This was originally added so that you could replace a T with
QAtomicInteger<T> in the same class and still keep ABI. However, for
legacy reasons, on 32-bit x86, types larger than 4 bytes keep an old
1990s alignment of only 4 bytes, but modern std::atomic<T> for those 8-
byte types enforces an alignment of 8 bytes. Therefore, the requirement
to keep alignment is not possible to guarantee.
In other words: you may not replace T with QAtomicInteger<T> or
std::atomic<T> and assume no ABI breakages in all platforms.
This is a requirement to implement atomicity. An 8-byte type aligned to
only a 4-byte boundary could cross a 16-byte boundary or, worse, cross a
cacheline boundary. Crossing the 16-byte boundary could be bad on some
processors, but crossing the cacheline boundary (addresses ending in
0x3C, 0x7C, 0xCC and 0xFC, or 4 out of 64 possible addresses or 6.25%)
is always bad: the CPUs cannot guarantee an atomic load or store
operation.
See also <https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71660>.
Task-number: QTBUG-67858
Change-Id: If90a92b041d3442fa0a4fffd15283e4615474582
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@qt.io>
We need to mark the object as deleted before destroying
it's declarative data, otherwise all sorts of bad things
can happen.
This fixes the qwidgetsinqml autotest in qtdeclarative.
Change-Id: I05a645ebe1ca7a50c8927e3dbd9ebb5aaf369a71
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
(cherry picked from commit 3e91625b58)
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
There had been a fake declaration for qEnvironmentVariable() in qglobal.h
thaqt was only visible to QDoc. It was removed in favor of documenting
both the actual declarations of qEnvironmentVariable(), one with a 2nd
parameter for passing a defualt value and one without that parameter.
But the one without the default value parameter was marked internal, so
it didn't appear in the docs.
When both functions were documented with a shared comment, a bug in
QDoc was revealed, because these functions are global, while the shared
comment functionality had only been implemented for class member functions.
Now the shared comment functionality has been implemented for global
functions, so these two functions are now documented with a shared
comment. We can, of course, reintroduce the #ifdef QCLANG_QDOC trick, if
that is pre3ferred.
Change-Id: I41d85def5daa3215a995d7697d064dfae37e8b2a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
That check is flaky on Windows. It doesn't seem to be testing Qt functionality.
I also don't see CreateFile2() documentation mentioning any guarantees that
opening the same file twice would give the same HANDLE each time.
Change-Id: Ica2e60571ae9fc39bf822803a2a9dd6add8323d7
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
This fixes qdoc warnings introduced by 67d5f79fe6.
Change-Id: I4b199e6243d9a7706befe4bc9549c78c11026d9e
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
We're relying on C++11 since a while, so lets not advertise creating
lists of strings with operator<<() anymore.
Change-Id: I14a3442ff852ac2c106d90c63504eb9ebb737609
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: hjk <hjk@qt.io>
We need to take into account the presence of a possible ',_debug' suffix.
Change-Id: I5655394b78723bbc6cc32e56849acc2366d288e2
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Once connection is established the socket emits 'connected', and then
you can start writing. But it will end up in the write-buffer and won't
get sent until 'transmit' is called. Some code (e.g. QWebSocket) relies
on QSslSocket transmitting once it's encrypted. This is done in the
OpenSSL backend but was not done in the WinRT backend.
Task-number: QTBUG-56558
Change-Id: I8cf5d3257f3597a4bb80f35369490a3816506a34
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Change the documentation to use nullptr and modify the related code
also while at it.
Change-Id: I6264a254828159cda54e90393835ea04e131350b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Use the match with the least modifiers (prefer Shift+9 over
Alt + Shift + 9) resulting in more missing modifiers.
Task-number: QTBUG-67200
Change-Id: I90463c0dfaadda29dcd24a08ba35c91fac8bd04c
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Add utility functions to QWindowsContext input messages sent to
native event filters (event dispatcher and window system interface).
Do not send input and similar events to the event dispatcher since
QEventDispatcherWin32::processEvents() also sends them.
Note though that QEventDispatcherWin32 does not receive all windows
messages.
In QWindowsKeyMapper, send the WM_CHAR/WM_IMECHAR events from
where the messages are removed.
Task-number: QTBUG-67095
Change-Id: I5f61294fcb3aee7e0eacd840a2010d128cd02a5d
Reviewed-by: Andre de la Rocha <andre.rocha@qt.io>
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
When a key press is received which is not spontaneous then it needs to
be manually sent as a shortcut override event to ensure that any
matching shortcut is triggered first. This enables emulation/playback
of recorded events to still have the same effect.
[ChangeLog][QtWidgets] Sending a key press event with sendEvent() now
sends a ShortCutOverride event first to the widget to trigger any
shortcuts set first.
Task-number: QTBUG-48325
Change-Id: Iafcc2cdb1773bffe89edaeb0abc44cd5a51088e6
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
[ChangeLog][Third-Party Code] Libdbus-1 code in Qt D-Bus is now properly
documented. The code is licensed under AFL-2.1 OR GPL-2.0-or-later.
Change-Id: I38bd6f38175562fff03ae84fc7a5c435de4cb972
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Wrap current example code snippet in std::function as a work-around,
as suggested in comment to QTBUG-61145 (see history tab) with the same
issue.
Task-number: QTBUG-67603
Change-Id: I6875b31d8e983e234b88384c7d76917ac144f953
Reviewed-by: Topi Reiniö <topi.reinio@qt.io>
In 5.11 robustness was added for nvidia drivers, this patch extends
that to other opengl drivers, and fixes the reported format.
Change-Id: Ia81934c3bcf13e7300fb4e75674ea63317039870
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
the 'info' variable was re-used too early. make a new one 'infoargs'
instead.
Task-number: QTBUG-67286
Change-Id: I77881ecbfce338d653358c5e5edac84e1c0c7de3
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
In iOS 9.0, the original application:openURL function was deprecated
and replaced with a newer one. As iOS 9.0 is no longer supported we can
safely switch to the new one. This is also required to prevent a crash
when the LSSupportsOpeningDocumentsInPlace and UIFileSharingEnabled keys
are set to true in the Info.plist file.
Change-Id: I59a7ee82e3ddb2777ef78e28b964ef8666c629af
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
This brings back Qt 4 behavior. The difference is that we only
ignore modifier-only key events, as it's done natively.
Task-number: QTBUG-49462
Change-Id: I02f2313e1164ba185336d80ac5cc16ce6d883b79
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
It use to be yellowish in the past, but modern versions
of macOS show it light gray.
Change-Id: I8cca5cbb37c73a6dfc79e633a746b9a7d7bced05
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Because we use toggle NSButton for selected tabs, the inactive
appearance doesn't follow what NSSegmentedControl would have
shown. Therefore, we fall back to our good old habits, i.e.,
render on a pixmap and do some pixel transformations.
Change-Id: I838a2f23abee5846219ba67328c79fa8cc359a9b
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Change QMacStylePrivate::drawNSViewInRect() signature
to remove all the unused parameters.
Reuse recent tab direction functions where appropriate.
Includes the infamous outter -> outer fix.
Change-Id: I8f92d79d8a6c3b5903bfbb13293afb6f72a5340b
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
The idea was probably to ignore it since macOS already provides
accessibility for a menu item in its native system menu. However, a Qt
Quick Controls2 Menu will instead show a non-native menu, which should
not be ignored.
Task-number: QTBUG-63522
Change-Id: Ib5ae16ad991ebd7a18fa73b8f576f20b1c14d4c8
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@qt.io>
In order to be able to build and test a testcase on an iOS device it
needs to be a bundle. So the app_bundle config should only be removed if
the testcase_no_bundle is set. This is already done by testcase.prf.
Task-number: QTBUG-45211
Change-Id: I4f16ea832ccff2a5db5fed0050fa0344b4ac9ad6
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
When rotating the device when the application is in the background iOS will
ask the root view controller to layout its views when then foregrounding the
application, but at that point the application state is still in the suspended
state, meaning we block any view resizing or rendering.
To ensure the views are resized correctly, we trigger a manual layout after
the application comes out of the suspended state.
Task-number: QTBUG-67719
Change-Id: I1ef0a4133d4b94edaac7b0f3cb4e49e367eb76d4
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
The watchOS and tvOS platforms also define __IPHONE_OS_VERSION_MIN_REQUIRED
for compatibility, so we need account for that in the ordering of the ifdefs.
Task-number: QTBUG-67534
Change-Id: Id86e684137550533470370ef29c3563d677d5865
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@qt.io>
With Cocoa it is possible to have a parentless QMenuBar so that the same
one can be used for multiple windows. Therefore if the top level window
is the QMenuBar when checking for the context then we can let it carry
on as if the top level window is the same as the active window.
Task-number: QTBUG-45453
Change-Id: Ifacf2111d5f9973afe8af30c6338918f130e51a4
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
It would be tedious to bump them every year, but we don't want
people to copy completely outdated year numbers either ...
Also consistently use https:// in contact address.
Change-Id: I9654417a3037fc18c6b9a0551c4883f5fc960084
Reviewed-by: Jani Heikkinen <jani.heikkinen@qt.io>
Commit 71404b0be1 added new header templates and LICENSE.GPL2,
LICENSE.GPL3, LICENSE.GPL3-EXCEPT, LICENSE.LGPL3 files already in 2016.
So it's finally time to remove some of the old licnse files.
LICENSE.LGPLv21 and LGPL_EXCEPTION.txt is not referenced anywhere
anymore. LICENSE.GPLv3 is referenced only in LICENSE.LGPLv3.
We unfortunately still have to keep LICENSE.LPGLv3 because it is up
to this day used in the standard header templates, but we can just
change the reference in there to LICENSE.GPL3.
Change-Id: Ia88b4de8a395757a39ca63d474b2214e3b1063fb
Reviewed-by: Jani Heikkinen <jani.heikkinen@qt.io>
The drive has to be defined for every Windows configuration (also
including winrt).
Change-Id: I94a3131b8aec20cda97dc78f55b1d87aa10240e4
Reviewed-by: Maurice Kalinowski <maurice.kalinowski@qt.io>
When releasing a STACK_OF(GENERAL_NAME). Actually, GENERAL_NAME_free is
a special function, not the same as OPENSSL_sk_free.
Task-number: QTBUG-57679
Change-Id: I3ed300bb95e8be35bd9cd06b6dbc6e59c7c6a4ee
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
The android mkspecs had their own way of doing the same as optimize_size
Change-Id: Id05822df6bdeb8b3aafada2901bd61530c490fe9
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: BogDan Vatra <bogdan@kdab.com>
We were using xcb_render_* APIs without checking if the server even
supports this extension. Attempting to use an extension which is
not present will always result in a crash. This patch adds the required
guards and refactors how we detect presence of XRender extension.
Also instead of falling back to some odd-looking bitmapped version
just leave the current cursor unchanged. That is how we did it in
Qt4 AFAICT.
Task-number: QTBUG-66935
Change-Id: I4f27f1d65a77563ec34f3e0e94492c9236d7f9a6
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
... introduced by 422838685c.
Instead of completely droping caching for bitmap cursors we can do
the same what is done in libXcursor - have a fixed size cache, with
oldest entries eventually being replaced with new bitmaps. This
fixes the original issue, where the hash was growing indefinitely
until running out of file descriptors and won't have the performance
penalty as in 422838685c.
Task-number: QTBUG-66897
Change-Id: I14f80b46f97fd0e2c920e17a31ffbc0441cd9d22
Reviewed-by: Uli Schlachter <psychon@znc.in>
Reviewed-by: Robin Burchell <robin.burchell@crimson.no>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
- removed the check for "m_segmentSize > 0" as according to the code it
will never be <= 0.
- wrap the entire logic in connection()->hasShm() { .. } as that is when
the logic becomes relevant. This makes the code more readable.
Change-Id: I572420df8e29cc46593f8a13c250f8c05c6a9108
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Avoid creating a QDateTime in the resource that will almost
never get used. Constructing the date time is expensive as we
convert the time stamp to local time.
Task-number: QTBUG-65713
Change-Id: I3638e108a8fbd237cd93e98aa2adc0ca2127822c
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Tuukka Turunen <tuukka.turunen@qt.io>
