m2e resolves the local repository from the Maven settings XML, never from
-Dmaven.repo.local, so JDTLS imports kept downloading into ~/.m2/repository
even when the IDE's own mvn runs honored the configured path. When a local
repository is configured but the user settings lack <localRepository>, derive
an effective settings file under <root>/.openidea/ and pass it as
java.configuration.maven.userSettings. Settings that already declare
<localRepository> are passed through untouched.
IDEA writes path options as URLs (`WORKING_DIRECTORY = file://$PROJECT_DIR$`). The run configuration importer expanded the macros but never stripped the `file:` scheme, so on Windows `file://C:/project` became the drive-relative `C:project` and was joined onto the project directory, and imported configurations failed to launch. Every path-valued option the importer reads (working directories, SDK home, script and settings files, pytest targets, env files including EnvFile entries, package.json, package manager, node interpreter, config file, test directory/file, JS file, interpreter path, path arguments) now goes through one conversion: drive letters in the authority position, `localhost` as a local authority, UNC for a real host, and IDEA's unencoded paths kept verbatim. Env-file lists no longer split at a URL's scheme colon on Unix.
Co-authored-by: Clarence404 <[email protected]>
Find Usages and usage counts searched again on every request: about two
seconds each for a common type such as String.
- The Java usages extension (1.12.0) keeps answers by search target until a
Java element, the classpath or an unsaved edit changes, and requests for
the same target share one search.
- Widening checks up to 5000 candidates against the server's analysis scope
before applying the 300-file limit, so names mentioned in many covered
files no longer report "too many files".
- Candidate scans skip the source folders the server covers fully
(openidea.workspace.coveredRoots) instead of reading every file.
- Commands a server registers after initializing are added to its commands.
* perf(java): count usage lenses in the project sources only
Usage lenses asked JDT LS for a full references search per declaration,
which also scans every dependency jar and the JDK. In a library class such
as java.lang.String that kept JDT LS busy for tens of seconds per pass,
and failed counts were asked again on every scroll.
- Scoped Java searches without libraries pass the "sources" scope.
- Lenses count declarations near the viewport (10 lines instead of 60).
- Incomplete counts are remembered per document version and asked again
once the server settles, instead of on every pass.
* perf(java): turn off the language server's own reference lenses
JDT LS shows a references lens above each declaration by default and the
editor resolved them all at once, a references search per declaration on
every open and edit. The editor's usage lenses already count usages.
* style(editor): keep the usage lens's unrelated lines as they were
Python SDK discovery only scanned ~/anaconda3-style install roots and
<root>/envs — an Anaconda installed off the home dir (D:\soft\anaconda3,
ProgramData) hid every environment from the Conda Environment picker even
though 'conda env list' saw them all.
Now: the conda binary on PATH reveals its install root (<root>/Scripts or
bin → prefix), 'conda env list' adds --prefix envs kept outside envs/ and
envs_dirs configured in .condarc, and the env probe looks for python.exe at
the env root on Windows — where conda puts it — instead of only under
Scripts/ (virtualenv layout).
A Surefire/Failsafe fork started with suspend=y never reaches the console
with its "Listening for transport" line: surefire buffers the fork's
output until the tests run, and the tests wait for the debugger. The IDE
found no port, never attached, and Stop left an orphan debug session.
Pick a free port up front (new debug_free_port IPC), pass it to the
-Dmaven.surefire.debug/-Dmaven.failsafe.debug agent as
address=127.0.0.1:<port>, and attach right away — java-debug retries the
connect while the fork starts, so no output line is needed. A duplicated
line that does arrive later is deduplicated by port.
A just-killed dlv can still hold this session's binary while the previous
run's lock is already released, so one remove_file attempt only ever
deleted the older leftover. Retry with a short back-off so the sweep
outlasts the handle release.
* fix(java): run jdtls on a new-enough JDK behind Scoop shims and project JAVA_HOME
Two ways jdtls ended up on a JDK too old for its bundles (#2227):
- jdtls_min_java resolved the command's own path to derive the install
root, but a Scoop shim (shims\jdtls.exe) hides the release layout, so
the bundles' real requirement fell back to JDTLS_MIN_JAVA (21) while a
1.64 install needs JavaSE 25. Extract the resolve/follow-shim flow
from launch_directly into jdtls::script_for and share it.
- configure_jdtls only wrote JAVA_HOME when the config had none, so a
project JAVA_HOME (JDK 8 for compilation) won over the chosen
server_jdk. Probe the configured value: a home too old or incomplete
for the server is replaced by server_jdk or any usable JDK; an
unresolvable explicit value stays (the user's call).
* test(java): compare JDK homes against canonical paths
jdk_at reports the canonicalized home (macOS /var -> /private/var,
Windows 8.3 short names), which configure_jdtls now writes to JAVA_HOME.
Have test_jdk canonicalize too, and assert the runtimes/JAVA_HOME values
against the Jdk.home fields instead of raw fixtures.path() joins, which
also differ on Windows when the join arg uses forward slashes.
* test(jdtls): compare JAVA_HOME against the canonical jdk home
test_jdk now canonicalizes home, so the duplicate-versions assertion must
compare against server.home rather than the raw tempdir join, which on
Windows carries 8.3 short names.
New agent tasks defaulted to Bypass whenever the provider offered it, whatever the user had chosen. A task now starts with the settings shown in the composer: the provider's reported mode, or Ask for Cursor. An explicit choice is kept, including Bypass, and running tasks keep their own mode. Cached provider settings could hold a Bypass written automatically with no record of its source, so a cached bypass is dropped on load, read and save while the option stays offered.
Fixes#740
`kill_tree` and `kill_tree_of` run `taskkill /T /F /PID` on Windows even when the child has no descendants, which is the common case for a program a plugin started, read from and closed. `taskkill` is a process spawn — ~0.7 s here, measured on this machine — and it runs inside host calls that are inside a plugin call's budget: dropping eight process handles at once overran a 5 s budget, so the plugin was trapped for something the host did.
When `descendants(pid)` is empty the child itself is all there is to kill, and both callers already do that (`kill_tree` follows with `child.kill()` + `child.wait()`, `kill_tree_of` leaves the child to its owner), so the spawn buys nothing. Skip it.
Verified:
- `cargo test -p ide-core` → 77 passed.
- `cargo test -p ide-host --test plugins` → 95 passed, including `a_plugin_keeps_a_process_running_behind_a_handle`, which fails on unmodified `main` on Windows (`plugin "misbehave" timed out after 5s` at its `childdrop` step, which drops eight handles) and now passes in 9.7 s instead of 20 s.
- clippy `--all-targets -D warnings` and `cargo fmt --check` clean; no component or WIT change, so nothing to rebuild.
Co-authored-by: chao <[email protected]>
Co-authored-by: l0ng-ai <[email protected]>
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.
New Agent Task showed no model, effort or mode for Claude until an agent had run a task. The `initialize` request that already loads commands also returns those options, so they are now remembered and shown, with one extra `get_settings` request to read the configured effort. No prompt or session starts, a failed lookup is not repeated, and a session's own report always wins, in memory and on disk.
Co-authored-by: panda-student <[email protected]>
The main toolbar and status bar could not be hidden, Tool Window Bars was a Settings row only, and widgets could be switched only from the status bar's context menu. View > Appearance now has Toolbar, Tool Window Bars, Status Bar and Status Bar Widgets, with check marks in the macOS and in-app menus, plus Settings toggles. Both new settings default to shown. With the toolbar hidden, the window keeps a compact header with the title, which Distraction Free now shares.
Controls stay reachable with a bar hidden: the Git branches popup and background tasks have their own hosts, and background tasks has an action.
Co-authored-by: lleviathe <[email protected]>
The refusal added for indexed-only renames missed code names in Go, Ruby and PHP, ran without the rename cancellation, and was English only. It now covers those node kinds, honours cancellation, and translates the hint.
When no language server answers a rename (none running, or no rename
support) and the indexed anchor under the caret is a code identifier,
prepare and rename now refuse instead of renaming only the plugin's
anchors, which left the declaration, imports and file name behind.
Anchors in strings, comments, attribute values or files without a
grammar keep their indexed rename.
Closes#2204
Some servers reuse one `$/progress` token for concurrent work (PHPantom sends begin, begin, end, end). The second begin overwrote the first task, which never got a done event and stayed at 0% in the background tasks popup. A begin on a token with a live task now ends that task first. The progress map keeps one entry per token, so busy status and idle waits count as before.
Co-authored-by: lleviathe <[email protected]>
Indexed markers (Edges.markers) were always drawn with the navigation glyph,
so a plugin couldn't tell its marker kinds apart. A [[contributes.markerIcons]]
entry can now claim them by key prefix (`markers`), with an optional
`tooltip` shown above the targets. The host matches the longest prefix and
sends the declaration id with each marker; the editor resolves the icon and
the translated tooltip from the projected declaration. Manifest only, under
the existing API 1.3 gate of markerIcons.
Closes#2194
Native Codex sessions never requested `skills/list` or published available commands, so slash completion showed no skills. New, resumed and forked sessions now publish enabled skills as `$<name>`, and `skills/changed` refreshes them. Failed or unsupported requests stay silent.
The Maven and Gradle tool windows disappeared in a workspace, because detection only inspected the workspace metadata folder. They now discover builds in every linked project; models, caches, Maven profile choices, refreshes and run directories use the owning project, and Java reload notifications target its build files. CMake already handled linked projects. Spring's endpoint filter is unchanged.
Fixes#932
Plugin tool windows could only target the primary split of their dock, because the view was never registered as secondary. An optional secondary = true on [[contributes.views]], gated to plugin API 1.3, places a view in Left Bottom, Right Bottom or Bottom Right. Existing plugins stay where they are, and a user's saved layout wins.
Co-authored-by: dengliu <[email protected]>
Launch and attach were cut off after 60 seconds, so a slow interpreter start failed with "debug adapter did not answer launch in time". They now wait for the adapter's result: an adapter error or a closed connection still fails startup, Stop cancels it without an error, and other request deadlines stay.
Fixes#1773
An installed JDK that cannot run, such as an Intel-only JDK on Apple silicon, was registered as a project runtime and could crash Java language server initialization even with a working JDK selected. Each JDK's executable is now probed before selection, off the UI thread and outside the cache lock; unusable JDKs are left out, working Java 8 runtimes stay, and macOS bundle paths are accepted. An explicit choice that cannot run reports its actual executable. Since JDT keeps VM installs in its workspace, a project with an excluded JDK gets one fresh workspace; everyone else keeps their path.
Fixes#1668
Run and Debug now share one Java preparation path: save documents, build with the configured policy, resolve the module classpath and JDK, and stop on a failed build. Incremental Make is the default; Rebuild, NoBuild and Build-no-error-check are persisted on the run configuration, and the form offers only what a build tool supports. Delegated Gradle goes through the same path, Debug reports save failures, Maven snapshots honour Rebuild, and scoped diagnostics are retained so incremental Maven cannot erase compilation errors.
Co-authored-by: misswell <[email protected]>
A child that stopped reading its input could freeze the whole window, because terminal input was written synchronously on the UI thread. Terminal, Run and debug stdin now share a per-session ordered writer queue bounded to 4 MiB and 1,024 writes. An overflow rejects the whole write and shows an error; closing or killing a session discards pending input without waiting for the child, and an explicit stdin EOF drains accepted writes first.
Refs #2160
Fixes found while verifying Ruby debugging. An entry point with no arguments reached Ruby's single-string exec in rdbg, so spaces and shell syntax were interpreted; it now goes through ruby -S -- and stays literal. rdbg sends no DAP output or process events, so its output streams are forwarded and drained at shutdown, and the launched process group is terminated on Stop or a failed start; attach leaves the process alone. Bundler lookup respects BUNDLE_GEMFILE and gems.rb, gem discovery uses the configuration's environment, and the missing debug gem message is translated.
Refs #2170
Co-authored-by: krmbzds <[email protected]>
A lone Alt press always opened or focused the main menu, with no way to
turn it off. Add `appearance.mainMenuAlt` (default on, mirroring IDEA's
ide.windows.main.menu.focus.on.alt) in Appearance > UI options; when off,
Alt key combinations keep working but pressing Alt alone does nothing.
Closesl0ng-ai/openidea#1155
Ruby files, `rails server`, Rake tasks, RSpec and Minitest configurations
can be debugged with rdbg, using the project's Ruby. In a Bundler project
it runs through the bundle; if the bundle lacks the `debug` gem, debugging
stops with an error asking to add `gem "debug"` to the Gemfile.
Reload Changed Classes ran an uncancellable JDT build with a fresh token. The UI now allocates a request id, offers Cancel Hot Swap Build and treats cancellation as cancelled rather than a failed reload; overlapping reloads no longer overwrite the active request. A cancelled build can write class files without publishing their build deltas, so java-debug cannot recover them on the next build: after a cancel, Hot Swap reports that the session needs a restart rather than claiming success with classes left out.
Co-authored-by: misswell <[email protected]>
Four ide-agent tests failed on a Windows machine with a normal setup: skill discovery walked up from a temp project into the real profile's skills, and a fixture repo inherited the global core.autocrlf. The fixtures now bound the walk at the project and turn autocrlf off.
Co-authored-by: panda-student <[email protected]>
Adds Maven toolchains file (-t) and active profiles (-P) settings, emitted before the goals. They combine with the Maven tool window's profile checkboxes, a configuration's own -t or -P overrides them, and repeated launches do not duplicate them. The settings do not reach jdtls's import: no toolchains file or explicit profile deactivation there.
Co-authored-by: misswell <[email protected]>
Adding projects to a workspace took one folder at a time. Choosing a folder that is not itself a project now scans below it, up to four levels and a shared entry budget, without following symlinks or descending into a project and skipping dependency, output and hidden directories. The projects found appear in a checklist with linked ones disabled; a folder with none is added as before. The scan is cancellable and runs in both the create and manage dialogs.
Fixes#2102
Claude Code reports its slash commands only after a session starts, so New Task listed only skills on disk. When the native provider has no remembered commands, a CLI process now runs only the initialize request, with a five-second timeout, shared by concurrent calls, and its result goes into the existing per-project store. A failed attempt keeps the disk list and is not repeated. No prompt or session is created.
ACP and other native providers still report commands after a session starts.
Fixes#2144
The shared twelve-result cap hid commands late in the order, and commands and skills looked the same. Command records now carry a source (stored tasks default to command; disk skills and Codex $skills are marked as skills), and the popup shows every match under Commands and Skills headers in the existing virtualized list, with keyboard selection skipping headers. File completion keeps its cap.
Fixes#2145
Serve the open projects to MCP clients and the IDE's agents: files, search, symbols, problems, edits through IDE documents, run and build with confirmation in the project's window, and Git status and diff.
Co-authored-by: haruka <[email protected]>
Co-authored-by: imacross <[email protected]>
Go files showed no semantic colours because gopls semantic tokens were not requested, and Color Scheme had no Go page.
gopls is now initialised with semantic tokens, constants, package variables and package references get Go-specific roles, and Color Scheme has a Go page with a preview.
Siblings covered: Go-specific attributes inherit from Language Defaults, Reset restores them, Chinese labels show in the Color Scheme tree, and a user's own gopls settings are kept.
Co-authored-by: shohruh-genius <[email protected]>
Kotlin files showed declarations, parameters, fields, annotations, KDoc and escapes in generic lexer colours, because the CodeMirror mode alone drove their highlighting.
Kotlin now gets supplemental backend marks from its tree-sitter grammar, which take precedence over lexer marks while LSP semantic marks stay innermost. The semantic provider also maps the mutable and operator-declaration tokens.
Siblings covered: other grammars keep their parameter colours, and constructor default values are highlighted as values.
Co-authored-by: shohruh-genius <[email protected]>
jdtls reports diagnostics for every imported project, so a module launch was treated as having compile errors because of unrelated sibling modules or its own test sources.
Launch errors now count only the launched module and the modules on its classpath, and application launches ignore test sources. A build with errors becomes a success only when the scope is known and nothing in it fails; if the classpath cannot be resolved, all errors are kept. What a launch does with reported errors is unchanged.
Siblings covered: symlinked and Windows-cased paths, Gradle and Maven output folders, and module paths.
Co-authored-by: misswell <[email protected]>
Removing a worktree could delete the folder the window had open: only the main worktree was blocked, so a linked worktree opened as the project, or one containing the project, went after one confirmation, and Force Remove discarded its changes.
`vcs_remove_worktree` now compares canonical paths and refuses a target that is the project root or an ancestor of it, with or without force, before running git. Worktrees below the project root stay removable, including a project opened on a parent folder.
Sibling covered: the Worktrees tab disables Remove for such a worktree and says why.
arookieofc
Clarence
Deng Liu
l0ng-ai
ppxb
SunZhiQiang
lleviathe
Pedro Pessoa
misswellтаmisswell
Kerem Bozdas
Shohruh Ismoilovтаshohruh-genius