Flatten the agent conversation into a dense tool-window layout and show command approvals as an inline strip with the request's own command (shell wrapper unwrapped), standard buttons and no horizontal overflow.
views.open-log-tab(key, title, process-log) (SDK host::open_log_tab) opens the read-only log tab a web view opens with open-log-tab, from a command or event of a window, reusing the Services pane's process-log.
The host checks the process grant and limits when called and remembers what each key runs in that window; the tab starts it by key (plugin_log_tab_spawn) through the web-view PTY checks: pinned program, trusted project, session cap, lifecycle and grant revocation. Remote windows report that these tabs aren't available there yet.
Refs #1907
Add Python Interpreter → Existing environment: Browse picks the environment folder, which is resolved right away (pyvenv.cfg, bin/python or Scripts\python.exe) and added to the interpreter list as "Python X.Y (<env name>)", then remembered. Picking the folder rather than the executable keeps the environment's identity, since the native file picker resolves the venv's python symlink to the base interpreter. Works for `python -m venv` and uv environments; a broken environment shows an error instead of being added. Probing runs off the command thread.
Co-authored-by: Inf1mum <[email protected]>
Closes#1720
Plugins can declare searchable multi-page settings (API 1.3): pages under the plugin in Settings, groups, string lists, editable tables and command buttons, while single-schema plugins keep working. A command button applies pending edits first and runs against the selected Global/Project layer (also remotely); Reset keeps other pages' edits, and invalid table/list edits never replace saved rows. Field and page order come from the schema without changing serde_json's map ordering elsewhere in the app.
Co-authored-by: arnaud-ritti <[email protected]>
Related #1947
Uninstall keeps a plugin registered but stopped until its private data, secrets and package are cleaned up; a failure is shown and Uninstall can be retried, also after a restart (a durable cleanup record outside the package restores the target). While cleanup is pending, enable, reload, consent, permission changes, reinstall, replacement and rollback are refused. Secret-deletion errors are reported, and cleanup never follows a symlinked ancestor outside plugin storage. The SDK mock checks private-path components and rejects traversal independently of filesystem grants.
Co-authored-by: arnaud-ritti <[email protected]>
A plugin web view can open an editor tab with its own title, language (or file name for detection), content and toolbar actions (API 1.3). Clicking an action sends `{ action, content }` back to the plugin, which replies with new content and/or a status message, or left/right content shown in the IDE's diff viewer. The tab follows the editor document lifecycle: modified marker, close-with-changes prompt, read-only, and saves whose reply arrives after newer edits (or that the plugin rejects) keep the document dirty.
Co-authored-by: dengliu <[email protected]>
Closes#1943
Find Usages and Rename use document-indexer definitions and references next to language-server results. An indexed rename previews every name anchor and applies through the existing atomic workspace edit with grouped undo, guarded by index synchronization, canonical paths and document snapshots; a superseded rename preparation can't revive an obsolete session. Language-server Find Usages and Rename are unchanged. Released WIT is unchanged.
Co-authored-by: arnaud-ritti <[email protected]>
Closes#1948
Plugins receive debounced batches of files created, changed and removed outside the editor in open projects, as SDK `Event::FilesChanged` (delivered through `other("files-changed", …)`, so released WIT is unchanged), under the `events` capability and the plugin's path filters. Activation on these events requires API 1.3. Batches are coalesced per plugin through its actor, survive cancellation without losing deltas, and are dropped when the project or host shuts down.
Co-authored-by: arnaud-ritti <[email protected]>
Refs #1083
A dev container's ports can be forwarded: the Ports tool window shows in a dev container window, `forwardPorts` of devcontainer.json are forwarded on open (keeping the local number when free, falling back otherwise), and ports forwarded by hand are restored per folder and configuration. Each forwarded connection is a loopback listener whose connections run `docker exec … openidea-server --connect localhost <port>` through the container's own network, so no published ports or extra tools in the image are needed. The container tunnel lives in `ide-remote::container_tunnel`; tunnels own their connections and Docker children, so stopping a forward, closing the window or losing the connection leaves no listeners or processes behind (the same ownership fix applies to SSH forwards).
Co-authored-by: andreabasile <[email protected]>
* feat(agents): chat-style conversation view with folded tool runs, tags, jump-to-latest and retract (#1980)
- Lay the conversation out as chat: user messages on the right, agent replies on the left, each with an author row, a card, actions in the header and a time/duration footer.
- Fold consecutive reasoning and finished tool calls into one collapsible 'Reasoning - N tool calls' row with +added / -removed counts; running tools stay expanded.
- Render @file and /skill references in message text as distinct tags.
- Load earlier entries in batches without moving the viewport, offer 'Jump to Latest' when scrolled away, and defer loading while the scrollbar is dragged.
- Add 'Retract Message': stop the agent, drop that turn and every later one, and put the original text/images back in the composer.
- Drop the catalog entry for the replaced duration tooltip.
* fix(agents): preserve conversation scrolling and retract state
---------
Co-authored-by: SunZhiQiang <[email protected]>
Co-authored-by: l0ng-ai <[email protected]>
Whole-line copies and cuts paste above the caret line from every paste entry point (shortcut, menu, history, plain text), including a copied blank line. Paste from History opens a dialog with filter, multi-select, delete and preview. Copy Path/Reference (⇧⌘C) opens a chooser for absolute path, file name, content-root/source-root paths, path with line and reference. Text pasted into a string literal is escaped (Smart Keys setting; Paste as Plain Text bypasses it), and missing imports are added after a paste (Auto Import setting: Ask / Always / Never).
Closes#1984
Worktrees are grouped under their repository; a linked worktree opened as its own root joins its repository's group instead of a separate list. The current marker compares against the roots actually opened, so a project opened from a linked worktree no longer marks the main worktree as current. A root's listing error is still shown when its repository isn't listed by another root.
Co-authored-by: panda-student <[email protected]>
The catalog test rejects Chinese text equal to its English source; CTest (CMake test configuration name) and the Tomcat dialog's URL: and JRE: labels were left as English.
Tasks can be renamed from the list and the tab; an explicitly submitted name is kept. Rerunning the first turn asks before clearing the conversation and resets nested subagent history. Codex questions use the shared question card and keep their ids, so identical questions keep separate answers.
Co-authored-by: panda-student <[email protected]>
Fixes#2039
CMake configurations bypassed the Build executor and Build Project ignored the selected target, so CMake projects had no hammer, configure/build output went to Run tabs, and generated names carried a "(build)" suffix (the build directory) that read like an action. Now the toolbar hammer and Cmd+F9 build the selected CMake target and profile without running it; configure/build prerequisites of Run, Debug and CTest run in the Build window and a failed build prevents the launch; generated executables are "CMake Application" configurations; generated names add profile/build directory only when needed to tell them apart ("hello", or "hello [Debug · build-debug/]"). Saved names and configuration ids are kept. Verified on macOS.
A plugin web view can open a read-only editor tab that streams a process's output, with plugin-defined toggles, Find, wrap, zoom and Clear. The process is authorised in the backend against the exact granted `process:<program>` and its pinned executable, output is bounded, and the stream survives tab switches but ends when the tab closes, the plugin is disabled or the window closes. Adds a web-log test fixture and docs. Verified on macOS; Windows/Linux not run.
Co-authored-by: dengliu <[email protected]>
A plugin web view can open interactive PTY sessions (spawn, input, resize, kill, streamed output and exit). Spawning requires the exact granted `process:<program>` and its pinned executable, with no PATH re-resolution or fallback shell; the cwd is the trusted project, arguments, environment (append-only) and sizes are validated, and at most eight sessions run per plugin. A frame can only drive the sessions it opened; output is acknowledged chunk by chunk and input/output are bounded. Sessions end on permission change, reload, disable, uninstall, frame teardown or window close. Requires API ^1.3; adds a web-pty test fixture. Verified on macOS; Windows/Linux not run.
Co-authored-by: dengliu <[email protected]>
Closes#1950
Plugins can declare `detail = true` on a tree or list view (any tool-window side, grouped or not) to show the selected item's detail-provider tabs beside the list. The pane follows selection and refresh, clears stale content when the item, project or remote host changes, shows an empty state when nothing is selected, and remembers its resizable divider. Services keeps its single split. Released WIT and IPC shapes are unchanged. Verified on macOS in light and dark themes.
Co-authored-by: arnaud-ritti <[email protected]>
Closes#1951
Gutter markers only recognised a short alias list and the runtime truncated icon references at 32 bytes, so ordinary icon names and package paths were dropped. Markers now accept any built-in icon name, and API 1.3 `contributes.markerIcons` declares package SVGs (with an optional dark variant) that are size-limited, sanitised and rendered through the icon registry. The shared SVG sanitiser now also rejects URL strings in any CSS function (e.g. image-set), not only url(). Verified on macOS in light and dark themes.
Co-authored-by: arnaud-ritti <[email protected]>
Adds a local Tomcat Server run configuration: Server / Deployment / Logs / Startup-Connection tabs, an Application Servers dialog (home, version, base directory, libraries), HTTP/HTTPS/JMX/AJP ports, update and frame-deactivation actions, JRE and VM options, and run/debug with redeploy.
Tomcat HOME and a configured base directory are read-only sources: every launch, build metadata write, sync and update goes to an IDE-owned CATALINA_BASE per configuration, which rejects paths equal to or overlapping either source (symlinks included). With "Deploy applications configured in Tomcat instance" on, the source's webapps and conf/Catalina/<host> descriptors are copied into it; off, the recreated base has none. Controls without a runtime consumer were removed; removing every server profile clears the selection. Verified on macOS (run, port changes, sync/redeploy, servlet breakpoint, source trees unchanged); Windows/Linux not run.
Co-authored-by: RuyimgByCN <[email protected]>
Closes#1954
Plugin file templates could only take a typed name plus a fixed extension, so dot-files such as .dockerignore and templates needing more than a name weren't expressible. Templates (API ^1.3) can now declare a fixed `fileName` and named `variables` that the New File dialog asks for, and plugin templates get the same Java/Kotlin/Go/PHP package inference as built-in ones. Unsafe filenames and existing files are refused. The dialog no longer submits on Enter while an IME is composing, and its variable form has proper padding. Adds a file-templates test fixture.
Co-authored-by: arnaud-ritti <[email protected]>
Closes#1834. A console result grid is editable in place when its rows trace back to one base table: a plain SELECT of column references / * from one table (schema-qualified or aliased, resolved through the console's active schema) whose primary key columns are all in the result. Edits, added and deleted rows reuse the table editor's workflow and are applied in one transaction, then the query re-runs. Aliased or computed columns stay read-only; joins, aggregates, subqueries/CTEs, unions and ambiguous column provenance show a read-only banner with the reason.
Co-authored-by: Inf1mum <[email protected]>
Refs #1083 (step B). A local project with a devcontainer.json can be reopened in its dev container (File › Reopen in Container, Find Action): the Dev Containers CLI brings the container up and the window connects to openidea-server inside it, as for WSL, so terminal, language servers, run, debug and git work in the container. The project's identity stays the local folder, so reopening from Recent Projects finds the same container. Missing CLI gives install guidance; trust of the local project carries over. Pure CLI output parsing lives in ide-remote; a failed connection no longer closes the current backend early, saved locations keep the container configuration, and Docker context precedence follows the CLI.
Co-authored-by: andreabasile <[email protected]>
Closes#1836. AppKit reports page swipes (three-finger NSEventTypeSwipe and two-finger page swipes) outside the web view's key/mouse events, so the desktop shell now forwards them as SwipeLeft / SwipeRight keymap strokes that can be recorded and remapped. Every macOS preset (IntelliJ, Eclipse, VS Code) binds SwipeRight to Back and SwipeLeft to Forward; two-finger swipes over horizontally scrollable content still scroll, and turning off the system page-swipe gesture disables navigation.
Co-authored-by: jaw52 <[email protected]>
Closes#447. Each SQL console gets an active-schema dropdown (PostgreSQL / MySQL / Oracle schemas; SQL Server selects the database, since T-SQL has no per-session default schema), remembered per console file; hidden for SQLite and ClickHouse. Every statement run through the pool first asserts the console's context, or the connection's default when none is selected (skipped when the tracked context already matches), and manual-commit sessions restore the default before the connection returns to the pool, so one console's schema never leaks into another.
Co-authored-by: Inf1mum <[email protected]>
Refs #1083 (step A2). Debug now works on Python configurations whose run target is a Docker image or Compose service: the run starts the program in a new container under debugpy (the image's own, else the IDE's managed copy mounted read-only), publishes the debug port on 127.0.0.1 only, and the debugger attaches like a Python Remote Debug configuration with the project folder mapped. Console scripts (pytest) run as modules and `uv run` keeps the manager in front; shell command lines are refused. Container debug startup is bounded (error when the container exits early or debugpy isn't ready in 120 s) and stopping the debugger stops the container.
Co-authored-by: andreabasile <[email protected]>
Fixes#1977. External Libraries only inspected the project root and cached incomplete results, so ecosystems in nested modules/workspace members (e.g. a pnpm frontend under a Cargo + Maven repo) and dependencies installed after the first listing never appeared. Providers now discover module/workspace roots (package.json workspaces, pnpm-workspace.yaml, Cargo members, nested manifests), follow pnpm symlinks, include local Cargo path dependencies, stream groups to the Project view as each provider finishes so a slow cargo metadata doesn't hide the others, log failures instead of dropping them, and don't cache cancelled or failed results.
Closes#1976. Settings › Build Tools › Maven gets a JRE field (per project or global): empty uses the Project JDK, else the chosen JDK runs Maven sync, builds, goals and tests (JAVA_HOME). One resolver (ide_sdk::apply::maven_jdk / maven_build_env) now serves the build model, compile steps, Maven run configurations and the mvnd Java 17 check; an explicit JAVA_HOME in a run configuration still wins.
Fixes#1913. The sign-in hint named the bare command (`cursor-agent login`), but the IDE launches agents through its own resolution (registry installs under the config dir, the login shell's PATH, PATHEXT shims), so the named command could be missing from the terminal's PATH. The agent spec is now resolved once before connecting, and when that executable isn't what a PATH lookup of the bare name finds, every built-in hint names the resolved path, quoted for PowerShell or POSIX shells. Hints for separate login tools (claude-acp, codex-acp, vibe) and custom agents are unchanged; the hint is translated through templates.
Native full screen moves a window's buttons into the bar revealed with the menu bar, and a project tab shown in its group's full-screen Space has none, so the title bar's slot for them stayed empty. The title bar now draws them there while the window fills the screen: close closes that project's tab, the green button leaves full screen, and minimize is disabled.
Adds the API 1.3 `endpoints-provider` export and `[[contributes.endpoints]]` (file globs, ^1.3, `project.read`). A plugin's routes join the Spring ones in Search Everywhere, `.http` URL completion and the declaration's Open in HTTP Client gutter action. The UI asks `plugin_endpoints` on demand; the host runs the plugin on a blocking thread, validates the whole list (bounds, project-relative locations, UTF-16 ranges) and caches it per project until a file matching the globs changes, then emits `EndpointsChanged`. A failed or invalid run keeps the last list; disabling, crashing or closing the project withdraws it, and late results can't restore it. Closes#1960.
Co-authored-by: arnaud-ritti <[email protected]>
Adds API 1.3 `ui.pick-directory` (native folder chooser, same window/budget rules as `pick`) and `workspace.open-project(path, new-window)`, which goes through the normal File › Open flow with its trust prompt, window reuse and unsaved-change handling. Paths are absolute on the UI machine and validated by the host whatever its OS. Closes#1959.
Co-authored-by: arnaud-ritti <[email protected]>
A plugin title, label or message equal to IDE text ("Pick", "Log", "Refresh") was translated by the IDE catalog. Primitives now translate given text through useGivenTx, which returns it as written below VerbatimTextContext (plugin views, service panes, status items, prompts); plugin commands, tool windows, menu labels and notifications carry a verbatim flag. Command categories and the IDE's own words stay translated.
Import Settings… also accepts a JetBrains settings.zip: colour schemes (and the current one), editor font, size, line spacing and ligatures, trailing-space stripping, final newline, sticky lines, whitespace, soft wraps and confirm-on-exit are mapped after a confirmation; keymaps, code styles, live templates, inspections and plugins are listed as not imported.
Refs #656
* feat(settings): JetBrains-style color scheme sub-pages
Move General, Debugger, Language Defaults and each language onto its
own page under Color Scheme, drop the redundant scope root from the
attribute tree, and turn the Color Scheme root page into a landing
with links to the sub-pages.
* feat(settings): per-scope preview samples with click-to-reveal
Give every scope page a code sample covering as much of its token set
as possible, render bg-only attributes in the preview, and let a click
on a preview token select its attribute and expand its tree ancestors.
* refactor(settings): fold attribute name groups into collapsible tree levels
Split the "›" groups of an attribute display name (Identifiers, String,
Comments, ...) into collapsible intermediate nodes so the tree matches
IDEA instead of repeating the group in every leaf label.
* feat(editor): preview the scheme's actual attribute state in settings
Attribute controls now mirror the scheme's effective appearance instead
of only showing user overrides: checkboxes and colours resolve imported
scheme values, theme tokens and built-in defaults. Removing an inherited
colour stores an explicit transparent override instead of falling back
silently, and redundant overrides matching the inherited value are
dropped. Also drops the stale "Incoming commits" catalog key.
* feat(settings): move the effect style onto its own right-aligned row
Put the effect checkbox and its colour on one line and the style
select on a narrower row below, right-aligned with the hex fields,
which now stretch to the panel edge.
* refactor(settings): render attribute tree labels in the plain font
Drop the per-token styling of the tree labels; IDEA lists them as
normal text, the preview keeps the colours.
* feat(settings): scroll the attribute tree to a preview-clicked token
Preview clicks now also reveal the attribute: after the ancestors
expand, the tree selects the row and centres it in the viewport.
* feat(settings): mark overridden attributes in blue up to the root
Show overridden attributes and every group containing one in the
accent colour, like IDEA, instead of a dot on the leaf alone.
* feat(settings): use the JetBrains demo as the Python preview and reveal cross-page tokens
Replace the Python sample with the classic syntax-highlighting demo (semantic
spectrum, f-strings, byte strings, escapes) while keeping coverage for every
Python attribute. Preview tokens outside the current sub-page now navigate to
their owning page and reveal the attribute there.
* fix(settings): keep locale entries, English preview text and default styles in the colour scheme preview
- Restore the "Incoming commits" translations that were dropped from other locales
- Replace the Chinese docstring lines in the Python preview with English text
- Use the existing Java operator attribute for separators in the Java preview
- Render default bold, italic and underline in the preview, matching the editor
---------
Co-authored-by: l0ng-ai <[email protected]>
A comparison window (Git Log > Compare with Local) of a remote project
rides its parent's session: remote_current answers with the parent's
connection, so its commands go to the host where the project is open.
But the window never opens a project of its own — it keeps only the
settings (restoring the project would overwrite the main window's
editor session) — and the useProject subscription read every settings
update as 'the window left the remote project', dropping the routing.
The next commands then ran on the desktop, where no project is open
for the window: Git Log > Compare with Local on a WSL project opened
an empty comparison window with 'IpcError: no project is open in this
window' (same for SSH remote projects).
The subscription now only ends the session when the window's project
actually leaves the session's root: it had the session's project open
and now holds none or another one. Opening a local project, LightEdit
and Close Project still end the routing (open_project /
light_edit_open / close_project route locally and end it themselves;
closeProject keeps the subscription as its safety net).
A package's locales/<tag>.json (English source text -> translation) is applied to the plugin's own manifest titles and labels and to its view data; new ui.locale() (API 1.3) returns the UI language for text the plugin builds itself. Catalogs only ever apply to their own plugin; text without an entry stays English.
Co-authored-by: arnaud-ritti <[email protected]>
Plugin API 1.3 adds what commit-message generators and before-commit checks need:
- `menu = "commit.message"`: a plugin command, with the plugin's icon, in the toolbar above the commit message of the Commit tool window and the Commit dialog.
- `commit.pending-commit()` (`vcs.read`): what Commit would commit right now, per file its repository, path, status and a `git diff` of only the included files and changed blocks against HEAD (1 MiB in all), plus the message, Amend and a revision. The diff is made by the same selection the commit uses, in a copy of the index, so the index and the working tree stay as they are; unsaved editors of the included files are saved first, as Commit does. Only in trusted projects.
- `commit.set-commit-message(revision, message)` (`vcs.write`): replaces the message only while the draft is still at that revision, so text the user typed meanwhile is never overwritten (`stale`).
- `[[contributes.commitChecks]]` and the optional `commit-checker` export (worlds `plugin-commit-checks` / `commit-checks`): checks listed in Commit Options (on by default, `vcs.commit.disabledPluginChecks` turns them off) and run after the IDE's own; their problems, optionally with a location, stop the commit until Commit Anyway.
The window answers the host through `CommitRequested` / `plugin_answer_commit`, also for remote windows. The Rust SDK gains `host::pending_commit`, `host::set_commit_message`, `openidea_plugin::commit` and their mock; a `commit` test fixture exercises them. All components are rebuilt for the WIT change.
Refs #1650
Following the active editor already worked, but the option kept IDEA's pre-2023 name and sat second under Behavior, so people didn't find it (#1666). It is now Always Select Opened File, listed first, and the header's Select Opened File button hides while it's on. Find Action still matches the old name; the setting key is unchanged.
Closes#1666
Co-authored-by: jaw52 <[email protected]>
* fix(agents): improve chat transcript interactions
* fix(agents): recover images when regeneration fails
* test(agents): use the native copy shortcut on every host
* perf(project-tree): measure row widths once per frame
The horizontal scroll range was measured in a layout effect on every render:
each pass forced layout, its setState re-rendered synchronously and measured
again, and the ResizeObserver was recreated each time, firing once more.
Expanding a 10k-entry folder went from ~26 ms to ~63 ms.
Keep one observer per tree and coalesce renders and resizes into a single
measurement on the next animation frame. Expanding is back to ~30 ms and
arrow-key p50 drops from 3.3 ms to 1.7 ms; the widths behave as before.
* perf(github): load the pull requests integration after start-up
Actions, the tool window and the timeline editor stay registered at start-up;
their implementations, the connection and review-mode tracking load on idle
or on first use. Takes ~60 KB gzip off the initial bundle.
* perf(templates): load Emmet when an HTML or stylesheet editor opens
The Emmet engine (~70 KB minified) was part of the initial bundle; it now loads
with the first editor that can expand abbreviations, or with Update Tag with Emmet.
* perf(search, run): load popups and Edit Configurations when first shown
Search Everywhere, Find in Files, the recent files/locations popups, File
Structure and the Run/Debug Configurations dialog were part of the initial
bundle although each only renders while open. `lazyWhen` mounts a lazy view
only while its condition holds; the idle prefetch still loads them right after
start-up, so opening one stays as fast.
* perf: load dialogs, the SQL console and diff views on first use
- Database dialogs, Assign Shortcut and the Logcat tool window load when shown.
- The SQL console extensions load with the first SQL editor
(`lazyEditorExtension` reconfigures the editor once they arrive).
- The changes tree loads the diff viewer when a file is selected, and
"Open Repository Log" loads the revision browser when run.
* perf: load plugin, debugger and project popups when first shown
The IntelliJ theme import, the debugger's popovers and dialogs and the title
bar's project popup now load when first opened. `lazyWhen` keeps a view
mounted once shown, so its own open/closed handling is unchanged.
* perf(angular): parse only text with Angular syntax as Angular
Every text node of a template got its own nested Angular parse, including
the whitespace between tags, which is most of them. Text without a `{`, `}`
or `@` holds no interpolation or block, so it stays HTML text. A full parse
of a 12k-line template takes about a fifth less work.
* perf(angular): reuse nested trees of unchanged text and attributes
An edit makes the mixed parser recreate many text and attribute nodes it
does not keep: the document's own text and the nodes up to the viewport. It
reuses nested trees only of nodes it kept, so each of those was parsed again
as Angular on every keystroke (~170 nested parses per key in a 12k-line
template). Their parse depends only on their text, so the trees are now
remembered by text, up to about 2M characters.
Typing in the middle of a 12k-line template: main thread per key 57 -> 22 ms,
p50 latency 30 -> 19.5 ms, no long tasks.
* perf: budget typing in a large Angular template
* perf: measure code completion over a 2,000-item list
The harness's probe can answer completion with generated items through the
command route; language server requests go through a route when one is
installed, not only under Tauri. The scenario opens the popup by typing,
narrows it letter by letter and moves the selection, with budgets for each.
* perf(run): show a flood of console output at most every 50 ms
While a process printed thousands of lines a second, the Run console
appended, re-rendered and laid out on every frame, so typing in an editor
meanwhile missed the next frame 35% of the time. A line after a quiet spell
still shows on the next frame; a flood now updates the console at ~20 Hz,
and typing during it misses 2-5% (p50 16 -> 9.7 ms) with the output done as
fast as before.
The performance harness gets a scenario for it: 50,000 log lines at 200 a
frame (a dev and perf build action, run.sampleOutput) while typing.
* perf(vcs): keep Git Log rows from re-rendering on every selection change
Commit rows were memoized but received a new Set and new closures on every
render, so moving the selection re-rendered every visible row, and each row
built a new Intl.DateTimeFormat for its date. Rows now get stable handlers and
shared remote-name sets, and date formatting reuses cached formatters.
* perf: measure the Git Log and splitter dragging
gitLog shows 20k generated commits served page by page through a fake vcs_log,
walks the selection with the arrow keys and wheel-scrolls the list.
resizePanels drags the side and bottom splitters with an editor, the Project
tree and the Log open. The perf probe's fake backends can now be combined.
* perf: translate the sample output action and keep fake completion details out of the text catalog
* perf(vcs): keep Git Log scrolling smooth as pages load
Each page of the log recomputed hidden long edges by scanning every row once per
lane, and the store laid out the whole graph a second time for fields nothing
read. Long edges are now found in one pass over the rows, the unused store graph
is gone, and the next page is requested while a page of rows is still ahead so
fast scrolling doesn't reach the end before git answers. The commit list also
contains its layout, so rows entering while scrolling no longer relay out the
workbench.
* perf: load the Git Log scenario's pages with git's latency and scroll across several
Right-clicking a commit with branch labels offered commit actions only. The menu now has a Branch 'name' submenu with the existing branch actions for each branch at that commit other than the current one, grouped under Branches when there are more than two, and keeps Rebase onto Selected Commit for commits without branches. The actions run against the commit's repository in a multi-repository project.
Co-authored-by: shohruh-genius <[email protected]>
The New Branch dialog refused any existing local name, so a branch could not be moved to a selected commit. When the name matches a local branch of the target repository, the dialog now offers an unchecked Overwrite existing branch option (git switch -C / git branch --force); the consent clears when the name changes, and Git's checkout checks keep the old tip and local changes if switching fails. New Branch from a Git Log commit also targets that commit's repository in a multi-repository project.
Co-authored-by: shohruh-genius <[email protected]>
Stop flipped the session back to Run as soon as the direct child exited, but a shell, Maven or the debug adapter can exit while the application it started still holds its port, so the next Run failed with "port already in use". The console also went quiet on a debug Stop because the DAP session was disconnected before java-debug drained the program's output, and closing a window killed running programs with SIGKILL at once.
Run and Debug sessions now stay in a stopping state from the Stop press until the whole process group (or the debuggee's pid tree) is gone. As in IntelliJ IDEA, the first Stop only interrupts and waits, and a second Stop kills; the Stop button turns into Kill process meanwhile. A rerun or restart waits for the old process without a time limit, and the termination time limit setting now only bounds a configuration's stop command. Debug launches record the program pid from the adapter's process/processid event, signal its pid tree and keep the adapter attached until the program exits; closing a project also stops debug sessions under its root. Closing a window with running programs shows a Terminating panel with Kill process and closes once they are gone.
Co-authored-by: misswell <[email protected]>
* fix(vcs): surface stage/unstage failures instead of swallowing them
The Commit tab's Stage/Unstage items ran the IPC call and only attached a
.finally(refreshNow) handler. call() rejects with IpcError on any backend
failure (require_trust, git add failure, path routing), so the rejection
went unhandled: no notification, no console line, and the UI silently
showed the file still unstaged — reported as 'Stage/Unstage doesn't
work' (#1890).
Route the four handlers through runOp(), the same helper every other
mutating VCS op uses: it logs the command to the Git console, shows an
error balloon with the backend's message on failure, and refreshes the
status afterwards (so the manual .finally is dropped).
This doesn't change the success path; it makes the failure visible, which
is what the reporter needed to diagnose the real backend error.
* fix(vcs): add text catalog entries for the stage/unstage op titles
The native Claude Code provider turned each AskUserQuestion question into a
permission request with only the option labels, shown in the permission
banner. Descriptions, headers, previews and multi-select were lost, and the
only way to give another answer was Skip, which tells Claude the user didn't
answer.
All questions now go out in one request carrying the full tool input, and
the conversation shows them as a card above the composer: each option with
its description, radio buttons or checkboxes for multi-select, an Other
choice with a text field, the option's preview beside the options on a wide
card (below them on a narrow one, sized to the tallest preview so hovering
never moves the options), number keys to pick, Esc to skip, and one Submit
for every answer.
Ordinary permission requests keep the banner, and Skip answers as before.
The answers go back in Claude Code's own format: question text to answer,
multi-select picks and Other text comma-joined, with an item that contains
the separator or a quote JSON-quoted so it reads back as one item.
Closes#1628
Co-authored-by: NealimeKenna <[email protected]>
Extends the VenvTool set with a "Conda Environment" option: the dialog
runs 'conda create -p <location> -y', pinning 'python=<major.minor>' from
the chosen base interpreter when it reports a version. An existing
conda-meta directory is refused before conda runs, and a missing conda
binary reports an install hint instead of a spawn failure.
Refs l0ng-ai/openidea#1724
* feat(sdk): create Python environments with uv and Poetry in Add Interpreter
Add a VenvTool parameter to sdk_python_create_venv so the Add Python
Interpreter dialog can build a new environment with virtualenv, uv or
Poetry, matching what the environment discovery already finds. Poetry
keeps envs in its own cache, so the dialog disables the location field
for it and resolves the interpreter via 'poetry env info --path'.
Fixesl0ng-ai/openidea#1776
* fix(i18n): update text catalogs for the new env-tool strings
The Add Interpreter dialog gained uv/Poetry labels and per-tool notes,
and its spinner text changed from 'Creating virtual environment' to
'Creating environment'. Add the new zh entries and rename the stale
key in every catalog so the text-catalog test passes.
l0ng-ai
Vadim LazukandInf1mum
Arnaud Ritti
Deng Liuanddengliu
Andrea Basileandandreabasile
RuyimgByCN
Vadim LazukandInf1mum
Jaw
LU XIANGRUI
Muuandl0ng-ai
Shohruh Ismoilovandshohruh-genius
misswellandmisswell
arookieofc
NealimeKennaandNealimeKenna