codemirror-vim only discards single-character IME commits in Normal mode, so multi-character commits (Chinese pinyin, Japanese conversions) were written into the document and could replace the selection. In Normal/Visual mode the composition now renders as usual and its whole span is restored when it ends; an ASCII single-character commit (dead keys) is replayed as a Vim key. Composition changes are marked transientChange: they stay out of undo history, the document model, the backend and other split panes. Insert/Replace mode is unchanged.
Closes#1056
Co-authored-by: xtr-hub <242808782+xtr-hub@users.noreply.github.com>
Dragging a tab showed no floating copy: the ghost cloned the tab's classes, and .tab { position: relative } loaded after .ghost { position: fixed }. The ghost is now positioned inline, and a tab dragged within its strip stays as an empty slot the other tabs move around; the insertion line for tabs from another strip is drawn inside the edge tabs so the scroller doesn't clip it. A terminal group is now a drop target like an editor pane: the middle joins the group, an edge splits off a new group (dropZoneAt moves to ui and is shared).
Closes#1052
Co-authored-by: fuqiuluo <61898844+fuqiuluo@users.noreply.github.com>
A task's agent process lived until the task was archived or the project closed, and stopping it killed only the direct child, so wrappers (npx, codex.cmd, claude shims) left the real agent running. Agents that support session/load are now stopped after 5 minutes idle and restarted with the session reloaded on the next message; agent connections, language servers and the native Claude CLI start in their own process group and are stopped with their whole tree.
New tasks run in the local checkout unless the user picks a worktree, and worktrees go next to the project (<parent>/<project>.worktrees/<task>), on the project's drive and outside its tree.
Fixes#1011Fixes#1003
Claude Code reports the account's 5-hour and weekly usage limits (rate_limit_event) and the session cost (result.total_cost_usd); only the limit status was used. The limits now show as meters next to the context meter in the task header, with the reset time in the tooltip, and Settings › Tools › Agents › Chat turns each meter (context, work time, limits, cost) on or off. Limits are per provider and kept in memory; cost is off by default.
Closes#1057
Co-authored-by: matthiasdemauroy <295796358+matthiasdemauroy@users.noreply.github.com>
A module whose maven-compiler-plugin release/source is configured in a parent POM or in <pluginManagement> had no Java level in the build model, so its JDK fell back to the newest installed one. The level is now read from build/plugins and build/pluginManagement/plugins and inherited down the parent chain. Launching a Java application without a project JDK also takes the module level from the build model.
Refs #753
Run/Debug Configurations offers Import from .vscode/launch.json when the project has one (also a Find Action command). debugpy/python launches become Python, Django test, module or pytest configurations with their interpreter; debugpy attach becomes Python Remote Debug; node/pwa-node launches become node or npm configurations, attaches become Attach to Node.js, and node-terminal commands map to npm, a program or a shell line. cwd, env, envFile and the per-OS blocks are applied. Imported configurations join the dialog like new ones and are saved on OK/Apply; names already present are skipped, and a notification lists skipped entries with the reason and keys with no equivalent.
Refs #991
Co-authored-by: andreabasile <7715018+andreabasile@users.noreply.github.com>
A remote window sends a command to openidea-server only when the server registers it; none of the db_* commands were registered, so they ran locally, where a remote window has no project ("no project is open in this window"). The command bodies move from src-tauri into ide_host::ops::db, the Tauri commands become thin wrappers, and the remote dispatcher registers the same ops, with cancellation for execute / table page / explain. db_export_file stays local because exports are written on the user's machine.
Fixes#1019
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
* fix(menu): keep dropdowns clear of the status bar
* fix(titlebar): hide right toolbar while hamburger menu is expanded
* test(titlebar): close expanded menu through the focused element
* style(titlebar): follow existing formatter layout for mocked menu
* test(titlebar): expect expanded menu titles instead of hamburger button
---------
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
Schannel curl fails every download with CRYPT_E_REVOCATION_OFFLINE when the CRL/OCSP servers cannot be reached (corporate proxies, firewalls, some regions), although the certificate is valid, so jdtls and other tools could not be installed. On Windows, pass --ssl-revoke-best-effort when `curl --version` reports 7.70+ with Schannel as the active TLS backend: the check is skipped only when the revocation server is offline, and actually revoked certificates are still rejected. Older curl, which rejects the option, is left as is. Fixes#801.
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
oxfmt ran only below an oxfmt configuration and reused the Prettier settings. It now has its own settings group and page (path, enabled languages, formatting provider, run on save) next to Prettier. For a language with an external formatter, oxfmt formats it when enabled for that language and either an oxfmt configuration is found above the file or oxfmt is the selected provider; otherwise Prettier, when enabled. Turning Prettier off for a language still leaves it to the language server. New Reformat with Prettier / Reformat with oxfmt commands run that tool regardless of the settings and report languages it cannot format.
Co-authored-by: d0whc3r <1378986+d0whc3r@users.noreply.github.com>
* fix(run): give the spring-boot:run fork a UTF-8 default charset on Windows
On Windows, mvn spring-boot:run passes the stdout/stderr encoding properties to the
forked application JVM through JAVA_TOOL_OPTIONS, but not file.encoding (it would
change how Maven reads sources). On JDK 8-17, logback / log4j2 console appenders
encode with the default charset (GBK on Chinese Windows) and ignore stdout.encoding,
so Chinese log lines showed as U+FFFD in the UTF-8 Run console.
For a Maven launch with a spring-boot:run goal and a UTF-8 console, add
-Dfile.encoding=UTF-8 to -Dspring-boot.run.jvmArguments, merged into the user's
VM options. It reaches only the forked JVM; Maven's own JVM is unchanged. A
file.encoding the user set (in jvmArguments or JAVA_TOOL_OPTIONS) wins.
Fixes#697
* fix(run): give Gradle JavaExec forks a UTF-8 default charset on Windows
Applications started by Gradle (bootRun, run, custom JavaExec tasks) kept the
ANSI default charset (GBK on Chinese Windows, JDK 8-17) while the Run console
decodes UTF-8. Code that converts text with the default charset then breaks:
Nacos config with Chinese text fails to parse (MalformedInputException) and
logback/log4j2 console output is garbled.
For a UTF-8 console on Windows, a Gradle launch now gets an init script that
adds -Dfile.encoding=UTF-8 to each JavaExec task when it runs, unless the task
or JAVA_TOOL_OPTIONS already sets file.encoding. Gradle's own JVM and compiler
source encoding are unchanged. Maven and Gradle are told apart so the
spring-boot:run rewrite only applies to Maven.
* fix(run): set Gradle charset before build task configuration
---------
Co-authored-by: DPhngNam <pnam3072004@gmail.com>
* fix(sdk): follow scoop shims to the jdtls release script
scoop puts a jdtls.exe shim (driven by a jdtls.shim descriptor) on PATH.
launch_directly only recognized the release's own script names, so a scoop
install kept the shim's command line and the configured JDK never applied.
On Windows, follow the shim's path entry to the release script; a shim
without a descriptor is still treated as someone's own wrapper.
Fixes#794
* test(sdk): make the scoop-shim fixture a full JDK
#815 made launch_directly require javac and the jdk.compiler/jdk.jdi/
jdk.unsupported modules, so the shim fixture's bare release file stopped
the launch before the shim path was exercised.
* fix(sdk): follow real Scoop Python shims for jdtls
---------
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
* fix(terminal): start POSIX shells as login shells on Windows too
Git Bash started argument-less on Windows (resolve_shell only passed
-l on Unix), so it ran as a non-login shell: /etc/profile never ran,
git-completion.bash was not loaded, and git <Tab> had no subcommand
suggestions. Start POSIX shells with -l everywhere, like the Git Bash
shortcut, Windows Terminal and IDEA do.
A bare login bash also runs with Git for Windows' MSYS default profile,
whose PATH leaves git and the mingw64 tools out unless <Git>\cmd is
inherited; default MSYSTEM to MINGW64 (the shortcut's value) unless the
user or the launching process already set it.
* fix(terminal): keep MSYS2 login shells in the working directory
MSYS2's /etc/profile runs cd "$HOME" on a login start unless
CHERE_INVOKING is set, so with the new login args an MSYS2 bash/sh
project terminal would open in the home directory instead of the
project. Default CHERE_INVOKING=1 for Windows bash/sh next to the
MSYSTEM default, still deferring to a value the user or the launching
process set. Git for Windows' /etc/profile has no such cd (verified
locally: a login bash stays in the project directory either way), so
Git Bash users see no change.
Diff tabs, including the preview tab, get Maximize Editor / Restore Size in their toolbar (reusing Hide All Tool Windows). Show Diff from the Commit view or the Git log now carries the whole change list: F7 / Shift+F7 at the last or first difference shows a hint, and pressing again moves to the first or last difference of the neighbouring file, skipping binary, oversized and unreadable files. The layout is restored after a preview is replaced and keyboard focus stays in the diff across files. Compare branches stays single-file.
Fixes#904
Co-authored-by: kkMihai <59262001+kkMihai@users.noreply.github.com>
Conflicts left by stash pop or an automatic stash now get the same editor banner (File has unresolved merge conflicts, Resolve conflicts…) as merge conflicts. Complete conflict markers are reported as errors in every language and listed in Problems (severity under Settings › Editor › Inspections › Version Control). The toolbar shows Resolve/Continue and Abort next to the branch while conflicts or an interrupted operation exist, and the Commit view and Branches popup offer Resolve Conflicts even when no operation is in progress.
Co-authored-by: ntbang0901 <121003585+ntbang0901@users.noreply.github.com>
The result pane starts from the base and every change, conflicting or not, gets Accept / Ignore next to its side; accepting one side of a conflict still lets the other side be appended, and the modifier key resolves using that side directly. Adds Resolve Simple Conflicts (word-level), Apply Non-Conflicting Changes from the left, right or both (optionally on opening), Revert Conflict Resolution, Accept Yours / Accept Theirs / Cancel / Apply, Compare Contents between any two versions, and undo back to the conflicted state. CRLF and UTF-16 files are written back exactly, deleted sides and binary files are handled, and the commands are registered as merge.* actions.
Refs #826
Co-authored-by: ntbang0901 <121003585+ntbang0901@users.noreply.github.com>
A mapping's local path may now be an absolute path inside the project (the form IntelliJ writes), and a deployment path of . means the server root; absolute paths outside the project are kept as they are. The project-tree Deployment submenu names the default server (Upload to <server>, Download from <server>, Sync with Deployed to <server>…) and gains Sync with Deployed to Default Server. Remote Host downloads use the backend-resolved mapping, which fixes downloads under symlinked or case-different paths.
Fixes#877
Co-authored-by: florentvdry <16673235+florentvdry@users.noreply.github.com>
Blocks that were moved unchanged, re-indented or moved and then edited are paired across the split diff: each side gets a link to the other (keyboard accessible), moved lines carry a left accent bar, edits inside a moved block get word-level highlights, and the toolbar counts moved blocks and whitespace-only changes. Staging and reverting still work on the original hunks.
Co-authored-by: shohruh-genius <7658451+shohruh-genius@users.noreply.github.com>
Clone Repository gets IntelliJ's layout: a source list on the left, a version-control selector, URL history, a directory picker, and a shallow-clone option (history truncated to N commits, git clone --depth N). Entries for hosting platforms without backend support were removed.
Co-authored-by: whyareyouserious <114078441+whyareyouserious@users.noreply.github.com>
The Terminal could only split panes inside one tab. Split Right / Split Down now create tab groups, each with its own tabs, + and ▾; tabs can be moved between groups by dragging or Split and Move Right/Down, Unsplit merges a group back, and closing a group's last tab focuses the neighbouring group. Split actions are registered as actions and appear in the tab context menu.
Fixes#774
Co-authored-by: StarryKira <59300016+StarryKira@users.noreply.github.com>
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 <11207890+ravuthz@users.noreply.github.com>
The Markdown preview had a fixed 14px body and notebook Markdown cells inherited the UI font size, so enlarging them meant scaling the whole UI. Languages & Frameworks › Markdown › Preview font size (application-level, 8–72, following the editor font size until set) now drives the preview and notebook Markdown cells; fenced code scales with it, and notebook code cells and outputs follow the editor font.
Fixes#768
Co-authored-by: jaw52 <53564781+jaw52@users.noreply.github.com>
Python puts the working directory first on sys.path for -c / -m. venv creation, pip (list/install/uninstall), the library-path probe and the IDE-managed Python tool probes and installs inherited the IDE's own working directory, so a project's typing.py / logging.py there broke them ("partially initialized module 'typing'"). venv creation now runs python -I; pip, the library probe and the tool commands run in an empty private directory (keeping pip install --user packages visible), and relative requirements (., ./pkg, .[dev]) resolve against the project root. The window-label part of the original change already landed in #960.
Refs #955
Co-authored-by: pcastro93 <2058578+pcastro93@users.noreply.github.com>
In Preview only the editor is display:none and the preview only followed it in the split layout, so Structure's Autoscroll to Source moved a hidden caret and Autoscroll from Source never saw the preview scroll. Caret moves without an edit now scroll a Preview-only preview to that line, and the preview publishes the source line on screen through editor/api (setShownLine/useShownLine) so Structure selects the heading being read; it is cleared when the editor comes back or the preview closes.
Fixes#977
Co-authored-by: jaw52 <53564781+jaw52@users.noreply.github.com>
Branch and tag labels in the commit details pane inherited the graph row's 180px single-line limit and were clipped. Labels in the details metadata now use the pane width and wrap, including names without break points; graph-row labels keep their single-line ellipsis.
Fixes#996
Co-authored-by: guozhiqiang123 <20378988+guozhiqiang123@users.noreply.github.com>
WebKit fires `dblclick` only for the second click of a burst. A quick
double-click right after another arrives as `click` with `detail` 3 and 4 and
no `dblclick`, so the second double-click on a folder in the Project view (and
on any tree row, tab or list item handled by `onDoubleClick`) did nothing.
A window-level shim now dispatches a `dblclick` on the click's target for every
later even click of a burst (detail 4, 6, ...), after the page's own click
handlers. Existing `onDoubleClick` handlers and their `stopPropagation` apply
unchanged; an engine that fires its own `dblclick` for that click gets it
dropped as a duplicate.
The Project view test now delivers bursts the way WebKit does (`dblclick` only
at detail 2) and checks both expand modes over 2–6 click bursts.
Refs #724
- List rows are no longer text-selectable. Settings pages enable text selection
for their content, and the shared List inherited it, so dragging or ⌘A
highlighted every row.
- An agent missing from PATH is shown as a muted row with a small "Not found"
label instead of an orange warning badge; a custom executable gets an accent
"Custom" label.
- The details header puts the status (with a status dot) and the launch command
on one line.
- Reset only appears when a custom executable is set.
- The sign-in hint gets its own "Sign in" heading.
- The list/details frame sizes to its content instead of filling the page.
The official Kotlin LSP answers navigation into a dependency with a
`jar:///…/lib.jar!/pkg/Name.class` URI but does not serve its contents through
`workspace/textDocumentContent`, so opening the target failed with "no
language server provides the contents of jar:…", locally and remotely.
`jar:` URIs are now resolved by the backend, where the jars are:
- An entry that is not a class (a sources jar entry, a resource) is read from
its archive.
- A class is shown as its source from the artifact's `-sources.jar` in the
Maven repository or the Gradle cache (`Name.java`/`Name.kt`, an inner
class's outer file, a `FileKt` facade's `File.kt`, a `@file:JvmName`
file); Go to Declaration then lands on the declaration of the name it was
invoked on.
- Otherwise the server's `decompile` command is used when it offers one, and
the error says that neither a source nor a decompiler is available.
- Library tabs name the archive and package (`spring-boot-3.5.16.jar ›
org.springframework.boot`).
Fixes#982
jdtls only checks the syntax of a Java file outside every module's source
folders (or outside any imported project) and answers navigation with
nothing. The editor then said only "Cannot find declaration to go to", and
the server's explanation was a warning squiggle on line 1.
- A banner above the editor explains it while the server reports it, for
jdtls (not on the classpath of a project / non-project file), gopls (no
package, build constraints — this replaces the Go-only banner) and
rust-analyzer (unlinked file). For Java it offers the Maven / Gradle reload
when available.
- Go to Declaration / Implementation / Type Declaration that finds nothing
in such a file gives the reason in its hint.
Fixes#827
The profiles checked in the Maven tool window only reached `mvn` (sync and
goal runs) and the Kotlin bridge. jdtls imported every module without them,
so a dependency whose version, or module, comes from a profile stayed
unresolved and its classes were reported as missing, while Build Project ran
without `-P` as well.
- A Maven sync records the profiles it ran with. jdtls gets a copy of the
user settings.xml with them in `<activeProfiles>` (m2e has no other input
for profiles), sent as a setting so a running server re-imports instead of
restarting; the file name follows its content so m2e reloads it.
- Toggling a profile reloads the Maven projects (once for several toggles in
a row, again after a sync that started with the previous choice).
- Build Project passes the synced profiles with `-P`.
- When jdtls finishes an import or project update it does not re-check open
files, so their errors stayed until the next edit; open Java files are now
re-checked on its ProjectStatus report. This also covers a POM edit or a
dependency added to ~/.m2.
Fixes#993
Settings › Tools › Agents is split into Built-in and ACP Registry tabs. The Built-in tab lists every built-in agent with its status (custom executable or not found on PATH) and, for the selected agent, its default launch line, an Executable field (empty = auto-detect on PATH) with Browse and Reset, and its sign-in hint. Edits are staged with the Settings dialog and written on Apply / OK; Cancel discards them.
The override is written to <config>/agents/<id>.toml, keeping hand-written name, driver, args and env; resetting a command-only override removes the file. AgentProvider now exposes the program, args, built-in defaults, login hint and whether the command is editable: built-ins started through a package runner (npx) and registry installs are read-only, with the reason shown.
Fixes#596
Co-authored-by: a-rookie-of-C-language <186315048+a-rookie-of-C-language@users.noreply.github.com>
Adds a PR template (linked issue, what and why, how it was verified, review focus), a minimal reproduction field on the bug form and required acceptance criteria on the feature form, and CONTRIBUTING.md with the issue and PR rules. A PR check labels needs-info and leaves one comment when the issue link or the verification section is missing; it only reads the title and description.
AGENTS.md is now tracked, with CLAUDE.md linking to it, and covers layout, rules, commands, the checks CI runs, pitfalls and parallel worktrees. The verify and issue-fix skills are shared under .agents/skills and linked from .claude/skills; the rest of .claude/ stays local.
Limit grid and export-preview cell text to 1000 UTF-16 code units plus an ellipsis. Edit large values in a wrapped dialog while preserving complete clipboard, export and database values.
Reproduced STATUS_STACK_OVERFLOW (0xC00000FD) in Windows WebView2 against a real PostgreSQL table with nine rows and 8.83 MB of jsonb data. Verified the fix through display, selection, editing, full clipboard copy, JSON file export and PostgreSQL writeback. All PR CI checks passed, including 5625 frontend tests and Rust checks on Windows, macOS and Linux.
Fixes#825
On Windows, a project file named after a device (nul, con, aux, prn, com1-9, lpt1-9; nul is typically left by a 2>nul redirect in Git Bash) aborts the Java import: Files.exists reports it missing, Eclipse's refresh builds a child path with an empty name and throws "Path must include project and resource name", so m2e never creates the Java project.
When JDT LS reports that failure on Windows, the server root is scanned for reserved device names (bounded, ignore files not honoured) and a one-time error notification lists them with how to recover. The name check covers COM0-9/LPT0-9 including superscript digits, case-insensitive, with or without extension or trailing spaces. Backend notification texts now go through txTemplate so templated catalog entries translate; zh text added.
Refs #976
Maven launches put a snapshot of each module's class directory on the classpath in place of the original. The originals were built as module.join("target/classes") (`...\app\target/classes` on Windows) while JDT LS reports `...\app\target\classes`; compared as text, the original stayed next to its snapshot, so every resource was on the classpath twice and MyBatis failed with "XML fragments parsed from previous mappers already contains value". Launch class directories are now joined per component, and classpath entries are compared by a key that ignores the verbatim prefix, separators, trailing separators and, on Windows, case.
Fixes#704
Tab switches and mode toggles unmount MarkdownPreview, whose only
scroll state was the DOM scrollTop; Preview Only started over at the
top. Record the preview's own scrollTop in the per-path live markdown
store and restore it on mount (split mode keeps its editor-driven
sync).
* fix(markdown): use editor-side tokens in the preview
The preview paints on var(--bg-editor) but took its text colors from
the UI tokens (--fg, --fg-strong, --fg-secondary), so with a light UI
theme plus a dark editor scheme (e.g. an imported dark .icls) it
rendered black-on-black. The code editor was unaffected because it
uses --fg-editor.
The preview now uses editor-side tokens: text var(--fg-editor),
headings var(--syn-markup-heading), code/pre/th background
var(--bg-editor-inline), blockquote var(--fg-editor) at 80%.
* chore: oxfmt src/features/run/types.ts (fixes format:check broken on main by 002401e8)
* chore: regenerate bindings after rebasing onto main
`@{u}..HEAD` fatally fails when the tracking config remains but the
remote-tracking ref is gone (the upstream branch was deleted on the
remote and the fetch pruned): the repository was dropped from the
multi-repository Push dialog, or, when every repository failed, the
dialog showed the raw fatal and disabled Push. Fall back to recent
history like the single-repository dialog does, so the branch stays
listed and pushable (a plain push recreates the remote branch).
The overview lane lies over the vertical scrollbar, so a mark the height of the lane (a new file added to git is one green VCS mark) hid the native thumb completely. The lane now draws the thumb itself, above its marks and without taking the pointer, at the position of the native one (scrollbarThumb, inset by its 2px border); it switches to the hover color while the pointer is on the thumb. The native thumb is made transparent in editors that have the lane, so the two don't add up, and still takes the drag. Editors without the lane (notebook cells, previews) keep the native thumb.
Fixes#1014
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A recursive symbol got a fresh id on every level (ids carry the parent
chain), so sibling dedupe never matched and each expansion requested the
same level again, forever. Identify items by their position key instead:
a child whose key repeats an ancestor is kept a leaf, marked recursive,
and never expanded.
In a workspace the tree's root is the workspace directory, and the linked projects are separate top-level roots outside it. Select Opened File (and other reveals) only looked inside that root and contributed folders, so a file of a linked project was never expanded, selected or scrolled to. Reveal now resolves the innermost linked project containing the path first; the Project Files pane also counts linked projects as shown.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Any ``` line closed an open fence, so a ```` block showing a ``` block ended early and the inner code was spell checked as prose. Fences now close as in CommonMark: the same character, at least as long as the opening fence, and no info string.
Co-authored-by: fhrmrk <81476595+fhrmrk@users.noreply.github.com>
Comments and Markdown went through the string-escape rule, which drops only the letter after a backslash, so the Doxygen command `\brief` was reported as the typo "rief". Outside string literals the whole word after a backslash is skipped now (Doxygen / TeX commands, escapes quoted in comments); string literals still skip only the escape letter.
Co-authored-by: fhrmrk <81476595+fhrmrk@users.noreply.github.com>
A request URL without a scheme defaults to http://, but the check was `contains("://")`, so `GET localhost:8080/login?next=http://…` was sent as is and failed to parse. Only a `://` after a leading RFC 3986 scheme counts now, both for HTTP Client requests and the proxy connection check.
Co-authored-by: fhrmrk <81476595+fhrmrk@users.noreply.github.com>
`/<token>/<key>` has no segments after the key, so the trailing-slash flag kept its initial value and the root's index.html was served at a URL without the slash; its relative links then resolved above the root and returned 404. The flag now comes from the path, so the root redirects to `/<token>/<key>/` like any other directory.
Co-authored-by: fhrmrk <81476595+fhrmrk@users.noreply.github.com>
After Reformat Code rewrote a file's indents (e.g. spaces to tabs per Code Style), the editor kept the indents detected when the file was opened: tabs rendered at the old width, the status bar showed the old indent, and Enter inserted the old indent, mixing tabs and spaces until the file was reopened.
Whole-file format edits from language servers and contributed formatters are now tagged `format.reformat`, and after such a transaction the document runs the open-time indent detection again and reapplies its settings, so every split picks up the new indent.
Refs #935.
Co-authored-by: jaw52 <53564781+jaw52@users.noreply.github.com>
xtr-hubandxtr-hub
fuqiuluoandfuqiuluo
l0ng-ai
matthiasdemauroyandmatthiasdemauroy
Andrea Basileandandreabasile
arookieofcanda-rookie-of-C-language
Bl0ckandl0ng-ai
kevinhandl0ng-ai
d0whc3r
kkmihaiandkkMihai
ntbang0901
Florent Vandflorentvdry
Shohruh Ismoilovandshohruh-genius
whyareyouserious
时雨遥
Ravuthzandravuthz
Jaw
Piettro
小小强andguozhiqiang123
Marius Ciepluch
Park Jong Hyun