's start,
output, failure (with the stack trace), skip and finish as TeamCity
service messages and keeps test tasks from being UP-TO-DATE. The console
hides those lines; the JVM parser turns them into `[class, method]`
results, named as in the XML reports, and drops Gradle's own repeated
`Class > test() FAILED` lines. Tests the run already reported keep their
output instead of getting the report's copy appended. The TeamCity
message parsing moves out of the PHPUnit parser so both share it.
Fixes#1268
Gradle's console names a test class by its simple name and a JUnit 5
method by its display name (`method()`), while its XML reports use the
qualified class and the method name. After a run the test tree showed
each class twice, counted the failure twice, and the report's copy of
the failing test had no source location.
Console results are now moved under the reported class (when the simple
name is unambiguous), Gradle method names drop the empty `()`, and the
report's stack frame is matched for Kotlin names with spaces and frames
behind a class loader prefix (`app//`).
Composer projects and PHPUnit suites are detected (Composer scripts, PHPUnit, artisan serve); PHPUnit runs report into the test tree through --teamcity with gutter runs for classes and methods, file-level runs, rerun failed and a PHPUnit run configuration type. Settings › Languages & Frameworks › PHP selects the CLI interpreter (discovered on PATH, Homebrew, phpbrew, phpenv, Herd, XAMPP, Laragon); it is applied to run configurations and PHP language servers through PATH. TeamCity service messages feed only the test tree and are kept out of the run console, sharing the stdout decoder go test -json uses.
Co-authored-by: ravuthz <[email protected]>
Django projects get a detected `manage.py test` configuration next to runserver and a Django tests configuration type (command, runner arguments, target, settings module, options). Results stream into the test tree with failures, errors, skips and subtests, failed tests navigate to their source line, Rerun Failed Tests narrows the command to the failed labels, and single tests run from the gutter.
Co-authored-by: andreabasile <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Go settings, execution, debugging and test reporting were disconnected, leaving Go projects without editable build/test fields and workspace maintenance actions.
- Add project build constraints that update gopls in place, and notices on excluded files.
- Add editable Go Build and Go Test configurations with directory, package and file targets, separate tool and program arguments, modules, environment, output selection and build-only execution, and pass the same settings to Delve.
- Decode Go JSON test output into the results view, with subtests, failures, skips and rerun failed. Add gutter targets for benchmarks and generate directives.
- Add module environment and vendoring settings, read go.work members, and offer tidy, download, sync and use from file menus.
- Add file and package generate actions, and format and organize imports through gopls on Save and Reformat.
- Show what a Go program prints in the debug console: Delve's debuggee shares its stdio, which no DAP event carries.
- Keep the default build output out of version control, and keep configurations produced by an earlier run out of the gutter choices.
Fixes#507Fixes#237
Adds knip (pnpm knip, knip.ts covering src, scripts, perf and e2e) and runs it in the CI frontend job. Removes what it reported: the unused codemirror and @types/dompurify dependencies, exports and test hooks used nowhere (AppLogo, Tooltip, barrel re-exports, duplicate default exports of lazily loaded settings pages), the unused iterateAll helper, and the zh entry for a deleted notification. rustfmt (used by lint-staged) is ignored as a toolchain binary.
Co-authored-by: jaw52 <[email protected]>
Run oxfmt once over the whole repository (formatting only, no logic changes) and add `pnpm format:check` to the frontend CI job so unformatted code no longer builds up. `.oxfmtrc.json` now skips the generated theme docs, test fixtures and code blocks inside Markdown.
Co-authored-by: jaw52 <[email protected]>
* feat: Second pass aligning interactions with IntelliJ IDEA
Result tool windows
- Next/Previous Occurrence (⌥⌘↓/↑) across Find, Usages, Problems, TODO,
Inspection and Hierarchy, with an "N of M" status hint
- Find: reuse the unpinned tab, Pin Tab, source preview, group by file or
directory, exclude moves to the next result, Rerun
- Usages: tabs with pinning, exclude and preview
- Problems, TODO, Inspection: no autoscroll by default, severity-first sorting,
context menu with quick fixes and copy description
- Trees: F4/⌘↓ jump to source; speed search Home/End and arrow handling
Find in Files, bookmarks, navigation
- Replace mode: Enter replaces the selection, ⇧⌥Enter replaces all, multi-select
- Pin Window toggle, highlighted read-only preview, search/replace history (⌥↓),
Alt mnemonics, search context filter (comments / string literals)
- Mnemonic bookmarks (⌥F3, ⌃⇧digit, ⌃digit), ⌘F3 bookmarks popup, descriptions,
⌘-click gutter toggles a bookmark
- Select In popup (⌥F1); Recent Locations double-click and Delete;
Search Everywhere remembers the query per tab
Code insight and refactoring
- Extract/Introduce continue with in-place rename and name suggestions;
expression and occurrence choosers with live highlighting
- Safe Delete confirms first; numbered Refactor This
- Quick Documentation history, in-popup links and a Documentation tool window;
Quick Definition switches between targets and shows the whole declaration
- Goto chooser lists symbol names and can pin to the Find window
- Parameter Info Tab/⇧Tab, intention preview, Reformat dialog and summary,
inlay hint ⌘-click and context menu
- Live templates: Enter moves to the next field, two-step Esc, ⌘J as a lookup,
language surrounders with mnemonics
- File Structure popup as a filterable tree; Show Usages filter and preview;
call hierarchy jumps to call sites
Editor
- Paste re-indents each line by default; Move Statement moves whole statements
in and out of blocks and re-indents
- Type over closing quotes of existing strings; ⌘D/⌘⌫ cover folded regions;
⌥⌫ deletes the preceding word
- ⌃G / ⌃⌘G whole-word rules and wrap hint; double-modifier caret cloning
- Visual-column block selection, nested Collapse All, recursive folding,
fold selection toggle, Go to Matching Brace, gutter drag selects lines,
sticky line and error stripe clicks, Insert toggles overwrite mode
Run, VCS, workbench
- Gutter icons show the last test result; Modify Run Configuration; test tree
hides passed tests, context menu and assertion diff
- Stack frames resolve by fully qualified class name; context run chooser
- Inline Smart Step Into
- Git Log multi-select filters, "me" and date range; favourite branches,
Fetch and tags in the branch popup; stash Keep index and Unstash options
- Terminal tab titles follow the shell; Distraction Free persists; gear menu;
sticky balloons move to the notifications tool window after a while
* fix(vcs): Label restored local changes as yours after Smart Checkout; name the branch checked out
Conflicts from restoring stashed changes show the user's uncommitted
changes on the left like other operations, and the checkout notification
names the new branch instead of the previous one.
* fix: Close the merge editor on Apply; confirm before terminating a running Run tab
Like IDEA, applying a merge result closes its tab, and closing a Run tab
whose process still runs asks before terminating it (with a Settings
toggle and "Do not ask again").
* fix(terminal): Don't start a new session in the background after closing the last tab
The hidden Terminal window stays mounted, so it reopened a session as soon
as the last tab closed. It now opens the Local tab only once the window is
shown again, as IDEA does.
* fix(vcs): Start the merge result from a three-way merge; close the editor on Apply
The result pane now begins with non-conflicting changes merged and each
conflict showing the base text, resolved from the side panes, instead of
raw conflict markers. Apply now reports success, so the tab closes.
fix(search): Match File Structure filter on word starts
* fix: Limit Smart Step Into to the paused line; drop usages hints on callbacks; show full intention titles
Adapters that list every remaining call in the function (debugpy) made Smart
Step Into fall back to a chooser that offered calls from later lines. Targets
are now narrowed to the calls on the paused line, so they are picked in place.
Anonymous functions and callbacks no longer get a 'no usages' hint, and
intention titles are no longer cut short when there is room for them.
* fix: Undo / Redo from the Edit menu and Find Action use the editor's history
They ran the webview's native undo, which misses programmatic edits such as
templates, completion and refactorings, so nothing happened in the editor.
l0ng-ai
Shohruh Ismoilovтаshohruh-genius
Pedro Pessoa
Jacob Kjeldahlтаkjeldahl
Ravuthzтаravuthz
Jaw