They are faster, and using them makes it paint commands be the most
CPU intensive part of lancelot instead of regular-expression matching.
Change-Id: Ifabf1081c48a83ce089660049051428fd3a43042
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
When the thread that got woken up by release() is supposed to release()
to wake up another thread, we were deadlocking. This happened because we
cleared the bit indicating that there was contention when the first
release(). Instead of storing a single bit, we now store the number of
threads waiting.
Task-number: QTBUG-66875
Change-Id: I72f5230ad59948f784eafffd15193873502ecba4
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Resizing is broken on Wayland EGL on Intel Mesa. Move resizing into a separate
test and skip it on Wayland it until it's fixed in Mesa.
Task-number: QTBUG-66848
Change-Id: I9450a5a588b0f5d8f0bd0210aae2dc72aa48d622
Reviewed-by: Pier Luigi Fiorini <pierluigi.fiorini@liri.io>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
QRasterPlatformPixmap::systemOpaqueFormat returned QPlatformScreen::format
without checking that the format was actually opaque.
This caused several QPixmap tests to fail on Wayland because Wayland
compositors don't communicate the native format of the screen, just a list of
supported pixel formats, so we just return ARGB32_premultiplied in
QWaylandScreen::format().
Rename the method systemOpaqueFormat to systemNativeFormat since that's how
it's used most of the time. And do a conversion when we actually care whether
the format is opaque or not.
Task-number: QTBUG-51748
Change-Id: I47dc1c3f185fb802016ca361206d47d02e8d3cf1
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
This fixes the following:
- tst_QSqlDatabase::recordMySQL() to account for performance
improvements done for small integral types
- tst_QSqlQuery::nextResult() so that NUMERIC results are seen
as doubles
- tst_QSqlQuery::timeStampParsing() so that MySQL accepts the
CREATE TABLE statement
Change-Id: I68fb1d06dac12d500bb4596463f5bdd65cc9c226
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
When using a virtual table inside a SQLite database it is possible that
it does not report the right number of parameters. Therefore we need
to account for this case to prevent it from crashing when trying to
bind parameters it thinks does not exist.
Task-number: QTBUG-66816
Change-Id: I3ff70bb1fe73091f43c3df53616f75858e451cfd
Reviewed-by: Jarek Kobus <jaroslaw.kobus@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
When network access is disabled, every QNAM request returns a
QDisabledNetworkReply instance, which emits error and finished
immediately. However isFinished() was still false, which could confuse
application code.
Change-Id: Ifd43c86364b11a9583a38fde536e6c09c109b55f
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Matches glibc commit 9a0cc8c1bd7645bf3c988890ffb59639c07a5812.
[ChangeLog][QtCore][QUrl] Fixed a bug in parsing IPv6 addresses with
more than 4 hex digits in a component.
[ChangeLog][QtNetwork][QHostAddress] Fixed a bug in parsing IPv6
addresses with more than 4 hex digits in a component.
[1] https://sourceware.org/git/?p=glibc.git;a=commit;h=9a0cc8c1bd7645bf3c988890ffb59639c07a5812
Change-Id: I2701038131d91eb108aebb3bec16278e4efe3de2
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
This reverts commit fdddb3a481.
After b85b272c26810e54a4ceb0707cf4569f87517b67, Windows 10
(msvc2017-x86) is tested on Windows 10 (msvc2017-x86_64). The
failure is gone.
Task-number: QTBUG-66798
Task-number: QTBUG-66216
Change-Id: Iba353b18b0be1346007fde674a9f768c4b9bf384
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
This reverts commit a0db55040d.
After b85b272c26810e54a4ceb0707cf4569f87517b67, Windows 10
(msvc2017-x86) is tested on Windows 10 (msvc2017-x86_64). The
failure is gone.
Task-number: QTBUG-66756
Change-Id: I4a8e8c62b4a021d144d570c38c4f4c10e8b1317c
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
This reverts commit 7257862fb2.
It has been reported that this patch causes widgets with certain effects
become invisible.
Task-number: QTBUG-60231
Task-number: QTBUG-66803
Task-number: QTBUG-66387
Change-Id: I9c3c4cf2f17ac639d1aee5489b665aa1e165af16
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Rename getter QHeaderView::firstSectionMovable()
to QHeaderView::isFirstSectionMovable(), turn into a property
and fix \since.
Change-Id: Ica2ca43d22f3fd78ff63178ba75014807fc9b823
Reviewed-by: David Faure <david.faure@kdab.com>
Commit f55c73ede2 added various backendConfig methods;
API review for 5.11 pointed out that Config should not be abbreviated.
Change-Id: I3b294b44a030b2a6e4cdd034fa27583c228dfe42
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
They seem to be running just fine
Task-number: QTBUG-66849
Change-Id: Ia48d47c0fff173eb1d6eba38bdba752a91bd2b56
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
QDir::isAbsolutePath(name) thinks any path starting with a slash is
absolute; however, to return a valid absolute path, we need to put a
drive prefix onto such a name. So use QFileSystemEntry::isAbsolute()
for that check (it believes in the need for a drive, or UNC prefix)
and handle the absolute-but-for-drive case when it arises.
Add a regression test and make related changes to existing tests.
Task-number: QTBUG-50839
Change-Id: Id5d2b2586bb1423fa2d9375a298a4bb5241cffe0
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Otherwise random debug messages from Qt might mess up the expected vs
actual results.
The setting of QT_LOGGING_RULES in initTestcase has been removed, as
the selftest overrides that for each invocation of a subtests, via
the processEnvironment() function.
Task-number: QTQAINFRA-1631
Change-Id: I855d31274f8261f8b125df23409353f7101be0e4
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
The Q_GADGET macro cannot be used in templates. It can however be useful
to derive a template class from a Q_GADGET enabled base class to benefit
from type safety features in C++ (e.g. the class could represent an id
or handle for some C++ type).
For proper wrapping of a QVariant with a gadget value in a QJSValue, the
QMetaType::IsGadget flag must be set for the registered template
instance type - which does not happen prior to the fix because
IsGadgetHelper requires qt_check_for_QGADGET_macro to be defined in the
registered class but not in an ancestor class - in other words: The
class must declare Q_GADGET.
To overcome this, IsGadgetHelper/IsPointerToGadgetHelper can now
differentiate between a Q_GADGET flagged class (allowing
automatic registration) and a derived class, e.g. a template class
(forcing Q_DECLARE_METATYPE to be used explicitly).
[ChangeLog][QtCore][QMetaObject] It is now possible to use template
class instances inheriting from a Q_GADGET in Qml
Task-number: QTBUG-66744
Change-Id: I7632ad45cff79fa422b3f852ca0b963f35fab155
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
That is correct C++. GCC (and apparently Clang) accept it after the name
for compatibility with the old __attribute__ syntax.
Change-Id: Ie7f3adaaed83198ca1c61bc0efdf51634e457b07
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If buddy is deleted then QLabel will still try to deliver the
shortcut event to it, in QLabel::event(), and cras.
Besides connecting to destroyed() which alone fixes the crash,
also hardened it and guarded against dereferencing null buddy,
in the unlikely event of someone adding features to QLabel.
The second part is suitable for backporting to the LTS branches.
Task-Id: QTBUG-66841
Change-Id: Ib6a36a9f99cee31afe1f88eb77f5b6cafb8b1086
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
The test has been flaky on top of QEMU. The test is clearly a sort of manually
rolled benchmark, not a regular autotest. Remove the test and replace it with a
benchmark in QObjectBenchmark.
Task-number: QTBUG-66823
Task-number: QTBUG-66216
Change-Id: I7a48293023f32141eed6fea50fbb63af18933a8f
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
When QStandardItem::insertRows/Columns is called with an empty list
or an insert count of 0, do not assert but return false.
Task-number: QTBUG-66491
Change-Id: I328598e08584fb9b0a35075458bfeb31c57ebfee
Reviewed-by: Samuel Gaist <samuel.gaist@edeltech.ch>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Since Qt 5.4 the QTabBar can be automatically hidden when it has less
then 2 tabs. Therefore the sizeHint should not consider the tabbars
size when the tabbar is hidden.
Task-number: QTBUG-64715
Change-Id: I2f248f88d9070de5354f7344c7628a78442ab499
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Postgresql needs a special value for nan and +/- inf. This was
considered during insert but not during select.
Also remove some pre-c++11 inf/nan - handling and replace it with
Qt equivalents.
Change-Id: I044ca58e9cf673f4b100b05a0d8e25c8a9c29ec5
Reviewed-by: Robert Szefner <robertsz27@interia.pl>
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
The code that updates a section size must also update length,
otherwise saveState can end up saving inconsistent state, and
restoreState() goes to an early-return, not doing anything.
The actual bug was fixed meanwhile because _q_sectionsChanged is called
again, which recalculates length. I still see this only as a safety
measure, every other code path that changes section sizes updates length
right away.
Change-Id: I6cc16261692d93b3640afafef600a5bdff8dca0c
Reviewed-by: Thorbjørn Lund Martsum <tmartsum@gmail.com>
Before running tests that depend on QWindow::requestActivate
Gets rid of several Wayland platform checks in tst_QWindow.
Change-Id: I7a5e029044a968dfcf87ecbb5105c01d52852d35
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Pier Luigi Fiorini <pierluigi.fiorini@liri.io>
And make it easier to fix if platformName == wayland-egl etc.
Change-Id: Ia2d62ba003796e08f3e8a5bbfd0c3fd9d185e4e0
Reviewed-by: Pier Luigi Fiorini <pierluigi.fiorini@liri.io>
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
By using qWaitForWindowExposed instead of qWaitForWindowActivated.
Task-number: QTBUG-66824
Change-Id: Idf604157070731d9c92ccf64d8349c8571960b7c
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
The test actually passes, so there's no need to skip it.
Task-number: QTBUG-66824
Change-Id: Id091776ff7ca7637fdcf0e0ced833982b5788d92
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Has been flaky in CI.
Task-number: QTBUG-66815
Task-number: QTBUG-66216
Change-Id: I2c3eb42507eae618486aa402474b4b3f85ff310e
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@qt.io>
This follows on from a232251992 which
covered a similar instance of this. As with that change, we should not
abort the compilation, just ignore it.
Task-number: QTBUG-63772
Change-Id: Ide958080a90f43ed19edd8a320e7d45de1c96821
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
When first starting an Android app we have invocation order issue, to
load the platform plugin we create the default QLocale (needed by the
resource locator code to see if :/qt/etc/qt.conf exists) so when the
android platform plugin loads and creates its own QSystemLocale, the
QLocale defaultLocalePrivate is already created and pointing to
globalLocaleData which means that systemData won't be called and thus
the code that triggers the call to QLocalePrivate::updateSystemPrivate
won't be called when calling QLocale().
I thought of two ways of fixing this, one was calling
QLocalePrivate::updateSystemPrivatea() from the QAndroidSystemLocale
constructor, but giving the responsibility to not break things to the
plugin seems a little fragile, so making the check on QLocale()
seems better.
Without this patch an Android app doing
QApplication app(argc, argv);
qDebug() << QLocale().name();
qDebug() << QLocale().name();
qDebug() << QLocale::system().name();
qDebug() << QLocale().name();
would print
""
""
"ca_ES"
"ca_ES"
now it correctly prints "ca_ES" the four times.
Task-number: QTBUG-41385
Change-Id: I2cf419f59aa008fa3aca11295fe7d42c40bcc32e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Silence debug output by setting QT_LOGGING_RULES
to turn off all debug output.
Task-number: QTQAINFRA-1631
Change-Id: I5c2366b4fe4bac341dcfd92f68b6da8071c5b089
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Evaluating testlib's XML output in COIN would fail with:
"XML syntax error on line 7520: invalid UTF-8"
for the toLatin1() tests due to some Latin1/UTF8 mixup.
Add a helper function to convert the data to plain ASCII.
Task-number: QTQAINFRA-1797
Change-Id: I1c64878d4c2a67b8c2689905b5ffe6707b5963c1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The test has been crashing flakily recently.
1) It is creating a QGLWidget
2) It is stealing the QGLContext of that widget and moves it into a separate
thread.
3) In that secondary thread it makes the context current.
4) Meanwhile the QGLWidget itself may receive for example a resizeEvent or
other events and - since it assumes that it owns the context - attempts to
make it current.
5) Attempting to call makeCurrent() on a QGLContext that is in a different
thread than the current thread (via QObject thread affinity) will result
in a call to qFatal() and consequently the test aborts.
The conclusion from Simon Hausmann is that this test is testing a pattern from
Qt4 times that may or may not have worked back then. Nowadays with the Qt5
QOpenGL* API we do support this properly and there appears little sense testing
this.
Therefore remove the test altogether.
Task-number: QTBUG-66411
Task-number: QTBUG-66216
Change-Id: Ie2d66705bc7c3914ace6abcba9557c7c67ad4db3
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
This test executable was not flaky in the normal sense that when run, it
sometimes passes and sometimes fails. Instead, in some builds it would fail
consistently and in some builds it would pass consistently.
The first test to fail was version(ok00, default to last version) which gives
"mylib" as the library name and -1 as the library version. The description
implies that QLibrary selects the biggest or last used version when given -1.
However, versions less than 0 are not used at all. Instead the loading uses only
the name to select the library. Change the description to match.
So why did the test sometimes pass, sometimes fail? The test uses two library
projects lib and lib2 which install two different major versions of libmylib.
That includes the symbolic links:
libmylib.so -> libmylib.so.1.0.0*
libmylib.so.1 -> libmylib.so.1.0.0*
libmylib.so.1.0 -> libmylib.so.1.0.0*
libmylib.so.1.0.0*
libmylib.so -> libmylib.so.2.0.0*
libmylib.so.2 -> libmylib.so.2.0.0*
libmylib.so.2.0 -> libmylib.so.2.0.0*
libmylib.so.2.0.0*
The key thing being that both set the libmylib.so symbolic link. In a
multithreaded installation it's undefined which happens to set the link last.
The test code expected libmylib.so to point to libmylib.so.2.0.0. Ensure that by
building and installing lib2 after lib.
Task-number: QTBUG-66722
Task-number: QTBUG-66216
Change-Id: Ic513c772902273049c28e43fc1d83d550aafcd23
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
[ChangeLog][QtCore][QJsonValue] fromVariant() conversion now converts
from QUrl and QUuid using special encoding forms to ensure best JSON
compatibility.
Change-Id: I56b444f9d6274221a3b7fffd150cdc5ca1f87ff1
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
For compatibility with other parsers that may expect it to be so.
Change-Id: I56b444f9d6274221a3b7fffd150cd66390f98fd5
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
After a37785ec76 went in, it become
apparent that multi mouse button state handling in qtestlib is
non-existent, for details see QTBUG-64030 and QTBUG-63786. What
happened behind the scenes often was not what one would expect based
on the provided QTest::mouse* input sequence - events went missing,
incorrect events were generated, each subsequent test function
started with a state set from the function that run earlier. It is
easy to see how a minor change in one test could easily affect outcome
of other tests.
With a37785ec76, Qt platform plugins
are now responsible for sending explicit mouse button type and state
information; qtestlib should take full responsibility now as well.
But using the new API from a37785ec7 alone in qtestlib is not sufficient.
We need to reset mouse state between each new test function run (we do
this at function scope as that fits with the current qtestlib API user
expectations). This patch implements the necessary reseting logic.
Updated tst_qwindow.cpp::generatedMouseMove() to use QTest::mouse* APIs.
That test requires pressing multiple buttons, it was not possible with
QTest::mouse* APIs before this patch.
Added an auto test for multiple mouse button pressing/release in
tests/auto/testlib/selftests/mouse/. And few other tests which are
currently QSKIP-ed, but should be considered when re-designing qtestlib
APIs.
Task-number: QTBUG-64030
Change-Id: I39fdcbc73a467a7463ce2aed622bf22484095635
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
