Cargo.toml had no way to tell that a dependency had a newer release.
Dependencies in [dependencies], [dev-dependencies], [build-dependencies], target tables and [workspace.dependencies] now get a weak warning when the newest stable release is outside the version requirement, with a quick fix that rewrites the requirement at its original precision ("0.7" → "0.10") in both string and inline-table forms. Versions complete inside the version string, newest first, without yanked releases. Path, git and workspace-inherited dependencies are skipped, prereleases are only suggested to prerelease requirements, and failures stay silent.
Versions come from the crates.io sparse index through the backend's HTTP client, with an in-memory and on-disk cache and coalesced requests, behind a dependency version source shared by file-type parsers so other manifests can be added later. semver is now used directly; it was already linked into the app.
Fixes#479
* fix(windows): keep maximized windows within the work area
* fix(windows): constrain final maximized outer placement
* Revert "fix(windows): constrain final maximized outer placement"
This reverts commit e481c900ca.
Inline debugger values looked up source identifiers in a plain object, so a name like toString on a nearby line resolved to an inherited Object.prototype member and crashed the editor's inline-values plugin. The execution-point subscription lived in that plugin, so after the crash the highlighted line stayed on the old position while the stack had moved on, and resume or stop did not clear it.
Locals are now kept in a prototype-free dictionary and looked up by own property, and the execution point is synced by its own plugin so an inline-value failure can't freeze it; hidden editors resync when remounted. Stepping, resume, run to cursor, reset frame and smart step clear the old execution point as soon as the request is sent, stale stack/locals responses from a previous stop are ignored, and another thread's continued event no longer clears the selected thread's execution point.
Fixes#668
CMake run configurations used fixed arguments (-B build, Debug) and only recognised existing build directories.
Projects can now define CMake profiles under Settings › Build, Execution, Deployment › CMake: build type, generator, CMake arguments, build directory (default cmake-build-<type>), build arguments and environment. Each profile gets its own Configure, Build, CTest and target run configurations, named after the profile when there are several. Configure runs first only while the build directory has no CMakeCache.txt, and a generator change clears the stale cache before configuring. Projects without profiles keep the previous detection. Arguments are stored as lists and edited with the same quoting rules as program arguments. Toolchain selection is not part of this change.
Refs #655
Java formatting comes from jdtls' Eclipse formatter, whose defaults differ from the IntelliJ IDEA code style. Settings › Editor › Code Style › Java now offers a formatting profile: Default (unchanged), IntelliJ IDEA, or a custom Eclipse formatter XML (absolute or project-relative path).
The IntelliJ IDEA profile is a bundled Eclipse formatter XML that follows IDEA's default Java code style item by item (indents, aligned multi-line parameters, annotation placement, else/catch/finally on the same line, array initializers, lambdas, blank lines, comments not reflowed). Manual line breaks are kept; lines longer than 120 columns wrap where necessary, which differs from IDEA's default of not wrapping. The profile reaches a running jdtls through didChangeConfiguration without a restart; an unreadable custom file shows a warning and leaves code intelligence running. Code style changes reconfigure language servers without scheduling Kotlin bridge builds.
Also drops two catalog entries whose texts are no longer used.
Fixes#649
On Linux the window-state plugin restored the saved size before GTK's decorations and output scale had settled, so the saved physical inner size went through the creation-time CSD/scale conversion and drifted on every restart (52 px per axis under Mutter, the remembered size collapsing on other setups), and maximize transitions overwrote the remembered normal size.
Linux now builds project and settings windows hidden and restores them after the first configure, keeping the normal size apart from maximize, fullscreen and minimize, protecting it during transitions, and snapshotting each window before close instead of querying GTK during exit. The state file format is unchanged; macOS and Windows keep the plugin as before.
Verified under Mutter and Weston: three restarts at 1234×678, three maximized restarts followed by unmaximize, the settings window, several project windows across a full exit, and window groups all restore exactly. macOS restore is unchanged. Not run: GNOME on real hardware, mixed-DPI monitors, X11.
Fixes#594Fixes#636
clippy::items_after_test_module failed the workspace clippy run because run_input was defined after the first test module in crates/ide-tools/src/process.rs. Move it above; no code changes.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
While a debug session is paused, Alt/Option+click on a variable or expression runs Quick Evaluate at the clicked position (the selection when the click falls inside it) and shows the value next to it; outside a paused session Alt+click keeps its editor behaviour. The editor context menu offers Evaluate Expression… seeded with the selection or the expression at the caret while paused.
Mouse shortcuts now pass the clicked editor and position to the action, and the binding is only consumed when its action is enabled.
Fixes#664
AppKit puts the title bar container and window buttons back at their defaults on every live resize step, title change and fullscreen exit. They were corrected from window notifications after that layout had been drawn, so the buttons bounced between positions, or stayed at the defaults when no notification followed. The frames are now corrected in a main run loop observer ordered just before the Core Animation commit, project windows are created hidden on macOS so the buttons are placed before the first frame, and repeated title bar height updates are skipped.
Choosing This Window for a local project in a remote window was routed to the remote host or skipped when the path matched the remote root. open_project and light_edit_open now run on the local backend and leave remote mode once they succeed, project trust for these opens is checked locally, and a failed open cancels the pending close.
Reformat Code only asked the language server for formatting, so files whose server has no formatting capability (Markdown with marksman, and others depending on the server) could not be formatted.
Languages can now declare an external formatter. Prettier is the first: a managed tool installed on demand, or the project's own node_modules/prettier, an explicit path, or PATH. It runs on the backend copy of the document with the file's path so Prettier resolves the project's config, overrides and ignore files, and its result is applied as one undoable edit. Language servers that format a file keep doing so unless Prettier is preferred in Settings › Languages › Prettier; formatting on save is opt-in. Parsers without range support format the whole file and say so.
Errors carry structured tool data, so a missing Prettier reuses the shared install prompt and Prettier diagnostics or timeouts show as non-sticky notifications.
Fixes#639
Java files in mixed Maven projects could not resolve Kotlin types from the same module or upstream reactor modules, because Kotlin compiler output never reached JDT's Java model.
A per-project bridge compiles saved Kotlin sources with the project's own Maven setup into private staging output (a bundled Maven extension redirects outputs and records each module's Kotlin compile result) and mounts the Kotlin classes as marked libraries through a bundled JDT command, leaving m2e/JRE containers and other entries untouched. The last successful snapshot stays mounted while rebuilding; a new one is published when every Kotlin target compiles, even if javac fails, so Java callers of a renamed or removed Kotlin API show their errors in the editor. Kotlin compile failures keep the previous API and are reported under the kotlin-bridge Problems source. Removing the last Kotlin file withdraws the libraries. Changes under build outputs and excluded directories never trigger a rebuild.
Part of #428
The editor's mousedown handler ignored every button except the left one, so a middle click fell through to the webview's autoscroll. A plain middle click now moves the caret to the clicked symbol and runs go-to-definition, suppressing autoscroll; a keymap shortcut bound to Button2 still takes precedence, and modifier + middle combinations stay unbound.
Fixes#663
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
java_class_file joins the package path with std::path, so on Windows the result uses backslashes while the test's expectation kept the literal forward slash from "p/Outer.java". Compare as Path so separator style does not matter.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
Run/Debug of a Maven module with Kotlin/Groovy/Scala sources rebuilt the module and its reactor dependencies with Maven before every launch, even when nothing had changed.
Each successful launch build now records a fingerprint of its inputs (sources and resource roots of the dependency closure, module and parent POMs, imported BOMs, wrapper/.mvn/settings/toolchains, JDK, command line and environment) together with an immutable snapshot of its outputs in the project cache. A later launch with the same fingerprint reuses the snapshot without running Maven; Run and Debug share it, application and test launches are kept apart. Inputs that cannot be resolved reliably (dynamic roots overridden by profiles or -D) always rebuild. A failed, cancelled or concurrently edited build never publishes a reusable record.
Superseded snapshots stay available while JVMs of the current session may still load classes and are removed when the project closes; snapshots left by earlier sessions and no longer referenced are cleaned on the first launch.
Fixes#530
Search Everywhere and Recent Files lists now support multi-selection: Shift+Up/Down and Shift+click extend a range from the anchor, Cmd/Ctrl+click toggles a row, and Enter opens every selected file, symbol or text result in list order, leaving focus on the lead row. Actions in a mixed selection are skipped; a single selection behaves as before. Selection resets when the query or tab changes.
The row-selection model used by Find in Files moves to src/ui so the shared List can opt into modifier selection.
Fixes#647
Adds a UI fallback font under Settings › Appearance › Use custom font, so glyphs missing from the chosen UI font (e.g. CJK in Menlo) render in a font the user picks instead of the system default. The fallback joins the UI font stack used by menus, the project tree and dialogs, and resolves macOS user fonts through the shared font loader.
Also keeps the custom font controls wrapping together at narrow widths and covers save/clear/disable, stack composition and settings round-trip in tests.
Fixes#654
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
In files no language server covers (Markdown, .properties, plain text), Shift+F6 only showed a transient "no running language server supports rename" hint at the caret, so the default rename shortcut looked broken.
Root cause: when prepare_rename reported no server and the syntax fallback declined, startRename fell through to the error hint. It now hands off to project.rename, which opens the file rename dialog for the active file; a running server that lacks rename keeps its hint.
Verified on macOS: Shift+F6 in .md and .properties editors renames the file, symbol rename in TypeScript is unchanged, and Shift+F6 in the project view still renames the selected file.
Fixes#651
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
Run/Debug of a Java-only module that depends on a Kotlin/Groovy/Scala module in the same reactor launched with stale classes or failed with Unresolved compilation problem, and a successful Maven build was still reported with the old JDT compile errors.
Root cause: the pre-launch external build only checked the launch module's own sources, and Maven shared target/classes with JDT autobuild.
Fix: walk the launch module's reactor dependency closure (transitive, test scope aware) for non-Java sources; when found, run Maven from the reactor root (-pl <module> -am clean compile, test-compile for test launches) and block the launch on failure or cancel with the Maven errors in the console. JDT autobuild is paused while Maven runs and restored afterwards, and the launch uses a per-launch snapshot of the closure's output so JDT cannot overwrite it. A successful Maven build no longer shows the stale JDT warning. Pure Java projects and Gradle keep their previous path; the debugger adapter choice still follows the launch module.
Verified on macOS with single- and multi-module Maven projects (new Kotlin values in Run and Debug, Kotlin error blocks the launch, cancel, pure Java does not invoke Maven). Editor type resolution for Kotlin in Java is handled separately.
Part of #428
Two folders can now be compared from the project view (Compare Directories with two folders selected, or Compare With… on one folder, also in Find Action). The result tree marks entries as only on left, only on right, contents differ or identical, with a name filter and a show-identical toggle; double-click opens the existing diff editor, and files or whole subdirectories can be copied to either side (overwrite and delete ask first, extra files in the target are kept). Comparison runs in the background with progress and cancel, honours the project's exclusions, and works for remote projects through the same file commands.
The comparison tree and diff preview are shared with the Git folder/revision comparison.
Verified on macOS (subdirectories, Chinese names, binary files, copy both ways, delete, cancel on large files); Windows and Linux not run.
Fixes#520
Opening a .class file showed binary content. Class files from library jars, project output directories, navigation, Go to Class and debugger frames now all go through one route to JDT class documents:
- Library classes open the attached source, or the FernFlower decompilation when there is none, read-only and marked as decompiled. java-debug's decompiled-source support is enabled, so breakpoints set in decompiled code map to the original lines and hit, and stepping into such classes lands on the right line.
- Output-directory classes (target/classes, build/classes) open their project source when it exists; otherwise the decompiled text is shown read-only with breakpoints disabled, because jdt.ls returns no line mapping for classes outside the Java model.
- The preferred content provider id is corrected to fernflowerContentProvider (attached sources still win), and the decompiled marker recognises the current FernFlower header.
Verified on macOS with a Maven project: library class from Libraries and Go to Class, breakpoint hit and step into decompiled code, output class with and without source. Not run on Windows, Linux or Gradle projects.
Fixes#394
Help › About showed no version, build or runtime in untrusted project windows and on the welcome screen.
Root cause: the restricted-mode allowlist added with project trust missed app_build_info and other application-level commands, so they were rejected wherever the window had no trusted project; the frontend then cached the failed build info for the whole session.
Fix: classify all registered commands and allow the 24 that only read application information or change the user's own application state (build info, log location, update checks, cache invalidation, file clipboard, JDK download, Docker settings, database export paths, patch export, system environment); JDK download now uses the application tool directory instead of project tool settings. Execution commands (run, build, debug, language servers, tool install from project settings) stay blocked. Build info is retried after a failure instead of being cached. Verified on macOS in an untrusted project and on the welcome screen; Windows not run.
Related to #519
With Transparent native-to-ascii conversion on, pasting or typing existing \uXXXX escapes into a .properties file and saving wrote them back as \u005CuXXXX, so the file on disk changed meaning.
Root cause: the shared properties encoder escaped the backslash of any valid ASCII Unicode escape. It now writes ASCII escapes as they are and only converts non-ASCII characters; a literal backslash right before a non-ASCII character is still protected. The same encoder serves every write path, so normal/auto save, find and replace in open and closed files, workspace edits, Local History restore and partial Git stage/commit are all covered.
Verified on macOS following the reporter's steps (paste escapes, save, reload externally); Windows and Linux not run.
Fixes#519
New terminals typed the virtualenv activation command for the wrong shell: on Windows the default terminal is PowerShell, but the activation line was chosen from $COMSPEC (cmd), so PowerShell just echoed "...\activate.bat" and the project interpreter was not activated.
Root cause: terminal_activation guessed the shell from $SHELL / $COMSPEC instead of the shell the PTY actually starts. Same class on macOS/Linux when the inherited $SHELL differs from the login shell the terminal uses, and for a blank configured shell.
Fix: the host resolves the shell once with the terminal's own resolve_shell and uses it both to start the PTY and to pick the activation command; terminal_activation takes the resolved shell and no longer reads the environment. Tests cover PowerShell/cmd paths, blank settings and a host-level run with an unrelated $SHELL. Verified in the app on macOS (default zsh and configured bash); not run on Windows.
Fixes#631
On macOS versions where WKWebView refuses user-installed fonts by family name (fonts in ~/Library/Fonts or /Library/Fonts), the editor, terminal, Markdown and UI fell back to the default font and Nerd Font icons showed as boxes; only a webfont loaded from the file's bytes renders.
Root cause: the shared font resolver only ever referred to fonts by local family name, so every consumer depended on the WebView resolving user fonts by name. A URL-based webfont is blocked by the existing font-src policy, which stays unchanged.
Fix: the shared resolver probes whether a family renders by name (on a document-attached canvas; a detached canvas falls back even when the font works) and only when it does not, fetches the enumerated user font files over a raw-bytes IPC (user fonts only, looked up by family, no paths accepted) and registers them with the FontFace API under a stable alias. Only the faces nearest regular/bold and their italics are loaded, once per family. Covers editor and editor fallback, diff/merge views, terminal font, fallback and built-in Nerd fallback (re-measured after loading), embedded terminal, settings preview, custom UI font, Markdown and Mermaid. A debug-only environment switch reproduces the WebView restriction for testing.
Verified on macOS 26 with the restriction simulated (icons boxes before, correct after) and without it (no font bytes transferred, no regression). Not run on the reporter's macOS version, Windows or Linux; other platforms keep the previous behaviour.
Fixes#592
lombok.config was ignored in Java builds (e.g. lombok.accessors.chain=true produced void setters), and jdt.ls wrote those classes into target/classes, breaking later Maven builds.
Root cause: Lombok calls getModuleClassLoader(boolean), which newer Equinox removed; the exception is swallowed, Lombok resolves the wrong source path and finds no lombok.config at all, so every config key falls back to its default. Affects every Java project using lombok.config.
Fix: ship Lombok 1.18.48 with the upstream fix projectlombok/lombok#4059 (lombok-1.18.48-openidea.1, embedded in the app, no download) as the jdt.ls agent; the versioned path also resets the jdt.ls workspace so stale output is rebuilt. A user-configured agent still takes precedence. Return to Maven Central once an official release includes the fix (source and checksum in plugins/lombok/README.md).
Verified on macOS: 5 clean builds with chained setters, identical output to official 1.18.48 without lombok.config, the app rebuilds already-poisoned target/classes, and Run works. Windows and Linux not run.
Fixes#565
Counts on the tool window stripe icons (changed files on Commit, problems, agent tasks needing input, running processes) can be turned off in Settings › Appearance & Behavior › Appearance › Tool Windows. On by default; the setting is stored in appearance settings and survives restarts.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
Agent sessions can be opened in the Agents tool window beside the editor (Open in Tool Window / Open in Editor in the task menus, or by dragging between the editor and the tool window); opening in an editor tab stays the default.
Fixes in review:
- A docked conversation kept the previous task's subagent view after switching tasks: the session's local state is now keyed by task.
- Moving a conversation between the editor and the tool window lost its scroll position: position and follow-output are kept per conversation (task or subagent).
- In a short bottom tool window the message area shrank to a few pixels: wide windows place the task list beside the conversation, narrow ones cap the list height.
- Drag and drop and list selection now go through the agents.dockTask / undockTask / openTask actions.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
applyCodeAction now resolves the picked action and executes it (lspResolveCodeAction → lspExecuteCodeAction) instead of calling lspApplyCodeAction, but refactorThis.test.ts and runKind.test.ts still mocked and asserted the old command, so main's frontend tests failed (4 tests). The mocks now follow the new commands; safeDelete.test.ts's negative assertion still holds.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
Go's test generation and interface stubs were reachable only through context actions, because Generate filtered their titles, and implementing an interface on a plain struct wasn't available since the client didn't support gopls's interactive commands.
Generate and the editor's context menu now list `Add test for ...`, `Declare missing methods ...` and `Add methods to ... implement ...`, and a generated test file opens in the editor. For gopls only, the client advertises `interactiveResolve` and runs its `command/resolve` handshake: a prompt titled from the server asks for the interface, shows validation errors in place, and Cancel writes nothing. Requests are cancellable and other servers and existing commands are unchanged.
Fixes#620
* fix(editor): pause expensive decorations during fast scroll bursts
Add a scrollThrottle plugin that flags the editor with a .cm-scrolling
class while the wheel is moving. While the flag is up the sticky-scroll
header and indent guides are hidden via theme CSS, and their view
plugins skip per-viewport re-measures/rebuilds; a trailing scrollSettled
effect re-runs them once the wheel rests for 140ms. This keeps fast
mouse-wheel scrolling from painting rows of stale decorations over
freshly rendered code.
Fixesl0ng-ai/openidea#375
* fix(editor): keep decorations painted during scroll bursts
The scroll throttle hid the sticky header and indent guides entirely while
scrolling, which flickered on every burst - and the sticky header is exactly
what you want to see while scrolling. Drop the display/background hiding from
the theme and keep the previous paint instead: the plugins skip re-measuring
during a burst and catch up once on the trailing scrollSettled update.
* fix(editor): throttle the first scroll event and queued sticky measurements
---------
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
While a new query was debouncing or validating, Find and Replace in Files kept showing the previous query's rows, count and highlights, and Replace stayed enabled. Replace, Replace All in This File and Replace All then sent the new pattern with the old positions, so `order` changed to `ord` could rewrite `order` as `Xer`.
From the moment query, flags, mask, scope or context change, old rows are dimmed and lose their highlights, replacement preview and count until the new search's first results arrive. All replace buttons and keyboard paths are blocked meanwhile, and a pending Replace All confirmation rechecks that its results are still current. Opening a retained row still works.
Refs #623
`.svelte` files got highlighting only from semantic tokens, because the language had no `ui-mode` and no CodeMirror language loaded. It now uses the HTML mode, which parses `<script lang="ts">` as TypeScript and `<style>` as CSS. Svelte blocks and markup expressions stay plain text.
Closes#626
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
Clipboard file-list reads and writes ran on background threads, while AppKit and WebKit use the general pasteboard on the main thread. Its type cache isn't safe against that, so the app crashed with a segfault or heap corruption in `-[NSPasteboard _updateTypeCacheIfNeeded]`. Rechecking the clipboard on window focus made this frequent, sometimes within seconds of opening a project.
On macOS these commands now run on the main thread and await the result without blocking a command thread; other platforms are unchanged. `ide-clipboard` documents and debug-asserts the requirement. A concurrent read stress test that crashed main within a second completed about 650,000 reads over four minutes with no crash.
The Replace in Files footer has more actions than fit beside its keyboard hint, so buttons shrank below their labels, overlapped, and ran past the dialog edge. Buttons and the Preview checkbox now keep their size, the hint truncates first, and the actions wrap to a second row aligned right. The results and preview areas shrink when the footer wraps or the window is short. Find in Files shares the footer.
Find Usages mixed reads and writes of a Java field, including Lombok getter and setter calls, because jdtls has no read/write classification in its references. The usages popup and Find window now have Read and Write toggles. Read-write accesses stay visible in either mode and unclassified results are kept.
Classification comes from a jdtls extension in `plugins/java-usages`, embedded as a jar and installed beside the other bundles. The checked-in jar was rebuilt from source with ECJ and matches byte for byte. Any extension failure falls back to standard references, so a jdtls upgrade can lose the filter but not Find Usages; rebuilding the jar on an upgrade is manual and described in the plugin README. Older installed jars are removed. Other languages keep their plain result count.
Co-authored-by: NingYuanLin <57001533+NingYuanLin@users.noreply.github.com>
Running Spring Boot main-class configurations had no Update action, and incremental Java builds did not publish edited resources to the running JVM. They now share the Update action, toolbar and build runner with Tomcat: resources go through the module's Maven or Gradle build and are published to the resolved runtime classpath, including Gradle's separate jdt.ls output directory. Class updates use scoped HotSwap when available and explain the restart requirement otherwise.
Both configuration types get "On 'Update' action" and "On frame deactivation" policies. A missing policy keeps the chooser and does no deactivation update. Focus tracking debounces external changes and prevents overlapping updates.
Verified on macOS with Maven, Gradle, Tomcat 10.1 and HotSwap. Windows and Linux focus detection were not run. On Linux a native GTK dialog can count as deactivation when no webview window is focused.
Fixes#580
Project Paste could reuse an earlier file copy after another app replaced the OS clipboard, and Cut files stayed dimmed. The tree now tracks the hand-off to the OS clipboard and rechecks it on window focus and just before Paste. A local Copy or Cut is cleared once the OS holds a different or empty file list; failed writes and clipboard read errors keep it. The native adapters now distinguish read failures from empty lists.
In remote windows, Copy and Cut write newline-joined server paths as plain text, tree Paste keeps the server source, and a later OS file list replaces the remote copy and imports those client files. A plain-text replacement stays undetectable there without a clipboard read that may prompt.
Verified on macOS, locally and against a remote host. The Windows clipboard code was not run.
Dropping a file on a terminal fell through to the window-wide drop, which opened it in the editor or asked to open a folder as a project. It now pastes the shell-quoted paths at the prompt, separated by spaces and with no newline, so nothing runs until the user presses Enter.
Each terminal, including the Run console, accepts only drops inside its own host. Windows paths use double quotes. Remote windows keep the existing drop, since a local path means nothing on the remote host. Verified on macOS with a simulated drop; a native Finder drag and Windows quoting were not run.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
Find in Files, Replace in Files and Search Everywhere have no window title bar, so they could not be moved off the editor underneath. Dialogs can now be dragged from any element marked `data-dialog-drag`: the headers of those search popups, the title of any titled popup, and the existing window title bars and alert bodies, which now use the same mark. Controls inside a handle never start a drag.
Also waits for a review thread inlay in a test where it can mount a tick later.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
Finding a declaration in a file meant opening the Structure view. A new Project view option, Show Members, lets files expand into their declarations, including nested ones, and clicking or activating a member opens the file at that symbol. It is off by default.
Outlines load when a file expands, including expansion restored on reopening a project or by turning the option on later. They refresh on save and on changes on disk, with stale in-flight results discarded. Files with no outline lose their chevron. File operations, drag and the context menu skip member rows. Also removes a duplicate Sort Folders First menu entry.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
The landing's three actions were flex items that hugged their labels, so the gap between icons followed the text length. They are now equal-width grid columns, which spaces the icons evenly in every locale.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
`.rs`, `Cargo.toml` and `Cargo.lock` used custom badges or the generic lock, and `.gitignore` used the git logo. They now have dedicated Rust, Cargo and Cargo lock icons with dark variants, and `.gitignore` and `.gitignore_global` share the ignored-file icon with the other ignore files. The icons' source and MIT licence are recorded in `THIRD_PARTY_LICENSES.md`.
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
The composers' primary actions were text buttons. Start, Send and Queue, plus Send in edit-and-resend, now share an icon-only submit button that keeps the translated name as its accessible label and tooltip, and shows a spinner while the action is pending. Stop is unchanged.
l0ng-ai
fuqiuluo
arookieofcanda-rookie-of-C-language
Ning YuanLin