[qt/qt-creator/qt-creator]: Summary of bulk changes made

Git repository change summary for qt/qt-creator/qt-creator
Pushed by mirror-service into branch 'master'.
Changed from fd4d4ca725f2967af40f12db286cf1d0be316027 to c1888cd27a6e0c277f0740062d26fed6c0104289
Acknowledgement was received that this change introduces only existing code that has been pushed to another public open source repository.

This change contains the following new commits:

Git commit bb56c85bc16bc67a8a18ea53e9f3ecb7fdcdb8c8 by Jarek Kobus on 23/07/2026 at 07:00..
Debugger: Fix Lldb hang from DEBUGINFOD_URLS env var

CreateTarget() hangs indefinitely if this is set and the
debuginfod server is unreachable (common offline/firewalled).
Unset rather than clear - some clients treat empty and absent
differently.

Change-Id: Ia399fc7ce86d947fbb6a34466aeeb66d706aa9c8
Reviewed-by: hjk <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/bb56c85bc16bc67a8a18ea53e9f3ecb7fdcdb8c8

Git commit a4beeb4d634e3c1c11c4711b568882951733ab53 by Jarek Kobus on 23/07/2026 at 07:24..
LldbEngine: Fix watchpoint insertion via insertBreakpoint()

Two bugs made every watchpoint insert fail: calling
SetScriptCallbackFunction()/SetOneShot() unconditionally raised
AttributeError for SBWatchpoint (only SBBreakpoint has them,
aborting before any reply was sent), and the trailing
describeError() fragment was concatenated onto describeBreakpoint()'s
reply without the separating comma GdbMi parsing needs.

Change-Id: I40559679030853a19473e93b4630d9f8c05893db
Reviewed-by: hjk <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/a4beeb4d634e3c1c11c4711b568882951733ab53

Git commit 8699f2a38a036a38a1efc002793cdd1b8e3afc59 by Jarek Kobus on 23/07/2026 at 07:24..
LldbEngine: Fix catch fork insertion via insertBreakpoint()

hasCapability() already claims CatchCapability, but
insertBreakpoint() had no BreakpointAtFork case, so it fell
into the generic pending/no-op branch. Break on the libc entry
point instead.

Change-Id: I2c890e6ce237a7e16703a2a9c7345caec473fd8e
Reviewed-by: hjk <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/8699f2a38a036a38a1efc002793cdd1b8e3afc59

Git commit 5efc22c79ac0544fd6c3002c14b2a0ba9a152b38 by hjk on 23/07/2026 at 08:53..
Debugger: Report signal/exception stops under LLDB

A crash such as a bad pointer access stopped the inferior under LLDB
with only a bare "Stopped" status; unlike GdbEngine, nothing surfaced
the signal. Have the LLDB bridge emit a signal-received notification on
a spontaneous signal/exception stop (with signal name and description),
and have the engine show the status message and, if enabled, the
"Signal Received" dialog, mirroring the GDB path.

Fixes: QTCREATORBUG-11049
Change-Id: Ibc26c6059b5b0f4427bbcfee6cd854e0c1c4bc62
Reviewed-by: Jarek Kobus <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/5efc22c79ac0544fd6c3002c14b2a0ba9a152b38

Git commit 960b3b529041397d8b93c44454c6f9b418aad704 by hjk on 23/07/2026 at 08:54..
ProjectExplorer: Honor the prompt-to-stop setting when closing a run tab

Closing a running application's output tab unconditionally called
promptToStop(), ignoring any "don't ask again" choice the user may have
made.

Route closeTab() through the prompt-to-stop machinery, but back it with
its own setting (PromptToStopOnCloseTab, default on) instead of sharing
the stop button's PromptToStopRunControl. Closing a tab is often a
clean-up action, and silently killing a still-running application on tab
close is more surprising than when explicitly stopping it, so the two
"don't ask again" choices are kept independent.

For that clean-up case, also add a "Close All Non-Running Tabs" entry to
the tab context menu, alongside "Close All Tabs" and "Close Other Tabs".

Assisted-by: Claude Code
Change-Id: Id59d6fbb94a9208b058e7cd4ffc40b1a8b305424
Reviewed-by: Eike Ziller <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/960b3b529041397d8b93c44454c6f9b418aad704

Git commit 76b468366737de3c3761a45d0c5d4b63646a7bee by hjk on 23/07/2026 at 09:00..
TextEditor: Return a plain widget from createProjectEditor

Add reusable createCodeStylePreview()/createCodeStylePreviewNote()
helpers and have createProjectEditor() (and the default per-project
editor) return a plain QWidget instead of a CodeStyleEditor. The
per-project panel only ever used it as a widget, and per-project settings
apply immediately, so no apply/cancel contract is needed there. The
settings page preview now uses the same helpers.

Assisted-by: Claude Code
Change-Id: Ic57b8709b71cf69fa176ed977831343dc38b9cc1
Reviewed-by: Christian Stenger <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/76b468366737de3c3761a45d0c5d4b63646a7bee

Git commit d1d0716dbcd66c4286d1c0fa7c2d4f1fcc28b4a9 by hjk on 23/07/2026 at 09:41..
FakeVim: Test that a Control modifier is not a plain command

FakeVim already leaves keys that carry the Control modifier (without
Alt) for Qt Creator shortcuts instead of matching them as plain letter
commands (see isAcceptableModifier), so e.g. Ctrl-Shift-X does not act
as "X". Add a regression test guarding that.

The remaining part of the report, Ctrl-Shift-U inserting a "u", is the
desktop input method entering Unicode code-point mode (IBus and
similar); it never reaches the application and is outside FakeVim.

Fixes: QTCREATORBUG-14369
Assisted-by: Claude Code
Change-Id: I4dcfb54c29cbab10c65b491d84fd8e426d8d495d
Reviewed-by: Cristian Adam <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/d1d0716dbcd66c4286d1c0fa7c2d4f1fcc28b4a9

Git commit 68eecaeb8c679588b69e789b689996890f1c419b by hjk on 23/07/2026 at 09:41..
FakeVim: Test the visual tab width matches the space width

FakeVim sets the editor tab stop distance to the tabstop times the space
width, so that tab-, space- and autoindent-indented lines line up. There
was no regression guard for that. Assert that the tab stop distance
equals the space width times the tabstop, for two tabstop values.

Fixes: QTCREATORBUG-10367
Assisted-by: Claude Code
Change-Id: I4d6c0f3d4ce3067a479d49cc747d9d5a006ae41c
Reviewed-by: Cristian Adam <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/68eecaeb8c679588b69e789b689996890f1c419b

Git commit c066f62e8efd399d2959d0c3743f9d44ff65e6ac by hjk on 23/07/2026 at 09:41..
FakeVim: Implement "gd" natively for Follow Symbol

Route "gd" through the same request as CTRL-] / :tag so users no
longer need to map a :gd Ex command. The mapped Ex command did not
reach a definition in the same file.

Assisted-by: Claude Code

Fixes: QTCREATORBUG-27191
Change-Id: I7eb42af5f70f7bc4eac1cc104457231c2c18c087
Reviewed-by: Cristian Adam <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/c066f62e8efd399d2959d0c3743f9d44ff65e6ac

Git commit eb73de204524406c2ccef77f436cf00ad728cb4b by hjk on 23/07/2026 at 10:10..
CppEditor: Expand doc comment for destructor declarations

Typing "/*!" before a destructor produced only an empty comment
continuation instead of a "\brief ~Foo" doc comment.

The generator bailed out early because its guard for the first
character of the declaration accepted only a letter, '_' or '[',
but a destructor starts with '~'.

Fixes: QTCREATORBUG-33584
Assisted-by: Claude Code
Change-Id: Ie8518d3b8b3888ba9de49f5de55cd069c06b74d6
Reviewed-by: Christian Kandeler <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/eb73de204524406c2ccef77f436cf00ad728cb4b

Git commit 170ad5b409079c491f7beea519caedb7caf9e7ce by hjk on 23/07/2026 at 10:10..
RemoteLinux: Test and tidy the "Access via" ssh -J plumbing

Task-number: QTCREATORBUG-30629
Assisted-by: Claude Code
Change-Id: I6bbb5abd91e87e9def64e1e56255aa20a259ed98
Reviewed-by: Christian Kandeler <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/170ad5b409079c491f7beea519caedb7caf9e7ce

Git commit 0edab689da8f2f9b79c764bd91f9e730b2a36d06 by hjk on 23/07/2026 at 10:25..
CppEditor: Shift doc comment body when reindenting a selection

Auto-indenting a selection (Ctrl+I) that contains a documentation
comment moved the opening "/*!" line to its new indentation, but
left the body and closing lines ("* \brief ...", "*/") at their
old column, breaking the comment's internal alignment.

The code formatter intentionally keeps a multi-line comment's
continuation lines at their current column instead of recomputing
it, since comment interior text is free-form. Extend
CppQtStyleIndenter::indent() to shift those lines by the delta
applied to the comment's opening line instead of leaving them
untouched.

Fixes: QTCREATORBUG-33585
Assisted-by: Claude Code
Change-Id: Ide1c240189d5c27a845703a31589e87244c50991
Reviewed-by: Christian Kandeler <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/0edab689da8f2f9b79c764bd91f9e730b2a36d06

Git commit 8b6aef454cac07a8a0cd42097d5470253b803581 by David Schulz on 23/07/2026 at 10:36..
QtSupport: use osType specific possible qmake commands

The default filter qmake* and qtpath* matches qmake.pdb on windows which
is not a valid executable qmake. Testing those candidates takes
unnecessary long in remote detection.

Change-Id: I8a167d165215ecf5f76ce2f4224b9d1117ce0e19
Reviewed-by: Christian Kandeler <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/8b6aef454cac07a8a0cd42097d5470253b803581

Git commit 83bb7b6a6016ebafa002e8ccb3fbe7e033a36980 by Christian Kandeler on 23/07/2026 at 10:36..
Plugins: Clean up includes

Resulting from a run of clang-include-cleaner, followed by manual
corrections.

Change-Id: I134c4d24519122246e7ec7a9f4fded7ba344f2b7
Reviewed-by: hjk <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/83bb7b6a6016ebafa002e8ccb3fbe7e033a36980

Git commit fef1b5e06527f10e54798a3e00c93d4542b1499c by hjk on 23/07/2026 at 10:47..
ClangFormat: Make the project code style editor a plain widget

Rebuild ClangFormatProjectEditor as a plain QWidget using the shared
createCodeStylePreview()/createCodeStylePreviewNote() helpers, which the
factory now feeds via the C++ snippet group and preview text. The
apply()/cancel() overrides are dropped: in the project view the global
config widget writes through immediately and its apply/cancel are no-ops.

Assisted-by: Claude Code
Change-Id: Iddaa55f68abeb10893fb354888fdafdc6b15e557
Reviewed-by: Christian Stenger <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/fef1b5e06527f10e54798a3e00c93d4542b1499c

Git commit c96f27f465450ff5f0708bb84a2bdf414abe3417 by hjk on 23/07/2026 at 10:48..
FakeVim: Keep cursor moves made by ex-mapped actions

An Ex command mapped to a Qt Creator action runs while m_inFakeVim is
set, so onCursorPositionChanged does not flag a cursor move the action
makes. pullCursor() then kept the stale position and the trailing
commitCursor() reverted the editor, undoing the move. Force the re-read
outside visual mode, where commitCursor() has extended the selection
for display and it must not be pulled back.

Assisted-by: Claude Code
Task-number: QTCREATORBUG-27191
Change-Id: I383587da330f608d1380d8afe7660b71afbffbb0
Reviewed-by: Cristian Adam <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/c96f27f465450ff5f0708bb84a2bdf414abe3417

Git commit 41711e426cbf7fe6824ca17fbd60f34ad971a1fc by hjk on 23/07/2026 at 10:58..
Core: Add Util.versionCompare() for reliable version checks

JSON wizard expressions compared Qt versions with JavaScript's own
operators, e.g. value('MinimumSupportedQtVersion') > '6.2'. That is
wrong for two-digit minor versions: as strings "6.10" < "6.2", and as
numbers 6.10 < 6.2 -- so anything gated on "> 6.x" silently misbehaves
from Qt 6.10 on.

Add a Util.versionCompare(v1, v2) JS helper backed by QVersionNumber
(returns <0, 0, >0) and use it for the Qt-version checks in the
autotest wizards.

Fixes: QTCREATORBUG-33685
Assisted-by: Claude Code
Change-Id: I1f50f7a3518e1fbccee971485e0e149873db3ede
Reviewed-by: Eike Ziller <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/41711e426cbf7fe6824ca17fbd60f34ad971a1fc

Git commit e4c1618f2f545490c6b93ebd20ae11b6f9540056 by pentek.imre (on behalf of Imre Péntek) on 23/07/2026 at 11:06..
QtSingleApplication: Don't hang forever when peer never acks

QtLocalPeer::sendMessage() waits for the receiving instance to
acknowledge the message with a timeout, but when blocking mode is
requested (-block) it then calls waitForDisconnected(-1)
unconditionally, even when the ack never arrived.

Connecting to a QLocalServer succeeds at the OS level (kernel listen
backlog) even when the peer never services its event loop. This
happens in practice: every process, including pure -client
invocations, registers itself in the instance list in the
QtSingleApplication constructor, so a client that is itself blocked
inside sendMessage() (or an otherwise frozen instance) can be selected
as the message target by the next -client invocation. The new client
then writes the message, times out waiting for the ack after 5 s, and
hangs forever in waitForDisconnected(-1) with no feedback. The
non-blocking variant correctly reports the failure and lets main()
offer the "Could not send message" retry dialog; the blocking variant
never gets there.

Only enter the blocking wait when the message was actually
acknowledged.

Steps to reproduce:
  qtcreator &                        # instance A
  kill -STOP <pid of A>              # stand-in for a blocked/frozen
                                     # first instance
  qtcreator -client -block file.txt  # hangs forever instead of
                                     # failing after the 5 s timeout

Task-number: QTCREATORBUG-34761
Change-Id: I0e545216743698daf6aec85721867b23e457120b
Reviewed-by: Eike Ziller <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/e4c1618f2f545490c6b93ebd20ae11b6f9540056

Git commit 090903bbbddc953337f2097672a2e6f513858c5e by hjk on 23/07/2026 at 11:15..
Debugger: Recognize gdb debuginfod downloads in the watchdog

The command watchdog is reset on every read from gdb's stdout, so the
"GDB Not Responding" dialog fires only when gdb stays silent on stdout
for the whole timeout. When gdb fetches debug info via debuginfod it
announces the download with a single MI console line and then downloads
silently, producing no output until it finishes. A download slower than
the watchdog timeout therefore looked like a hang, and stopping the
debugger from that dialog produced a confusing "crashed" message.

Recognize the "Downloading ... separate debug info" console line, show
it in the status bar, and remember that a download is in progress so
that, if it stalls past the timeout, the watchdog dialog explains it is
a debuginfod download (possibly a slow or unreachable server) instead
of claiming gdb is not responding.

Assisted-by: Claude Code
Fixes: QTCREATORBUG-34811
Change-Id: I813de143a6d9fa58d8334ca55a994542b5861d61
Reviewed-by: Jarek Kobus <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/090903bbbddc953337f2097672a2e6f513858c5e

Git commit c7f286a7c602cc35a1042142fd93160d51559c9c by hjk on 23/07/2026 at 11:21..
TextEditor: Remove the now-unused CodeStyleEditor::setupPreview

Its two callers (the default and ClangFormat per-project editors) now
use the free createCodeStylePreview()/createCodeStylePreviewNote()
helpers.

Assisted-by: Claude Code
Change-Id: Ife9d1d838166ec01209c01da61af5a525b6d8669
Reviewed-by: Christian Stenger <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/c7f286a7c602cc35a1042142fd93160d51559c9c

Git commit bba247aeeb484242a9d71a639fdec8d4f9030bf2 by hjk on 23/07/2026 at 11:21..
Debugger: Test writeMemory in the LLDB dumper test

Add a "WriteMemory" case to tst_dumpers that overwrites a local via
the writeMemory bridge command and checks the fetched value. The
test harness resolves the variable address and issues the write
before fetching variables; a new MemWrite dumper option drives it.
LLDB only, as only the LLDB harness performs the memory write.

Assisted-by: Claude Code
Change-Id: I84da139fd822535f4c3b522c2b46b58507d81961
Reviewed-by: Jarek Kobus <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/bba247aeeb484242a9d71a639fdec8d4f9030bf2

Git commit 88bba2849e462a34ebf97eac2d4bc798db4aa5ff by hjk on 23/07/2026 at 11:30..
FakeVim: Match Vim when repeating a visual paste with "."

Repeating a paste over a visual selection with "." produced mangled
text: the dot command was recorded without the visual re-selection,
so "." ran as a plain normal-mode paste.

In Vim the redo of a paste over a visual selection records only the
deletion of the selection, not the put text. So "." re-selects a
same-sized region and removes it without inserting anything (the
deleted text going to the unnamed register). Record the dot command
that way. Verified against Vim 9.1 for character- and line-wise
selections.

Assisted-by: Claude Code
Fixes: QTCREATORBUG-18298
Change-Id: Iab8a311490fe53bb5077afe0968d62744e9c0c0f
Reviewed-by: Cristian Adam <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/88bba2849e462a34ebf97eac2d4bc798db4aa5ff

Git commit 0c2a8a6b6f9eb7a63149ad1d542268d9fdb4f157 by hjk on 23/07/2026 at 11:40..
TextEditor: Retire createSettingsEditor, unify on value editors

Every code style page now goes through setValueEditorCreator. A plain
value editor edits the preferences live and CodeStyleAspect wraps it
with the shared selector and preview; a value editor that derives from
CodeStyleEditor (only ClangFormat, whose settings live in .clang-format
files and the global settings) lays out its own selector and manages its
own deferred apply/cancel, which CodeStyleAspect detects via
qobject_cast and routes to.

CodeStyleEditor is slimmed to that self-managed contract; its layout
helpers, createSettingsEditor and setSettingsEditorCreator are removed.

Assisted-by: Claude Code
Change-Id: I363c20ca17218da03255c93e2cd902a5898fd247
Reviewed-by: Christian Stenger <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/0c2a8a6b6f9eb7a63149ad1d542268d9fdb4f157

Git commit 4088aa1564dc4044aad630bce8855243493a08ab by Jarek Kobus on 23/07/2026 at 11:49..
LldbEngine: Support plain remote-server attach without a platform

AttachToRemoteServer/AttachToRemoteProcess with no lldbPlatform set
(e.g. the generic "Attach to Running Debug Server" dialog) fell
through to "Unsupported platform" - connect via the raw gdb-remote
plugin instead, like a plain gdbserver attach.

Change-Id: I57e4c7f68bd503277a4c985198aabcef760f20fc
Reviewed-by: hjk <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/4088aa1564dc4044aad630bce8855243493a08ab

Git commit c1888cd27a6e0c277f0740062d26fed6c0104289 by Christian Stenger on 23/07/2026 at 11:50..
SquishTests: Fix handling of save as dialog

The respective label starts invisible, but gets visible on
show. Do not use the first as it is handled as if a different
object.

Change-Id: Iae3e82adc282e651148dc570084007c639ac8e8e
Reviewed-by: Jukka Nokso <[email protected]>
Reviewed-by: Christian Stenger <[email protected]>
https://invent.kde.org/qt/qt-creator/qt-creator/-/commit/c1888cd27a6e0c277f0740062d26fed6c0104289