[qt/qt/qtbase]: Summary of bulk changes made

Git repository change summary for qt/qt/qtbase
Pushed by mirror-service into branch 'dev'.
Changed from b687b72dbffc3cc60db6ffe3d55ec7b08b30efe2 to 780ef39c3bcdfc7a80093a558b6008e2010829af
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 84b1dc8c26119b5419ffe961b6406ee388876870 by Dennis Kim on 31/07/2026 at 12:58..
ohos: Fix window position after destroy() and recreation

QOhosPlatformWindow::initialize() stored the initial geometry through
the top-level QPlatformWindow::setGeometry(), bypassing the client/frame
conversion that QOhosPlatformWindow::setGeometry() performs based on
QWindowPrivate::positionPolicy. When a position is requested before the
platform window exists (setFramePosition() then show()), the policy is
WindowFrameInclusive and the requested top-left is frame-relative, so
QPlatformWindow::geometry() ended up holding a frame-relative position
instead of the client one.

The problem was masked on the first show(): the native layout reported
the real client geometry back through handleGeometryChange(), correcting
QWindowPrivate::geometry. After destroy() the core resets the policy to
WindowFrameExclusive and the stored geometry already holds client
coordinates, so on the next show() nothing corrected it and
QWindow::position() no longer matched the value seen before destroy().

Call QOhosPlatformWindow::setGeometry() explicitly so the client/frame
conversion is applied and m_lastRequestedWindowFrameGeometry is derived
the same way as during a normal geometry change. The class is named
explicitly to reach this level directly rather than the
QOhosFloatingWindow override, which would additionally emit a screen
change from within window creation.

Task-number: QTBUG-148052
Task-number: QTBUG-148261
Pick-to: 6.12
Change-Id: I5b8268362dc3c8a5b9894efe38d95f44facd8218
Reviewed-by: Liang Qi <[email protected]>
Reviewed-by: Zbigniew Chyla <[email protected]>
Reviewed-by: Seokha Ko <[email protected]>
Reviewed-by: Dawid Śliwa <[email protected]>
https://invent.kde.org/qt/qt/qtbase/-/commit/84b1dc8c26119b5419ffe961b6406ee388876870

Git commit 3ff390109260e05ca0807e0fb55f763b96a8e699 by Eskil Abrahamsen Blomfeldt on 31/07/2026 at 14:39..
wayland: Don't send wl_data_offer.finish on version 1 offers

QWaylandDataDevice::data_device_drop() called wl_data_offer.finish()
unconditionally when accepting a drop from another client. The finish
request only exists since wl_data_device_manager version 3, so calling
it would trigger a protocol error on compositors supporting v1 or v2
(e.g. Qt Wayland Compositor).

Fixes: QTBUG-137533
Pick-to: 6.11 6.12
Change-Id: Iedcfaccbe28418ffea388994e1b0dd70d47d33a9
Reviewed-by: David Edmundson <[email protected]>
https://invent.kde.org/qt/qt/qtbase/-/commit/3ff390109260e05ca0807e0fb55f763b96a8e699

Git commit e080cd4c9dd394f6a587b9962640947c4bdc7002 by SanthoshKumar Selvaraj (on behalf of Zbigniew Chyla) on 31/07/2026 at 16:52..
ohos: Extras: make OnContinue responses read as commands

OnContinueContext exposed the continuation outcome through set-prefixed
names, two of which took no argument (setRejectResponse(),
setMismatchResponse()). A setter with no argument is misleading: these
are not property writes but commands that select one of the platform's
three OnContinueResult outcomes (AGREE, REJECT, MISMATCH).

Rename them to command verbs:

 - setAgreeResponse() => sendAgreeResponse()
 - setRejectResponse() => sendRejectResponse()
 - setMismatchResponse() => sendMismatchResponse()

Task-number: QTBUG-147194
Pick-to: 6.12
Change-Id: I483a5acc20decc7828114139dbf6c08b971c6765
Reviewed-by: Marc Mutz <[email protected]>
Reviewed-by: Liang Qi <[email protected]>
https://invent.kde.org/qt/qt/qtbase/-/commit/e080cd4c9dd394f6a587b9962640947c4bdc7002

Git commit ebeca6b9490146c3f46b52b8a317f2d60ff8c595 by Tor Arne Vestbø on 31/07/2026 at 16:52..
visionOS: Use the Swift driver to generate the bridging header

We were invoking swiftc with -frontend, which bypasses the driver.
The driver is what works out the macro plugin search paths for the
target platform, passing -plugin-path, -external-plugin-path, and
-in-process-plugin-server-path on to the frontend.

Without them the frontend can't expand any Swift macros, which broke
as of the Xcode 27 SDKs, where SwiftUI's @State is a macro rather
than a property wrapper:

    error: external macro implementation type 'SwiftUIMacros.StateMacro'
    could not be found for macro 'State()'; plugin for module
    'SwiftUIMacros' not found

Passing the plugin paths ourselves is not much of an option, as
SwiftUIMacros lives in the platform directory rather than the
toolchain, so it needs the per-platform -external-plugin-path with
the plugin server path appended. We let the driver work it out.

Pick-to: 6.12 6.11
Change-Id: I98ba5c3bc8d9a9d0ff8f4b0caf3a28f6e3e49090
Reviewed-by: Alexandru Croitor <[email protected]>
https://invent.kde.org/qt/qt/qtbase/-/commit/ebeca6b9490146c3f46b52b8a317f2d60ff8c595

Git commit bafd17e0c8dbb51a970a22f79c9e1fa3e1aa9e78 by Andy Nichols on 31/07/2026 at 17:16..
rhi: gles2: Synchronize compute-written indirect draw buffers

drawIndirect() and drawIndexedIndirect() did not register the indirect
buffer with the pass resource tracker, and the blanket buffer barrier
mask emitted at pass start did not include GL_COMMAND_BARRIER_BIT. As
a result, when a compute pass wrote draw arguments into a buffer that
a later render pass consumed via an indirect draw, no glMemoryBarrier
covering indirect command sourcing was ever issued and the draw could
read stale arguments.

Register the indirect buffer with BufIndirectDraw access when
recording the draw, mirroring the Vulkan backend, and add
GL_COMMAND_BARRIER_BIT to the buffer barrier mask. The new
AccessIndirectDraw usage state is a read, so it never triggers a
barrier by itself; it only makes a prior compute write to the buffer
visible to the barrier emission at pass start.

The other backends already handle this hazard: Vulkan tracks
BufIndirectDraw explicitly, D3D12 transitions the buffer to the
INDIRECT_ARGUMENT state, and Metal and D3D11 track hazards
automatically.

Pick-to: 6.12
Change-Id: I9ecbfc0157a8a65c8813bed855e9840e05db94b2
Reviewed-by: Laszlo Agocs <[email protected]>
https://invent.kde.org/qt/qt/qtbase/-/commit/bafd17e0c8dbb51a970a22f79c9e1fa3e1aa9e78

Git commit 780ef39c3bcdfc7a80093a558b6008e2010829af by Christian Ehrlicher on 31/07/2026 at 17:16..
SQLite: Update SQLite to v3.53.4

[ChangeLog][Third-Party Code] Updated SQLite to v3.53.4

Pick-to: 6.12 6.11 6.8
Change-Id: Ie3137ce8442ad0824a46e043a1b5f14a0a154175
Reviewed-by: Axel Spoerl <[email protected]>
Reviewed-by: Kai Köhne <[email protected]>
https://invent.kde.org/qt/qt/qtbase/-/commit/780ef39c3bcdfc7a80093a558b6008e2010829af