Make QFreetypeFace::getFace() not return the same font
face for all memory fonts.
QtFreetypeData has a cache which maps QFontEngine::FaceId
to QFreetypeFace instances. Each FaceId should then
have a unique hash value, which is accomplished by
by hashing the font file name and a stored uuid.
However, we were not setting the uuid in all cases,
which means that in-memory fonts (which do not have
a file name) would hash identically.
Fix this by setting the uuid to a value based on the
address of the FontFile, which should be unique.
Pick-to: 6.7
Change-Id: I45de37b512fb666c26490551c831c0db824b5f35
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Unloading plugins after each test case in QtDeclarative autotests leads to crashes (signal 11) during type registration in subsequent test cases.
The issue arises due to the unloading process, where unloading and then re-registering types in the next test case leads to sig11.
Disabling the unloading of libraries prevents the crashes.
Pick-to: 6.7
Task-number: QTBUG-115777
Change-Id: I6712f687134fd452fb0b962363a8f0fde85344ff
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If a commit adds new baseline test items, they would not be reported
to the server when doing a LanceBot test on that commit, and so not be
visible in the report either. This commit fixes the client side of
that issue; the server is already updated.
Change-Id: I60c2958def5e7b54ddc2789cf5727edcbb1a90e1
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Reviewed-by: Jonas Karlsson <jonas.karlsson@qt.io>
This fixes a memory leak due to a missing call to release
on the localized names in populateFamilyAliases(). The
DirectWriteScope makes this automatic when it goes out
of scope.
We did release in the other places, but for hardening
the code a bit, this also adjusts them to use the scope
guard.
Change-Id: I88402fad930e36cdd3a78244941fb53ca214520e
Reviewed-by: Lars Knoll <lars@knoll.priv.no>
To enable setting a default priority for all requests to a given site.
Fixes: QTBUG-121669
Change-Id: Icd8a9c59e4afb0432c65f84cc8115ae0c7f85506
Reviewed-by: Juha Vuolle <juha.vuolle@qt.io>
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Add the reference to unity builds in the documentation of
NO_PCH_SOURCES to fully describe the semantics.
Change-Id: Icfd3ee401efa154da4b8ed9a307d29f2605d948a
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
onLoaded and the initial expose/paint should be sequenced
such that onLoaded is fired first, followed by the expose.
This makes sure that we don't spend any time on painting
frames before Qt is completely initialized.
Add a "requestUpdateHold" mode to QWasmCompositor (initially
on) which disables requestUpdate calls, as well
as releaseRequestUpdateHold() which enables requestUpdate
calls again. This is a one-way transition; the mode
can't be enabled again.
This amends commit f2e22774 which implemented the concept
of startup tasks, where onLoaded can be delayed until
for instance font loading has been completed. After
this commit the expose event and initial commit will
be delayed as well.
Change-Id: Icc784306726174fbabe8785d54485860e968745a
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
Reviewed-by: Piotr Wierciński <piotr.wiercinski@qt.io>
When cross-compiling, host rcc might generate zstd compressed
resources, even though the target might not support zstd
decompression.
To avoid that, we made sure to disable zstd compression when using
cmake api like qt_add_resources if the target platform does not
support it.
We did not do it for CMAKE_AUTORCC though.
In such a situation, the linker would fail with:
error: undefined symbol: qt_resourceFeatureZstd
Add the --no-zstd option to AUTORCC_OPTIONS for targets that are
created by Qt CMake public API like qt_add_executable and
qt_add_library if the target platform does not support zstd
decompression (check via the QT_FEATURE_zstd variable).
This in turn applies to our own qt_internal_add_ API as well.
Allow opting out via the QT_NO_AUTORCC_ZSTD CMake variable.
[ChangeLog][Build System] Targets created with qt_add_executable
and qt_add_library will now add the --no-zstd option to AUTORCC_OPTIONS
when the target platform does not support zstd decompression. You can
opt out via the QT_NO_AUTORCC_ZSTD cmake variable.
Pick-to: 6.5 6.6 6.7
Fixes: QTBUG-121948
Task-number: QTBUG-106466
Task-number: QTBUG-101353
Change-Id: Ibdcfecd9a4b1e206479a3f4588b1b624dd91e122
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
Change-Id: I508edd50f4fa8a799dcfd746a0a916e3e39be84d
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
We cannot remove the overload using QT_REMOVED_SINCE, because a qint64
lvalue in the new overload will not bind to an uint& parameter, so the
old code would not compile.
Deprecate the old overload, and add a unit-test that makes sure that
it still behaves correctly.
This commit also introduces the new deprecation macros that are
required to do the deprecation in Qt 6.11.
Amends fd48ce0b73
Found in 6.7 API review
Pick-to: 6.7
Change-Id: I02893bfbe040df736f8e746384e0261a0f0041d3
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The operator was not converted to using a new streaming version, so it
supported only 32-bit length.
Fix it and add a manual test for serialization/deserialization of
large c-style strings.
Amends fd48ce0b73
Pick-to: 6.7
Change-Id: I83704edec021d400b992de810feba5da52d5ffe1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
If the underlying QDataStream's device already contains all data that
we want to read, we can optimize the allocation algorithm, and try to
allocate all needed memory at once.
Use bytesAvailable() to determine if the underlying device can provide
all requested data, and adjust the initial block size based on the
result. If not all data is available, fall back to the geometric
growth algorithm.
Change-Id: I6384d2caa16c238c2dbb77b2ad761cbd8a44df6c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The algorithm tries to allocate the memory using 1 Mb blocks.
When the input data is large, this results in a lot of reallocations,
which is slow and inefficient.
This patch modifies the algorithm in such way that the allocation size
increases at each step, resulting in geometric growth of the buffer.
Pick-to: 6.7 6.6 6.5
Change-Id: I071df68d51ba1dbd8b2eb5f94eb078a33223505f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Set this status when the stream tries to write more data than the
current serialization format supports.
Update the methods that write containers to return early if they fail
to write the container size, and do not try to serialize the elements.
Convert the manual tst_manualqdatastream test into a data-driven
test, allowing us to specify various stream versions. Adjust the test
code to check that the SizeLimitExceeded status is set when the
stream version is <= Qt_6_6.
Amends fd48ce0b73
Found in 6.7 API review
Pick-to: 6.7
Change-Id: If4c62ea53ac9bccd423f00f0f03afd6ba6bdc4f5
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Do that to avoid narrowing at the call site on 32-bit platforms.
Amends fd48ce0b73
Found in 6.7 API review
Pick-to: 6.7
Change-Id: I31142399385521d973b2ed3789745569e44d5d63
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
By wrapping the use of the timeout value in QTRY_IMPL in a lambda that
feeds the user input through the std::chrono::milliseconds constructor
with std::chrono_literals in-scope, the macros continue to work with
raw integral values as well as chrono literals not finer than
millisecond granularity.
Port all higher-level macros to pass a chrono literal and port some
uses in tst_selftests.
[ChangeLog][QtTest] The QTRY_*_WITH_TIMEOUT macros now also accept
chrono literals (was: int milliseconds).
Fixes: QTBUG-121746
Change-Id: Ib38406fc005a0a2c4ae3fd009760f73ad8bed355
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
According to QUIP-18 [1], all tools file should be
LicenseRef-Qt-Commercial OR GPL-3.0-only WITH Qt-GPL-exception-1.0
[1]: https://contribute.qt-project.org/quips/18
Pick-to: 6.7
Task-number: QTBUG-121787
Change-Id: I89e08859871d29a9cf8c7a56b30b07b2c2f34b42
Reviewed-by: Kai Köhne <kai.koehne@qt.io>
According to QUIP-18 [1], all examples file should be
LicenseRef-Qt-Commercial OR BSD-3-Clause
[1]: https://contribute.qt-project.org/quips/18
Pick-to: 6.7
Task-number: QTBUG-121787
Change-Id: I185ebdff2faa5e4af6ac0ee77c3ae33faae06e7d
Reviewed-by: Kai Köhne <kai.koehne@qt.io>
According to QUIP-18 [1], all build system files should be
BSD-3-Clause
[1]: https://contribute.qt-project.org/quips/18
Pick-to: 6.7
Task-number: QTBUG-121787
Change-Id: I58689f832ed3a5c8475c9bd7e0205e2128fb7cc2
Reviewed-by: Kai Köhne <kai.koehne@qt.io>
Using Selenium for WebAssembly testing enables us
to test user interactions, which is very valuable.
Turning this test into automated allows us to run
it in CI pipeline. This will help with detecting
regressions.
Two of these tests are currently failing on CI
machine and they have been temporarily disabled.
Change-Id: I754dd05955e55eb031070f5328ef715b7826c2b5
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
When using a style sheet, Q(Plain)TextEdit::isReadOnly() might get
called during the initialization before d->control is properly
initialized which lead to a crash.
This amends 43ce457cbd.
Fixes: QTBUG-121697
Fixes: QTBUG-121790
Task-number: QTBUG-1857
Pick-to: 6.7 6.6.2 6.6 6.5 6.2
Change-Id: I15c357c9eef7f6559bcc2ad89033a3d8e7fcbfef
Reviewed-by: David Faure <david.faure@kdab.com>
It looks like we don't even support nested namespace(RCC namespace
mangling fails). Assuming that QT_NAMESPACE should be
the valid C/C++ indentifier.
Change-Id: I5c2073c21964eb96abb3e894d83e1df65f7730b4
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The perivous version generated weird condition, and seems changing
the QT_NAMESPACE after qtbase configuration is noop, we may replace
the generated condition with the conditional generation.
Change-Id: Ifa09dba4db00099a07da2cff5505e6fd0b008289
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
This covers the following use case:
QtModuleX is pre-built and installed, it's imported. The plugin has
a PLUGIN_TYPE that is associated with QtModuleX and is built with
application that links QtModuleX. When deploying the application
it's expected that the plugin is deployed, as the one that belongs
to the linked QtModuleX.
This ensures that we udpate the internal _qt_plugins property that
is used in the __qt_internal_collect_plugin_targets_from_dependencies
function.
Pick-to: 6.5 6.6 6.7
Change-Id: I9824351ebab5a24509800da4db69f1e282a35884
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
1XX statuses are not final statuses, and testing 'error' on it is not
necessarily very meaningful. Furthermore the internal handling of
these statuses changes in Qt 6.8.
Fixes: QTBUG-108068
Pick-to: 6.7
Change-Id: I9c18c579cd310266273eccd2aee2e032f60538e0
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Make QNetworkAccessManager to discard all informational (1xx) replies
from HTTP servers with the exception of 101.
According to RFC 9110:
"A client MUST be able to parse one or more 1xx responses received
prior to a final response, even if the client does not expect one.
A user agent MAY ignore unexpected 1xx responses."
Status code 101 is an exception because we use it to upgrade to http2
cleartext when it is enabled.
Fixes: QTBUG-108068
Change-Id: I415ff053180a43f0c7ff1b4b2a60fd114ce08ac9
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Juha Vuolle <juha.vuolle@qt.io>
Android 13 and above requires the notification permission
"android.permission.POST_NOTIFICATIONS" to be requested at runtime.
Pick-to: 6.7 6.6 6.5
Fixes: QTBUG-121668
Change-Id: I38512ffc07c1a3ffe7cddffa937b98332f172d15
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
If extract_android_style meta data is set to none currently the app
still tries to extract full stlye and crashes. If this option is set
no style data should be extracted after this patch.
Pick-to: 6.7
Fixes: QTBUG-121667
Change-Id: I764e1eb6a582073196f991ca930d245d17a1f7e5
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
It is possible for a QRhiD3D12 instance to be created and destroyed
before the cmdAllocators list is initialized. This change simply
guards the cmdAllocators so that Release is only called if the
element is not nullptr.
For an example of how this can happen see QRhi::create. The
QRhiD3D12 is created but may be released immediately if
QRhiD3D12::create fails. One way this may happen is if the
ID3D12Device is removed but in practice many different errors may
cause create to fail.
Pick-to: 6.7 6.6
Change-Id: I395d247a952f9584122be083ac5ca6a3caddf300
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
isTimeZoneIdAvailable() is significantly slower than just trying to
initialize the timezone and see if that worked.
Even in the x86 emulator the difference for this is from 2+ms to no
longer measurable here, on less powerful ARM devices it's even more
extreme. This matters in particular for code creating many QTimeZone
instances, e.g. for calendaring.
Pick-to: 6.7 6.6 6.5
Change-Id: I5f175137b8b71816347a8debb492214427a51104
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Some unix-like concepts are supported by VxWorks VSB layer - UTILS_UNIX.
One of such methods is getgrgid(). Include it in
`qfilesystemengine_unix.cpp`, so that we don't need to exclude VxWorks from code that uses it anymore.
Task-number: QTBUG-115777
Change-Id: I72b301647bfdb208cb6859bb0f9994e3537fc345
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
[ChangeLog][Third-Party Code] Updated Valgrind header used by QtTest.
The change only affects portability of s390 inline assembler.
Task-number: QTBUG-121346
Pick-to: 6.7 6.6 6.5
Change-Id: I7eca3b13d4f5475693c548b85b837d5d4b4b090f
Reviewed-by: Kai Köhne <kai.koehne@qt.io>
Neither win_outputDebugString_helper() nor OutputDebugString() add a
newline at the end of the message, so one needs to be added before being
passed to the output handler.
A newline was previously present but removed when deduplicating calls to
qFormatLogMessage().
Fixes: QTBUG-121947
Pick-to: 6.7
Change-Id: I9f1c1d8726e4234f24999f47c52340140d5a4614
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The GDI font backend is missing support for certain modern features,
and has a lot of work arounds for missing APIs. DirectWrite is the
modern way to handle fonts on Windows, so we make this the default
now, but keep the old backend as a fail safe.
Fixes: QTBUG-119420
Change-Id: I0ea5cdfdcd759ccc894efb01b2410826c44aa1ea
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Lars Knoll <lars@knoll.priv.no>
The Android TextureView class wants to know whether the view is
opaque, while the condition has been evaluating whether it should be
transparent. Invert the condition to let Android know correctly whether
the TextureView should have transparency.
Pick-to: 6.7
Change-Id: Ic636f78dd3c691e85456c579e4559b8bc7a077a2
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
CarbonCore defines `DEBUG`, which can nameclash with the `DEBUG` define
in the json parser when using unity builds
Change-Id: Ic9f666a1da98aaebe30836abf877228f2f83004c
Pick-to: 6.7
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When linking Qt statically, QtCore and QtGui are not separated into
different DSOs.
when both statically linked versions of QtCore and QtGui are linked into
the same binary with LTO, gcc emits:
```
/usr/src/debug/qtbase/6.3.0-r0/git/src/gui/painting/qicc.cpp:105:
warning: type ‘Tag’ violates the C++ One Definition Rule [-Wodr]
/usr/src/debug/qtbase/6.3.0-r0/git/src/corelib/kernel/qtranslator.cpp:78:
note: an enum with different value name is defined in another
translation unit
/usr/src/debug/qtbase/6.3.0-r0/git/src/gui/painting/qicc.cpp:106: note:
name ‘acsp’ differs from name ‘Tag_End’ defined in another translation
unit
/usr/src/debug/qtbase/6.3.0-r0/git/src/corelib/kernel/qtranslator.cpp:78:
note: mismatching definition
```
we therefore define the `struct Tag` in an anonymous namespace or the
QIcc namespace
Change-Id: Ib4edeede35c51322ab1959d70fb87359b196e59b
Pick-to: 6.5 6.6 6.7
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The docs say "This example illustrates function calling with 0, 1 and 2
parameters" because it was copied from the synchronous call()
documentation. But the snippet didn't do that, because back in Qt 4 when
this was created, it was too difficult to exemplify that for
asynchronous use. It now no longer is, with lambdas.
Drive-by update the docs for both functions to refer to each other.
Fixes: QTBUG-121873
Pick-to: 6.6 6.7
Change-Id: I664b9f014ffc48cbb49bfffd17b089b7f77c1cde
Reviewed-by: Ievgenii Meshcheriakov <ievgenii.meshcheriakov@qt.io>
readdir_r() has been deprecated since glibc-2.24; all usage of
QT_READDIR_R in qtbase has been ported to readdir() since 2016
4b6784b49c (which explains in details the
reasons behind the deprecation).
What's left is the QT_READDIR_R, user code that still uses it, can
switch to readdir_r() (which is not advisable).
[ChangeLog][QtCore] Removed QT_READDIR_R macro; readdir_r() has been
deprecated since glibc-2.24 and it's recommended to use readdir()
instead. For more details see:
https://man7.org/linux/man-pages/man3/readdir_r.3.html
Change-Id: Icca2dca7e696533dcb983a82ba97a13baadcf015
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Also use C locale for this, rather than QLocale().
This makes the ICU backend's override redundant.
Change-Id: I3d668dd3a784b48d0a5fff7d11cc25a6e1423c84
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The two differed only in one selecting UNIX, the other WIN32; not sure
we have any platforms for which neither of those conditions is true
but, in any case, if such a platform has ICU and enables feature
timezone, I guess we should be using ICU for its timezone backend.
Change-Id: I3a7812c143f18732bf4dae7e8de8e42a662f0f39
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Morten Sørvig
Jacek Poplawski
Michał Łoś
Eirik Aavitsland
Eskil Abrahamsen Blomfeldt
Ahmad Samir
Mårten Nordheim
Tim Blechmann
Alexey Edelev
Alexandru Croitor
Jari Helaakoski
Ivan Solovev
Marc Mutz
Lucie Gérard
Piotr Wierciński
Christian Ehrlicher
Juha Vuolle
Mate Barany
Assam Boudjelthia
Oliver Dawes
Volker Krause
Krzysztof Sommerfeld
Edward Welbourne
Johannes Grunenberg
Tasuku Suzuki
Tinja Paavoseppä
Thiago Macieira