Some QSpinBox tests start failing after reverting to using legacy mouse
messages to handle mouse input in the Windows QPA. It seems to be caused
by a test that runs before it and moves the mouse cursor. Then when the
QSpinBox tests run, they create widgets that appear below the mouse
cursor, causing some mouse events to be generating and messing with the
events synthesized by the test itself. With the pointer messages being
used for mouse input, the legacy mouse messages that are generated under
this condition were being ignored. But by reverting to the old
implementation, the legacy messages are handled again, causing the test
to fail. This change moves the mouse pointer to a safe position during
the test initialization, so it does not depend on the state left by
previous tests. This change needs to be integrated together or before
the change in the windows QPA.
Change-Id: I91f7e9376dc495ee61250e0a7d908c1c2b685bc8
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
It was needed as a workaround for the window border and window content
becoming out of sync when we were drawing and flushing to the layer
from another thread using OpenGL. Since we've disabled ThreadedOpenGL
we can remove this workaround for now.
The workaround also had other problems, as windowWillResize: did not
turn out to be a reliable signal to know when a window was about to
resize. Since we override windowWillUseStandardFrame:, we would not
get the callback when zooming windows. Conversely, if we removed the
windowWillUseStandardFrame: override, we would get a single callback
for windowWillResize:, and then multiple callbacks that the window
had resized. In addition, windowWillResize: is not only used as a
callback, but also as a way to let the window delegate restrict the
size, so it's called when e.g. AppKit determines the standard frame
of the window -- an operation that doesn't involve an actual resize
operation.
If we re-introduce a workaround using screen-update disabling we need
a better hook than windowWillResize:, and we need to track when we
disable screen updates so that we can pair them up with corresponding
screen update enables, otherwise we'll get log messages in the console
about "SLSReenableUpdate: unbalanced enable/disable update".
Change-Id: Ifca8892083c8666976391a4ada8f8d1471493943
Fixes: QTBUG-73726
Task-number: QTBUG-69321
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
The display-link callback comes in on a secondary thread, so there's a possible
race in using QGuiApplication::instance() to check whether or not we're on the
main thread.
Change-Id: Ic26bca8f5f54847a1e1b11dc92e786693c86e9de
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
The API is often called from main before entering the event loop,
and the making the window first responder and key will autorelease
both the window and view.
Change-Id: Ie2a7dc14652015cbe802b57696e4a82d564e2dc0
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Since include paths are fully resolved, we must remove the code that prepends
the SDK path to the OpenGL include paths.
Change-Id: I80d74629c7fc989a89c3f1d95d6de43b4c1de17a
Fixes: QTBUG-73736
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Calling QWindow::setVisible doesn't have the same semantics, so we
split off QWidget::setVisible into QWidgetPrivate::setVisible and
call that instead from QWidgetWindow.
Task-number QTBUG-67504
Change-Id: Ie50938d4a1d33ad4b59c742e75e3ca30f1b19399
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
For some reason, QCalendarWidget gets filtered press events that were
intended for Qt Virtual Keyboard's input panel (QQuickView), so we have
to make sure that the window is indeed a QWidget - no static_cast.
Change-Id: Ibc9dce956918ac50d1fed8231a445b7338aef09c
Fixes: QTBUG-72925
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Use xilink for ICC and lld-link for Clang.
Change-Id: I13c74339ae9e3e5c97210afd20a53c7e474b873b
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Top level windows already get their geometry changes via windowDidMove
and windowDidResize.
Change-Id: Ie6370aa290ef48c8b3ac770e77adb57ce43cbb47
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
We don't need to react to updateTrackingAreas, as we only have a
single tracking area that we can add once and forget. By asking
AppKit to track all events in the visible rect, we can also pass
a zero-rect for the tracking area.
Change-Id: I2545712adc49b51904d5adc11f1faca36901b49d
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Core Text doesn't actually have a concept of DPI internally, as it
doesn't rasterize anything by itself, it just generates vector paths
that get passed along to Core Graphics.
In practice this means Core Text operates in the classical macOS
logical DPI of 72, with one typographic point corresponding to one
point in the Core Graphics coordinate system, which for a normal
bitmap context then corresponds to one pixel -- or two pixels for
a "retina" context with a 2x scale transform.
Scaling the font point sizes given to HarfBuzz to an assumed DPI
of 96 is problematic with this in mind, as fonts with optical
features such as 'trak' tables for tracking, or color glyphs,
will then base the metrics off of the wrong point size compared
to what the client asked for.
This in turn causes mismatches between the metrics of the shaped
text and the actual rasterization, which doesn't include the 72
to 96 DPI scaling.
If a 96 DPI is needed, such as on the Web, the scaling should be
done outside of HarfBuzz, allowing the client to keep the DPI of
the shaping in sync with the rasterization.
The recommended way to do that is by scaling the font point size,
not by applying a transform to the target Core Graphics context,
to let Core Text choose the right optical features of the target
point size, as described in WWDC 2015 session 804:
https://developer.apple.com/videos/play/wwdc2015/804/
GitHub-PR: https://github.com/harfbuzz/harfbuzz/pull/1484
Change-Id: I830f0cd7a82552422bbe09226e2d571e246fe3f4
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Remove wrong code changing the Bido level of line separators. This
lead to wrong ordering of the string in case the line separator was
meant to be ignored and the string should be rendered in one line. Line
breaks are anyways already reset to the paragraph level by the algorithm
and reordering is done on a line by line basis, so this will work
correctly when doing proper line breaking.
Secondly fix a small bug found while testing the above change, where
we wouldn't set the correct levels for boundary neutrals and explicit
embedding chars because we did that processing before we were fully
done with the BiDi algorithm.
Change-Id: Id88f91cd58d2ab29be864aef34ca1727c1586611
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
The algorithm has been treating DirB inconsistently so far.
initScriptAnalysisAndIsolatePairs was treating it differently
than generateDireationalRuns leading to assertions.
It wasn't visible in our test data, as DirB is in almost all cases the
paragraph separator, where we split strings anyway.
Change-Id: I7dc0e7bbcf30ee84d8781ea06097da023e371f05
Fixes: QTBUG-73238
Reviewed-by: Robert Loehning <robert.loehning@qt.io>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Constructing a QStringRef directly from the string, offset and a
length is UB if the offset + length exceeds the string's length.
Thanks to Robert Loehning and libFuzzer for finding this.
QString::midRef (as correctly used in both changed uses of QStringRef,
since 432d3b6962) takes care of that for us. Changed one UB case and
a matching but correct case, for consistency.
In the process, deduplicate a QStringList look-up.
Added tests to exercise the code (but the one that exercises the
formerly UB case doesn't crash before the fix, so isn't very useful;
the invalid read is only outside the array it's scanning, not outside
allocated memory).
Change-Id: I7051bbbc0267dd7ec0a8f75eee2034d0b7eb75a2
Reviewed-by: Anton Kudryavtsev <antkudr@mail.ru>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The timeout will never be larger than numeric_limits<quint64>::max(),
especially on platforms with 32-bit longs.
Instead, test if the timeout is exactly numeric_limits<unsigned long>::max(),
which matches the ULONG_MAX value which is documented
to indicate no timeout.
Change-Id: Ib663eddb5703797c50c04fd4eae60bd64f379d1c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
WINAPI_PARTITION_PHONE_APP is defined for all our winrt mkspecs nowadays
so the code can be used unconditionally.
Change-Id: I4f2b60a0b9bba5b407ebbc213c44a0e5b4057855
Reviewed-by: Maurice Kalinowski <maurice.kalinowski@qt.io>
Our bundled ANGLE library only partially supports OpenGL ES > 3.0 so warn
users that there might be dragons.
Change-Id: I16711fe9f449e85dd8b2369e1fcec6c9f81d5ae0
Reviewed-by: Andre de la Rocha <andre.rocha@qt.io>
Reviewed-by: Miguel Costa <miguel.costa@qt.io>
Even though our ANGLE versions only partially supports OpenGL ES > 3.0,
there are users who want to use functionality that is available. By
setting major and minor version we can support this use case.
Fixes: QTBUG-72762
Change-Id: I9a1d3009355693baa971deb3c4bbf14c595edf0b
Reviewed-by: Andre de la Rocha <andre.rocha@qt.io>
Reviewed-by: Miguel Costa <miguel.costa@qt.io>
Our current ANGLE version (chromium/3280) relies on validation to be done
when doing the rendering, as the validation at the same time completes the
caching. Skipping the validation caused asserts and rendering issues.
The part of the validation that failed before is now deactivated in Qt's
copy of ANGLE as it is not relevant for our use case, so that validation
can be re-enabled now.
This reverts commit a1dec825f9.
Fixes: QTBUG-73317
Change-Id: I5fd176eaa0bc28d93ca93019b7092211fe5bcce5
Reviewed-by: Miguel Costa <miguel.costa@qt.io>
Reviewed-by: Andre de la Rocha <andre.rocha@qt.io>
We run into validation issues when using the ES2 code path when blitting
widgets on winrt. By using ES3 we not only avoid this issue, but there
might also be performance gains.
We now call window()->format() instead of window()->requestedFormat as the
latter will not respect the values that were set on initialization of the
native window (which is done in QWinRTWindow's constructor).
Change-Id: I5ed7a9326691375f9c9cb5d8d22ee8d1b643fbd0
Reviewed-by: Andre de la Rocha <andre.rocha@qt.io>
Reviewed-by: Miguel Costa <miguel.costa@qt.io>
Update our cflags and lflags with the ones found in android.toolchain.cmake
Fixes: QTBUG-73274
Change-Id: Id9fd9bf04df959239abd3100090a1485e872b2f0
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
The patch to elide the QToolButton text when there is not enough space
introduced a regression with multi-line text.
Fix it by using the newly introduced common function to elide multi-line
text.
Fixes: QTBUG-72226
Change-Id: I066ebbd2f360add93406cc29bb4bbbebf599ba42
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Factor out the calculation of the elided text from
QCommonStylePrivate::viewItemDrawText() so it can be used by other
painting functions.
Change-Id: I28e6bfd2fe4d7c552848446fa9913df78589d15b
Reviewed-by: Christian Andersen <csandersen3@gmail.com>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Try to better describe what it is and what it does. Also mention
its strongest use case.
Change-Id: Ib5c3e8a3c9b96169c139c5d7e8995a6a49d7d5e1
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
Tidied up the existing float tests in the process.
(In particular, s/SUCCESS/PASS/ since that matches real test output.)
These verify that QCOMPARE() handles floats and doubles as intended.
Extended the existing qFuzzyCompare tests to probe the boundaries of
the ranges of values of both types, in the process.
Revised the toString<double> that qCompare() uses to give enough
precision to actually show some of the differences being tested there
(12 digits, to match what qFuzzyCompare tests, so as to show different
values rather than, e.g. 1e12 for both expected and actual) and to
give consistent results for infinities and NaN (MinGW had eccentric
versions for these, leading to different output from tests, which thus
failed); did the latter also for toString<float> and fixed stray zeros
in MinGW's exponents (which made a kludge in tst_selftest.cpp
redundant, so I removed that, too).
That's further complicated handling of floating-point types, so let's
just keep an eye on how expensive that's getting by adding a benchmark
test for QTest::toString(). Unfortunately, default settings only get
runs that take modest numbers of milliseconds (some as low as 40)
while increasing this with -minumumvalue 100 or more gets the process
killed - and I'm unable to find out who's doing the killing (it's not
QProcess::kill, ::kill or the QtTest WatchDog, as far as I can tell).
So results are rather noisy; the integral tests exhibit speed-ups by
factors up to 5, and slow-downs by factors up to 100, between runs
with and without this change, which does not affec the integral tests.
The relatively modest slow-downs and speed-ups in the floating point
tests thus seem likely to be happenstance rather than signal.
Change-Id: I4a6bbbab6a43bf14a4089e96238a7c8da2c3127e
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Search the previous item or the next item in a model instead
of searching them on visual layout. This way the cursor will
not stop at the beginning or at the end of a row or a column.
Fixes: QTBUG-14444
Change-Id: I0ef203a4dcd876e4c50559fb87e61585f07434d1
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
GCC for 64-bit Windows has a bug that it fails to properly re-align the
stack pointer for use with 256-bit memory addresses (AVX). Therefore,
there's about a 50/50 chance that any function using AVX will have an
improperly-aligned stack. In release mode, stack accesses should be
rare, but in debug mode they happen frequently. Either way, this is a
ticking time bomb, so we disable.
Clang is not affected.
32-bit MinGW is not affected.
64-bit in other OSes with GCC are not affected.
Fixes: QTBUG-73539
Change-Id: Id061f35c088044b69a15fffd1580967808f31671
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Since the official v4l-utils-1.12.0 release, media_get_entity_by_name() function expects
only two arguments instead of three as in older versions thus breaking build of eglfs_kms_vsp2
backend.
Cf. https://git.linuxtv.org/v4l-utils.git/tree/utils/media-ctl/mediactl.h#n253
Fixes it by creating an overloaded wrapper function that will choose the right
version based on the signature of the media_get_entity_by_name function pointer
argument.
Fixes: QTBUG-73427
Change-Id: Idab52558b6f2f23137456c21e33ece1ef0e9aa4e
Reviewed-by: Johan Helsing <johan.helsing@qt.io>
If we have have AA_EnableHighDpiScaling on and have loaded a @2x image, layout
calculations can't use the real pixels of the QPixmap directly.
Fixes: QTBUG-73401
Change-Id: I1891411a0e359e0148476f73b6cc3a128893a374
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
It is the flaky test causing most failures in qtbase at the moment.
Task-number: QTBUG-73545
Change-Id: Id9c5db27ebd08a4cf3c119d2fada12fdf1a5d2a0
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
Header files of modules that specify generated_privates are usually
not yet available at qmake-time. Thus, the installation rule must not
check for the file's existence.
Change-Id: Ifc7ff95422912d255744c9006382ff181176ae77
Fixes: QTBUG-71340
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
By only looking at the function prototype, it's tempting to assume that
the handler needs to be registered with the usual SLOT macro. That is
not the case.
The code snippet is already included in the class description, but it
doesn't hurt to repeat it here.
Change-Id: If24fdca41a4bd976ebd1156c9e1106469388265c
Fixes: QTBUG-50484
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Reviewed-by: Topi Reiniö <topi.reinio@qt.io>
The non-qualified hue() and saturation() etc. refers to those values
in the HSV, not HSL, color model. Make this more explicit in the
documentation to avoid confusion.
Task-number: QTBUG-73129
Change-Id: Ief337672966ac72d0d0c3606d8d68acf01ffe7ee
Reviewed-by: Venugopal Shivashankar <Venugopal.Shivashankar@qt.io>
We translate all pure gray colors into cmyk having c,m,y=0 and only
the k value expressing the darkness. But a fix introduced to avoid
division by 0 caused rgb(0, 0, 0) to be an exception to this; it ended
up being translated as c,m,y,k=1 instead.
Fix by catching the potential div-by-0 situation earlier and directly
set the orthodox cmyk translation: c,m,y=0,k=1.
Fixes: QTBUG-73171
Change-Id: I3774eaf9d96e096ac5c47c55d28881bea2bd1309
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
For some overly tight beziers where the start or end point and the
next control point are closer than the pen width, the stroker's
shifting algorithm will produce a start/end tangent pointing in the
opposite direction from what is expected, for one of the sides. This
would break the square and round capping logic. Fix by detecting the
situation in the capping function and reversing the tangent when
necessary.
Change-Id: I48f4f017403d7b289b0483dd2b3a7ff1bbd0cf2a
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
The read_xbm_header() function is used to check for valid file header,
containing valid width and height values. But in case of an invalid file,
the check could depend on uninitialized variables.
Change-Id: I9f933ed6e38d86109e5b5a8d55fe763ab928d749
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The rendering of a css styled menu item with icons or checkmark was
partially fixed with aa1bc47942 but
introduced some other painting regressions, especially in RTL mode.
Fix it by syncing the code with fusion and windows style. With this
patch a menu text with a check mark but no icon is now aligned exactly
the same as a menu text with a icon.
Fixes: QTBUG-66380
Fixes: QTBUG-70491
Fixes: QTBUG-72817
Change-Id: I83a95d15eb130e7f72471820b53c3cd5554d9334
Reviewed-by: Nick D'Ademo <nickdademo@gmail.com>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
tst_QHeaderView::defaultSectionSizeTest() fails on High-DPI screens
because the default minimum section size is greater than the values used
for testing the header sizes. Therefore the test will fail.
Fix it by explicitly setting the minimum header size to something
smaller than the test values.
Also add a debug line to output the default minimum section sizes so
other failures due to this problem can be debugged better.
Fixes: QTBUG-73309
Change-Id: I257f341cef9381f140aa4d4f68376c5edadc39cc
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
If a window needs display due to e.g. being resized, we need to wait
until the corresponding expose event has been delivered before we
resume update requests, otherwise the update requests may result
in partial paints that do not fully cover the area needing display.
Change-Id: Ibfb54bfe3c2e85b606ef67d34a6a5fdb85456edd
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
