This will make two floating points containing NaN compare as equal,
instead of the regular nan != nan IEEE behavior (which isn't very useful
in a unit-test framework).
Note that this does not apply to indirect comparisons, for example via
QVariant.
Change-Id: I39332e0a867442d58082fffd150851acfdd18c23
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Various pieces of code have to be disabled in this case.
Change-Id: I83b133f17e9f024016a79c9103293627185449d2
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Fixes the default C version used with gcc < 5
Change-Id: I948dece961caed8e6b181e1c6e6b9dc43c46583f
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Implement QCocoaWindow::isEmbedded() which detects this
property based on parent view and window type. This avoids
having to use a setter function to set the state.
The detection can’t handle all cases, but should be
sufficient for our use case.
Change-Id: I12a5b90b4e4a7e10714f7275ae001e99c9361e2c
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
When performing gestures such as Exposé or Mission Control with the
mouse button down, the position of the mouse release event is
returned with NaN values. This causes
QGuiApplicationPrivate::processMouseEvent to be called recursively
and ultimately crash.
Task-number: QTBUG-67194
Change-Id: If1536bc4dc2075c498cdd6c5afe57c86bdaac13b
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
We can use the QT_LIB_FOO #defines instead of our own WITH_FOO ones to
determine if a library is available. Also, it doesn't currently make
sense to refer to libraries which are not part of qtbase here. We might
add that in the future, but QtScript (being deprecated) is probably not
the first one we should add.
Change-Id: I7f2397ca5499ba6003088478161182e960e815fb
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
It depends on QAbstractItemModel, so we need to switch it off if
itemmodel is unavailable.
Change-Id: I97246767a5e387b7a2cee90c34125a8411ef1c4e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
This follows the naming convention for global headers and causes
qttestglobal.h to be added to the top of the generated QtTest header. It
is necessary to have it there when other headers depend on features
defined in configure.json.
[ChangeLog][QtTest] The qtest_global.h header is now deprecated. Include
qttestglobal.h instead.
Change-Id: Iaed639d4e13dd99cee6387fce9d15d6e55b0b1e8
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Previously it was only enabling use of TLS1.0, unlike our openssl
backend, which understandably caused some confusion among some of our
users.
Seeing as this is also the default value in QSslConfiguration it is nice
to have it negotatiate more secure ciphers.
Task-number: QTBUG-67112
Change-Id: Ie216703da1ec4e6b973a881040e14816ad4c0a32
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
This should be tied to the CustomizeWindowHint flag:
if that flag is set we start out without NSResizableWindowMask
and possibly add it later on if WindowMaximizeButtonHint
is set.
Change-Id: I7e826d4bd357a8a17c60cfef948af25d61b66ebf
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
This helps to distinguish timeouts from real crashes in COIN.
This is already done for Windows.
Change-Id: I4daeafa36f50482d20cea4bd1106647081ff7abe
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Qt::SvgMiterJoin is more important than the name suggests: It is used
not only by SVG but also by PDF. It differs from Qt::MiterJoin in the
miter limit behavior.
Task-number: QTBUG-52640
Change-Id: I8ad04b1231958628caab18a233d54d42ea6449e7
Reviewed-by: Sze Howe Koh <szehowe.koh@gmail.com>
FT_LCD_FILTER_H wasn't defined because we weren't including the header.
To fix it just remove the checks, as was done for Linux and assume
sub-pixel is there. If it's not then no harm done, it won't use any.
Change-Id: I76f50cb17e41621c45c03cb7d5c75c110557ea68
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
The added xcb-xinput code was produced by build of libxcb 1.13
with xcb-proto 1.13.
The following parts were removed from it:
- Pointer Barriers API (requires xcb-xfixes 1.9 with xcb-proto 1.9)
- SendExtensionEvent API (requires definition of xcb_raw_generic_event_t
from libxcb 1.13)
[ChangeLog][Third-Party Code] Sources of xcb-xinput 1.13 were bundled and
are available via -qt-xcb.
Change-Id: I43d2f43bee0ba874d099c9fb858e74b0e3edc970
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
...and update bundled xcb sources to libxcb 1.9.1 with xcb-proto 1.8.
These are the minimal versions of libxcb and xcb-proto available
on officially supported platforms (they are present on RHEL 6.6).
Remove support_libxcb_versions_where_xcb_sumof_not_available.patch
and revert it for xkb.c, because libxcb 1.9 implements xcb_sumof().
This change makes it easier to bundle xcb libs from newer versions
of libxcb (e.g. xcb-xinput).
[ChangeLog][Third-Party Code][X11] The minimal required version of
libxcb is now 1.9. Bundled xcb sources were updated to libxcb 1.9.1
built with xcb-proto 1.8.
Change-Id: Iebcd05656c4a5ed5dd95e898d497acef857423f0
Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
We report it in the overview, but left no way of controlling it
Change-Id: I1d31f2e31bb32566f47069c3776e41033ffb1891
Reviewed-by: hjk <hjk@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
The resize event should not be used to draw, as there will be a follow-up
exposeEvent delivered just after, and we don't want to draw twice.
This is how QRasterWindow operates too, except it defers the backingstore
resize to beginPaint() instead of resizing in the reizeEvent.
It's also how QOpenGLWindow operates, which also has a note for the
virtual resizeGL method saying: "Scheduling updates from here is not
necessary. The windowing systems will send expose events that trigger
an update automatically."
Change-Id: I2a9740018508c2eb129149f53237ee8e378c03b1
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
The updateLayer function is only called when the layer is a
_NSViewBackingLayer instance, which is what the default
implementation of [NSView makeBackingLayer] returns.
Once we move to optionally returning a CAMetalLayer, we need
to use the more generic displayLayer: API, so we do that now
as a first step.
This matches the way we draw and send expose events on iOS.
Change-Id: I49721ff005ca9dfddebff645705f96b5ab46abb4
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
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>
Use QT_CONFIG(regularexpression), disentangle it from QT_BOOTSTRAPPED,
switch it off in the bootstrap build, remove the #ifdefs from
qregularexpression.{h|cpp}, and add QT_REQUIRE_CONFIG(regularexpression)
to the header.
qregularexpression.{h|cpp} are already correctly excluded in tools.pri
if !qtConfig(regularexpression).
Change-Id: I21de154a6a118b76f99003d3acb72ac1e220d302
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Require it in the headers and exclude the implementation from the build
if disabled.
Change-Id: Ida3303f8595f47b469e92d68e8bccc3957d943b6
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
QRegExpEngine is immutable once created, so we can easily share
them between different QRegExp instances. This requires a QHash for
engines that are currently in use in addition to the cache for
currently unused engines.
Task-number: QTBUG-65710
Change-Id: I165c12a7b065c488ecd14258a7cdfe0e62666632
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Loading empty glyphs in some fonts would cause the application
to crash with an infinite recusion on the stack between qfontengine
and qfontengine_ft:
12 0x00007ffff6954395 in QFontEngine::alphaMapForGlyph (this=0x7fb0c0, glyph=50, subPixelPosition=..., t=...) at text/qfontengine.cpp:877
13 0x00007fffee7d104e in QFontEngineFT::alphaMapForGlyph (this=0x7fb0c0, g=50, subPixelPosition=..., t=...) at freetype/qfontengine_ft.cpp:2096
14 0x00007fffee7d0ea8 in QFontEngineFT::alphaMapForGlyph (this=0x7fb0c0, g=50, subPixelPosition=...) at freetype/qfontengine_ft.cpp:2078
15 0x00007ffff6954395 in QFontEngine::alphaMapForGlyph (this=0x7fb0c0, glyph=50, subPixelPosition=..., t=...) at text/qfontengine.cpp:877
16 0x00007fffee7d104e in QFontEngineFT::alphaMapForGlyph (this=0x7fb0c0, g=50, subPixelPosition=..., t=...) at freetype/qfontengine_ft.cpp:2096
17 0x00007fffee7d0ea8 in QFontEngineFT::alphaMapForGlyph (this=0x7fb0c0, g=50, subPixelPosition=...) at freetype/qfontengine_ft.cpp:2078
18 0x00007ffff6954395 in QFontEngine::alphaMapForGlyph (this=0x7fb0c0, glyph=50, subPixelPosition=..., t=...) at text/qfontengine.cpp:877
...
Fix this by trusting the freetype fontengine that it could load the glyph,
as the base class anyway can't do better.
Task-number: QTBUG-62331
Task-number: QTBUG-66617
Change-Id: I6c7c24d24ec0f71a66fa519c04a336f276e418f6
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
The parsing of the DISPLAY environment variable was wrong,
and only worked by accident for DISPLAY=':[0-9]'.
Task-number: QTBUG-62068
Change-Id: I6860e3907c9b1ad6e538d1b5d08628cd306b4aa1
Reviewed-by: Alexander Volkov <a.volkov@rusbitech.ru>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Generated files should be added to RESOURCES with an absolute path.
Task-number: QTBUG-67011
Change-Id: Ief82b576824df9abd0901970f076e30dfe57b7d0
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
If we pass nullptr as a buffer parameter, ERR_error_string uses
a local array with static storage duration, which makes ERR_error_string
non thread-safe. Instead we can use ERR_error_string_n with our
own buffer. As for the size: docs are inconsistent, sometimes
they say 'at least 120 bytes long', sometimes 'at least 256 ...'.
Their code (ERR_error_string implenented via call to ERR_error_string_n)
has buf[256] and so we do. I know this will enrage our Mr. Bot, but I've
removed a stray whitespace.
Change-Id: I4b8cc7b6b9af1a34fc87a760927a493332cdd0a5
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
This makes the pointer that was in the moved-into object be destroyed
before the return of this function (if the reference count was 1),
instead of letting it live in the moved-from object.
Task-number: QTBUG-66322
Change-Id: I3debfc11127e4516b505fffd151209292bd3adaa
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Since we won't use the high bit of the low 32-bit word at all, we don't
need the AND with 0x7fffffff either. Just cast.
Change-Id: I72f5230ad59948f784eafffd151aa5a7dee995db
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
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>
No longer needed. The comment about missing constexpr support is
incorrect: MSVC 2015 does have constexpr issues, but they don't affect
our use of std::atomic.
Change-Id: Ie9d9215342d449c48a11fffd151d11208137f00d
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
MSVC atomics still use the generic version, instead of qatomic_cxx11.h.
The implementation of fetchAndSub is implemented on top of fetchAndAdd,
but produced a warning with unsigned types.
Change-Id: I72f5230ad59948f784eafffd151aa53435b75298
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
selectFile() is a no-op if the dialog is visible and
its line edit has focus, which will happen at some
point after show() is called.
In other words this is a race, and the test will pass/fail
depending on the timing of the platform implementation.
Task-number: QTBUG-66536
Change-Id: I03957edb6496d17af7a8ed1d4ca86435375e0d40
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Don’t assume that there is an active modal widget
on QApplication::focusChanged(). That signal can also
be emitted on focus clear during dialog destruction.
Task-number: QTBUG-66536
Change-Id: I20c64339c56a52b39c26a7683b62779deba576d9
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Add null-pointer checks to QCocoaEventDispatcher::
currentModalSession():
- window->handle() may return nullptr if the window
has been destroyed.
- We call beginModalSessionForWindow, which processes
events. This can potentially destroy or delete the
current window; pointers to it must be checked again.
This also has the effect that currentModalSessionCached
is not set to a session that does not have a window,
also prevents clearing the cleanupModalSessionsNeeded
flag for such sessions.
Task-number: QTBUG-66536
Change-Id: Ie055933c872a8ede3c938882fb2d4f7cf8ff3c81
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Add qWaitForWindowExposed() calls in addition to the
existing qWaitForWindowActive().
Depending on the platform, window activation events may
be synchronous, which means that a window can (and will)
become active before it becomes exposed.
This causes test failures for tests that count paint
event, and does not wait-for-exposed, when the expose/paint
event is delivered after waitForWindowActive() has returned.
We need to keep the waitForWindowActive() as well: the
test has several qWait() calls with he comment:
“Increase the probability of window activation not
causing another repaint of test items”.
These qWait() calls can possibly be removed in a future
commit.
Task-number: QTBUG-66536
Task-number: QTBUG-61967
Change-Id: Ie61bba058b583fdd1d80e600475aff3efccc32eb
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
This test function was removed from the blacklist as
a part of the cleanup in 2056bc6b, but is still flaky
on CI.
Change-Id: I348ed167dff6fd618ede28456ada6da86ee341cd
Reviewed-by: Paolo Angelelli <paolo.angelelli@qt.io>
Previously we were using a treeview that wasn't very clear it
was editable, now we simply use QComboBoxes that are much
more clear the user can change the values
Change-Id: Ied9bca195f4cb275115213631e21cd6a15544311
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@qt.io>
This patch adds several macros, functions and typedefs,
needed by DTLS, into our qsslsocket_openssl_symbols.
Change-Id: I9e4dccc0c576b26b3f629cee6e3245e707604674
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@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>
Thiago Macieira
Ulf Hermann
Allan Sandfeld Jensen
Marcel Krems
Morten Johan Sørvig
Nathan Collins
Mårten Nordheim
Friedemann Kleint
Qt Forward Merge Bot
Kai Pastor
Tor Arne Vestbø
Sergio Martins
Alexander Volkov
Lars Knoll
Mikhail Svetkin
Simon Hausmann
Timur Pocheptsov
Albert Astals Cid
Johan Klokkhammer Helsing