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

Git repository change summary for qt/qt/qtdeclarative
Pushed by mirror-service into branch 'dev'.
Changed from debbdc9a5d2e59ac52f70aaf11267a4d74612672 to 54e6b16ee7940f61cd3c9b46bf5e59062408ca68
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 d6d1fde95e91ddcc5a3df0c2b263a14d2f7f672b by Kwanghyo Park on 06/08/2026 at 07:27..
quickvectorimage: implement pattern fill support for ItemGenerator

Render pattern tiles into an off-screen container and apply them as
QQuickShapePath::fillItem via a repeating ShaderEffectSource. Both
userSpaceOnUse and objectBoundingBox patternUnits are supported.

Task-number: QTBUG-147844
Change-Id: Ic1d7201cd3015be5affd4a30cf339c2067e6f796
Reviewed-by: Eskil Abrahamsen Blomfeldt <[email protected]>
https://invent.kde.org/qt/qt/qtdeclarative/-/commit/d6d1fde95e91ddcc5a3df0c2b263a14d2f7f672b

Git commit 616a70496e624a87673b12739cbd9ae75935a31b by Olivier De Cannière on 06/08/2026 at 12:40..
Compiler: Un-quadratify populateReaderLocations

Despite recent performance improvements, populateReaderLocations still
has quadratic complexity in the size of the input function's instruction
count. This complexity can become unmanageable on rare but still
realistic inputs such as the array of 5000 strings of the linked issue.

This patch improves the algorithm to hopefully sub-quadratic complexity.

Instead of walking the reachable downstream instructions for every write
in search of readers and/or another write invalidating the first (the
source of the quadratic complexity), sweep the basic blocks once and
keep a state tracking what is reachable where. Every read then resolves
against that state with constant time lookups into that state.

Significant performance gains are observed in the targetted problematic
long functions. No meaningful performance impact is observed for other
functions. Performance comparison results for the HugeArrayCompilation
benchmark can be found on the linked issue.

As its name implies, the function populates m_readerLocations. In order
to verify the change is safe, the output was compared with the output of
the baseline (parent of 8e7626a8525719173b517133559f4ef2f11d213a)
function when building qtdeclarative/src and tst_qmlcppcodegen, and the
they were compatible. All members of the RegisterAccess values were
equal except for registerReadersAndConversions that now contains more
conversion locations. This is safe as it can only lead to types being
overgeneralized. This may in theory lead to missed dead-store
eliminatation opportunities but none were observed in pratice. Generated
code output was identical in both cases.

Also documented the members of RegisterAccess.

Fixes: QTBUG-148164
Pick-to: 6.12
Assisted-by: Claude Sonnet
Change-Id: I8a5b4a5547b37e823a9cd6a056d9fae72279b45a
Reviewed-by: Sami Shalayel <[email protected]>
https://invent.kde.org/qt/qt/qtdeclarative/-/commit/616a70496e624a87673b12739cbd9ae75935a31b

Git commit 559fc0e3acaf82ec9d5af601d14e146dfad3b8de by Tor Arne Vestbø on 06/08/2026 at 12:40..
CMake: Report MACOS_BUNDLE_POST_BUILD as a no-op instead of ignoring it

In 8cae7ea4ac7ecbc0063d37a713332e252739e745 we retired the build-tree
QML deployment step and kept MACOS_BUNDLE_POST_BUILD as a silently
accepted no-op, for source compatibility. A project that relied on
that step to populate its build-tree bundle therefore keeps
configuring and building cleanly while the bundle quietly stops being
populated, with nothing to point at the qt.conf that replaced it.

Say so instead. A notice, not a warning: passing the option is not a
mistake, it just no longer buys anything.

Also drop the three in-tree callers that
2967ad70d2601d6d8d5451f40897cdf37febc209 missed: it swept the examples
but left stylekitcustomdelegates, and it did not touch the two tests.

Task-number: QTBUG-101338
Pick-to: 6.12
Change-Id: Ie0955911811700f71641aa174630a9708db161ad
Reviewed-by: Orkun Tokdemir <[email protected]>
https://invent.kde.org/qt/qt/qtdeclarative/-/commit/559fc0e3acaf82ec9d5af601d14e146dfad3b8de

Git commit c207c83257b381b41163c2e2f46f146b042bc39b by Olivier De Cannière on 06/08/2026 at 12:40..
qmllint: Don't crash on FunctionSignatureBehavior: Ignored

When the pragma was set to Ignored, we exited early before setting the
code on the function object in populateSignature. This then led us to
crash in the first bytecodeHandler pass of the compiler, the basic
blocks, that assumes there is code to decode.

Don't exit early to make sure the code is set on the function. This is
fine because we already don't attempt to compile functions without type
annotations. But this now let's us lint them.

Adapt tst_qmllint to handle snippets starting with pragmas.

Fixes: QTBUG-148633
Pick-to: 6.12 6.11
Change-Id: I89edd827775f1d602225fca6defbcff5b9bcfd33
Reviewed-by: Sami Shalayel <[email protected]>
Reviewed-by: Semih Yavuz <[email protected]>
Reviewed-by: Fabian Kosmale <[email protected]>
https://invent.kde.org/qt/qt/qtdeclarative/-/commit/c207c83257b381b41163c2e2f46f146b042bc39b

Git commit 54e6b16ee7940f61cd3c9b46bf5e59062408ca68 by Olivier De Cannière on 06/08/2026 at 12:40..
qmllint: Don't depend on QmlDom

The dependency was added for eventual potential qmllint fixes using the
Dom. Those don't exist yet and qmllint doesn't need the Dom.

Remove the dependency.

Amends a6f8cf4274ddc7c1465410d6a15677040262f8a7

Change-Id: I98d5bf82c7c038ee66594f67b5418d153f926b3a
Reviewed-by: Semih Yavuz <[email protected]>
Reviewed-by: Sami Shalayel <[email protected]>
https://invent.kde.org/qt/qt/qtdeclarative/-/commit/54e6b16ee7940f61cd3c9b46bf5e59062408ca68