QMAKE_PRL_LIBS contains absolute paths to libraries, e.g. libqtpcre2.a.
On "make install" the libdir is replaced with the installation target
libdir. If the libdir is in QMAKE_DEFAULT_LIBDIRS (e.g. /usr) then the
replacement was empty. That worked fine for include paths but not for
paths referencing files in that libdir:
/my/build/lib/qtbase/lib/libqtpcre2.a would become /libqtpcre2.a.
Add another replacement that takes care of file paths and inserts
$$[QT_INSTALL_LIBS].
Fixes: QTBUG-75460
Change-Id: I4e84478a50c24d4143ad5695493cad2992735cf2
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Samuli Piippo <samuli.piippo@qt.io>
- Never for global inputs
- Otherwise only if the temporary is referenced more than once
-> meaning it's actually caching the result of some operation
Tests updated accordingly.
Change-Id: Ic76615370d23dee3965ca6350d5257a8be5a3e22
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
The old code in QtActivityDelegate.updateHandles() and
EditPopupMenu.setPosition() could use size of EditPopupMenu.m_view to
calculate position of context menu before that size was calculated
during an asynchronous layout pass. In particular m_view reports size
0x0 when context menu is opened for the first time after start of the
application. In this case the context menu was displayed on top of the
text editor instead of being displayed above it.
This patch fixes that problem by moving all positioning code from
QtActivityDelegate.updateHandles() to EditPopupMenu.setPosition() and
adding an OnLayoutChangeListener which calls setPosition() again each
time the size of m_view changes, including when it changes for the first
time from 0x0 to a real value.
Change-Id: I670fef811a4dcba5524f7520ea41a47978dd10f1
Reviewed-by: BogDan Vatra <bogdan@kdab.com>
The old code used size of m_cursorView to calculate position of the
cursor handle popup. But since on Android layout process is
asynchronous, both width and height of m_cursorView was 0 upon first
call to setPosition(). This resulted in selection handles being
initially displayed at a wrong position for a fraction of second and
then quickly moving to the correct position. In some cases handles
stayed at the wrong position until touched by user.
This patch replaces use of m_cursorView's size with use of m_popup's
size. Width and height of m_popup may be used immediately because they
are explicitly assigned in initOverlay(). The size of m_popup should be
always equal to the would-be size of m_cursorView because of how it is
calculated.
Change-Id: I9868c9a5ce0103d8328b2478cf82feaceba7f404
Reviewed-by: BogDan Vatra <bogdan@kdab.com>
The file system appears to cache too aggressively, so if the reported
storage size doesn't change after flushing to disk, ignore the failure.
Change-Id: Iba7dce79591447fac296bfe92c2dc993d36d0c2a
Fixes: QTBUG-69868
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When we trace a scope, then we pass the scope args to the _entry trace
point. There is no need to do that also for the _exit trace points, it
just blows up the trace data for no obvious gain. Any decent tracing
consumer can easily find the args for the _exit call by matching it to
its _entry call.
Note that this is standard practice in trace points, and also done like
this in the Linux Kernel trace points for example.
Change-Id: I273293b0c7e799767acc1960b50ab675fc765a36
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Use Q_TRACE_SCOPE and the corresponding naming scheme. Additionally,
don't change the behavior of the code when tracing is enabled, i.e.
continue to return early if possible.
Change-Id: I9ba9679869db1541a19bc832beede902224c52f2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The remaining diff to clean 1.6.37 is archived in the qtpatches.diff file.
[ChangeLog][Third-Party Code] libpng was updated to version 1.6.37
Change-Id: I589bff09beec1977be8c6ca2a60aadf05f337f38
Reviewed-by: Liang Qi <liang.qi@qt.io>
Change mentions of parameter value being "zero" to nullptr.
Clarify that when nullptr is passed to moveToThread() event processing
is stopped because the object is no longer associated with any thread.
Also, reitarete this fact in the paragraph about processing of new
events.
There's an exception to the rule that QObjects cannot be "pulled" by
moveToThread that is buried in the implementation and not mentioned
in the doc. This information is worth noting explicitly.
Change-Id: I816ff737c48d8057b39e36b566079710aeb8e690
Reviewed-by: André Hartmann <aha_1980@gmx.de>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
The helper in QOpenGLDebugLogger did not account for the extra "KHR"
suffix in the function names and results in a crash when ANGLE is being
used. This commit fixes this issue.
Change-Id: I439d8bfc53b010be5410286b86c090aff171aaef
Fixes: QTBUG-62070
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
CreateCursor only works with standard sizes (32, ...) depending on the
display hardware. No longer apply the scale factor for the blank cursor, because
it might lead to unsupported cursor sizes resulting in random pixels.
Change-Id: I48d84bd913d2dd8f62129126c9a41e58ee2cbcae
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
The sysrootification of QMAKE_INCDIR_OPENGL on macOS must happen only
once. Commit 49ef3773 addressed this but stored the sysrootified
QMAKE_INCDIR_OPENGL in qt_lib_gui_private.pri. For installer packages,
these paths are the paths of the build machine and most likely wrong
on the user's machine.
This reverts commit 4949ef377349ba4dae840c2d5caa36e2d516707baa and
restores the sysrootification in sdk.prf. The original include paths
are assigned to QMAKE_EXPORT_INCDIR_OPENGL and stored as
QMAKE_INCDIR_OPENGL in qt_lib_gui_private.pri.
Fixes: QTBUG-75374
Task-number: QTBUG-73736
Change-Id: I4c0f65866d60660c632363dba3adc7ea2e344bfc
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
QFile::copy() didn't have the syncToDisk() call that QSaveFile::commit()
has. So add it.
[ChangeLog][QtCore][QFile] Made QFile::copy() issue a filesystem-
synchronization system call, which would make it less likely to result
in incomplete or corrupt files if the system reboots or uncleanly shuts
down soon after the function returns. New code is advised to use
QSaveFile instead, which also allows to display a progress report while
copying.
Fixes: QTBUG-75407
Change-Id: I95ecabe2f50e450c991afffd1598d09ec73f6482
Reviewed-by: Henrik Hartz <hhartz@gmail.com>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Don't go into an infinite loop breaking pages, when an image is about
as large as the page. Correctly take top and bottom margins into account
when calculating whether the image could fit on one page.
Amends change 416b4cf685.
Fixes: QTBUG-73730
Change-Id: Id311ddf05510be3b1d131702f4e17025a9861e58
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Fix warnings
qtbase/src/corelib/io/qprocess.cpp:776: (qdoc) warning: Class CreateProcessArguments has no \inmodule command; using project name by default: QtCore
qtbase/src/corelib/serialization/qcborstream.cpp:1441: (qdoc) warning: Class StringResult has no \inmodule command; using project name by default: QtCore
Change-Id: I1c85ca32aff1f89f70898af7b11cfead96c80349
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Using getaddrinfo, which implements RFC 6724, implies that addresses
that are not needed will be trimmed. In particular, IPv6 addresses are
often not returned.
Also move the implementation detail documentation down in the text, it's
a detail with little relevance for the usage of the class, but makes for
a good opener regarding this behavior.
Change-Id: I516a64f0b39a6a06621a63c1d5236544b7758049
Fixes: QTBUG-31865
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Otherwise, AppKit, while rendering 'detached' (not in any view hierarchy)
controls and cells will use NSAppearance.currentAppearance, which is not
guaranteed to be the same as NSApplication.effectiveAppearance.
Task-number: QTBUG-74515
Change-Id: I82dcebf2230932ecfcbf33c422a3b7bd0aed61d7
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Keep spinning the main event loop if we can't acquire the semaphore,
this way the Android UI thread can post events on it.
Fixes: QTBUG-74076
Change-Id: Ia87e0535f94c67728176918ab928ff5ce8b00f8e
Reviewed-by: VaL Doroshchuk <valentyn.doroshchuk@qt.io>
There is an effort to make KDE software accessible, which exposed the
missing roles. Check that they are complete with an assert.
Change-Id: Ibaff0a90e1cee316983569ecee7759a13212e3c3
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
When building in a unix style build system (i.e. msys), QMAKE_DIRLIST_SEP
is a colon, not a semicolon. Thus, always split the incoming string
(after the fixup regex) using semicolons on windows.
This matches the code for gcc, further up, which does:
equals(QMAKE_HOST.os, Windows): \
paths = $$split(line, ;)
else: \
paths = $$split(line, $$QMAKE_DIRLIST_SEP)
Change-Id: I6a0175f9d14ae9ca188553483b7868f0549c784a
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
stdcpp-path is needed to set the correct stdc++ library in libs.xml file.
This reverts commit 1366c4f046.
Change-Id: I79b398c5d97c1e98bf503ef7b95b2e9f0f18bc11
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
Images in an ico file contains transparency information stored as a 1
bit mask. However, when the depth is 32 bit, it means there is an
alpha channel present, and the mask should be ignored. The Qt ico
handler failed to do that. This has gone unnoticed, since the mask in
such images is typically set to all 0s, and so makes no difference to
the result. But ico files exist that contain junk mask data, so fix
the reader to ignore it properly.
Fixes: QTBUG-75214
Change-Id: I1b4456d71689ec783076a582f2fb215e7dc56e62
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The full stroker does not produce good results for aliased lines
thinner than 1 pixel. Avoid it by making sure that such thin lines
are painted by the cosmetic stroker, even when they have
non-uniform transformation.
Fixes: QTBUG-73866
Change-Id: I7b5f0fa555903246e0c3fd92cd435cc8c0b15a24
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The dash stroker did not check for inf/nan coordinates.
Fixes: QTBUG-47887
Change-Id: I1e696cd15cc37d8fcb6a464cac3da33c3a8b95c2
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
We don't support CMake version 2 anymore. Instead of just updating
to a newer (but fixed) version let's link to the latest
documentation. This might create a bigger risk that links get
stale, but hopefully let people find always the latest information.
Task-number: QTBUG-72159
Change-Id: I082de80cf9ee107b5d017ab8ad6369f2448b0e1b
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
This fixes the "could not parse compiler option" warning when
generating VS project files.
Fixes: QTBUG-75275
Change-Id: Idd98ae5fdb8ebf5a4e311cbb6cd3ed1daba74ca4
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
For gcc's -I and -L arguments a = prefix is replaced by the sysroot.
Since we're resolving include paths and library paths, we have to
support this feature.
For example, the linux-rasp-pi3-g++ makes use of this and is broken
without this patch.
Change-Id: Ie39e63322bd35e2a93aa8e55d52260164b8c6a6b
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
Injected headers were made relative to MODULE_BASE_OUTDIR by syncqt
and made absolute by resolving against REAL_MODULE_BASE_OUTDIR.
This breaks for modules that reside outside the original Qt source
tree (if the directory depth doesn't coincidentally match).
Now, we resolve injected headers against build_basedir, which is
REAL_MODULE_BASE_OUTDIR. To emphasize the equivalence of
REAL_MODULE_BASE_OUTDIR and syncqt's build_basedir, use the former for
syncqt's -output argument.
This commit amends 2aa779e8.
Fixes: QTBUG-70587
Change-Id: I2935d87d7ee681fa4aa795a270b94ab7a43abe59
Reviewed-by: Dominik Holland <dominik.holland@pelagicore.com>
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
If the client of the backingstore fails to pick up dpr changes, and
tries to flush the backingstore without a repaint, we will end up
flushing a back-buffer with a stale dpr. Detect when this happens,
warn the user, and smooth out the situation by adjusting the layer
accordingly.
Change-Id: If4596a8976a3902252c81d8e28c7aeb9fdd908bf
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
We now use QCFType to track the CGContextRef, instead of manually
maintaining the lifetime of the context. A bunch of unused methods
were removed, including completely broken ones like isNull().
Change-Id: Ib5a05aadbf8f228192e74c9a4c8919580b831497
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
For WM_APPCOMMAND messages that also trigger WM_KEYDOWN/WM_KEYUP, let the
latter messages be handled, instead of stripping them and synthesizing a
press/release from the command code, in order to get the correct scan codes
and autorepeat info.
Fixes: QTBUG-73879
Change-Id: I936cd76be87a76dc6b6223eeb246e4e7aee3a4ac
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Changes QWindowsDropDataObject to only ignore non-CF_HDROP formats when
the drop contains only "text/uri-list" mime data, and the URIs are for
local files, to avoid messing with custom formats set by the developer,
while still fixing the case reported in QTBUG-62662.
Fixes: QTBUG-74232
Change-Id: I946ced222377716876d0aea54b3eb05d40e7fa44
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Add 'source' window offset. This covers the cases where platform
window is created besides one full screen window (like for popups),
where content has qquickwidget / qopenglwidgtet. In that case
fbos/textures from those widgets have offset according to 'source'
window. Note backingstore texture has geometry of 'source' window.
Task-number: QTBUG-69533
Change-Id: I2514b36fd3a6b9b86f51999df1c2b3e9565aafde
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
WebEngine HTML based popups depends on setting Qt::WA_ShowWithoutActivating,
to keep forwarding events to chromium event handling. This works well with xcb,
windows or coca windows backends, however was not respected on eglfs.
Add check before activating the window.
Task-number: QTBUG-69533
Change-Id: I66b249ec497af890c8a2228eee3bac3c806e77ed
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
If the shared context had robustness set then all child contexts must
as well, otherwise we will fallback to a non-shared context breaking
threaded rendering.
Change-Id: Ie5526e632ad21289b6164c1ca06e54ec714187c7
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
If the displays are not cleaned up on dll unloading, profilers might
report memory leaks.
Change-Id: I04cbc3c2448bfb450f7d840e216827f86856e963
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Andre de la Rocha <andre.rocha@qt.io>
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Calling destroy from the QEglFSWindow dtor() triggers the virtual
invalidateSurface() to be called on a partly destroyed object. As the
child windows deregister themselves from their screens on
invalidateSurface() this is dangerous: It leaves a dangling pointer in
the screen.
Fixes: QTBUG-75075
Change-Id: Idd3fea18562d41973f364340df875a50dbd5691e
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
this avoids the scenario where the linker would pick up the wrong qt
libraries for LIBS_PRIVATE because LIBS added the "wrong" path first.
this is also consistent with configure-supplied dependencies as of
recently.
as a side effect, this also removes pretenses of lsb linker handling, as
it makes no sense after the change and is certainly obsolete anyway.
Fixes: QTBUG-50921
Change-Id: I84398c9143f393c2eefb3c69a31bd9f633669924
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
[ChangeLog][QtCore][QMetaObject] Non-copyable lambdas can now be used
with invokeMethod(). For consistency reasons, the functor object is
now always moved.
Fixes: QTBUG-69683
Change-Id: I66ff5e21d6d1926f0f7c5f8c304bed1a90b69917
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
Screen readers like NVDA and Narrator were not reading the contents
of a QComboBox when changing its value using the keyboard, without
expanding it, due to missing UI Automation notifications in this case.
This change should also help in other cases where updated string
values were not notified to screen readers.
Fixes: QTBUG-75066
Change-Id: Id7f488380aec5ad27fd11b3cf854d44ab1b28688
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>