| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
| |
Adds test for user-agent override and fixes the override so that it
also works when loading with content instead of URLs.
Change-Id: I3f61b1d91b7b0908e35216722054168d1c514a87
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
| |
BrowserContextAdapter is not a form of shared data, and using
QExplicitSharedDataPointer on it as always been a misuse of the class.
Instead we should switch it to QSharedPointer, which also allows us to
use QWeakPointer.
Change-Id: I8eb489b4a12d3fdddcde55821be294814a156a9d
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
Reviewed-by: Michal Klocek <michal.klocek@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: If20959319ab8fb1ff7281bba6a1dfcfaabcc1989
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
|
|
|
|
|
|
|
| |
Assert and thereby also document thread assumptions.
Change-Id: I44cd8b015ed8a6e38b9328f334f55d9d2f72b50d
Reviewed-by: Michael Brüning <michael.bruning@theqtcompany.com>
|
|
|
|
|
|
|
| |
Pulls in update that makes remote inspector compatible with Chrome 50+
Change-Id: I5e5db318228fa77c650f058fc834568c1d5bb6ad
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
| |
ICU .dat file and .pak files were already copied to
the developer build directory in core_module.pro. Let's
do the same for translations, and remove the logic
handling this in resources.gyp.
Change-Id: I4926cd6012f2664feed27aa042eb3ceb282a93d9
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I008b649b43069978b6ce19b2faf5a557dd67c756
Reviewed-by: Michal Klocek <michal.klocek@theqtcompany.com>
|
|
|
|
|
|
|
| |
Pulls in security fixes from Chromium 50
Change-Id: I46f331fd37b17366231469c1b5aa2df9bea75291
Reviewed-by: Michael Brüning <michael.bruning@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When exiting fullscreen while viewing a full screen video, the video is
replaced with a black screen, and does not switch back to the initial
page.
This happens when hide() is called on a native window while being in
fullscreen, which hides the window, but does not exit full screen mode,
thus showing a black screen.
The workaround fix consists of two parts:
1) Make sure to delete the fullscreen window object, to release the
OS virtual screen it occupies (visible in mission control as a black
rectangle).
2) Hide and show the main page window, to force exit fullscreen mode.
Change-Id: I364d4427e84957ed97f57a14e2cfc2c95aed6f48
Task-number: QTBUG-52058
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
| |
Get rid of dangling pointer on storage removal in
URLRequestContextGetterQt. This is accessed later
in generateHttpCache.
Task-number: QTBUG-52468
Change-Id: I03c0b3186b01046e17258af838bf5f9adc435da0
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
I could not reproduce the issue, nor the CI. Anyhow, it doesn't hurt to
include the header explicitly, instead of relying on indirect includes.
Task-number: QTBUG-52460
Change-Id: I78d04f3af853aa5951cdfc066089f06f71f72988
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: Iee69fd2b033cfe09d5347e8502812632d1309b19
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
While it is better to use the global shared OpenGL context, we can
fall back to using the module type which is usually correct.
Change-Id: I441bfdc38db593abfd10cb08ba90af09dcd31a2a
Task-number: QTBUG-51379
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
| |
If a URL interceptor changes the request but doesn't block it, we
currently do not send the navigation requested signal to the API layer.
Instead continue as normal when not blocking.
Change-Id: Ida6aad5767fcaef58093c50aa3932092334f499c
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I27981327fb563fdd6697352582c2cbc99b173ccc
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
Pulls in build fix for MSVC 2015 Update 2.
Task-number: QTBUG-52367
Change-Id: I57f17388b280eb79397cbba3e6b1b546d30be096
Reviewed-by: Michal Klocek <michal.klocek@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
The property was already revisioned, but the signal was not. Also
mark the property as new in the documentation.
Change-Id: Iab50bea0d593e9dafb3e58b7f99dd5b95655a1a9
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It is needed to let local content access the qrc resources by making them
part of the same security origin. This is stated in the docs as the default
behavior.
This also backports change f7d343000d77c80f34a115fe9c78f101929292b1
from 5.7, which fixed some race condition with the qrc registration.
Task-number: QTBUG-52085
Change-Id: Ia704fa6dbce8529f82695ab543244020cb9f8b8a
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
|
|
| |
The test passes now.
Change-Id: Ifc96c995036759d554938dbbc80ca05a5680d496
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
|
|
| |
Pulls in security fixes from Chromium 49.0.2623.108
Change-Id: Icb318aa21772eecb4b39f2bbe92ccdc0f020aee3
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
| |
For example, @include, @match, and @exclude.
Change-Id: I16d29b4e72452980e43c8e7c5702e21cccf159d5
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
The mimeType property of the WebEngingeDownloadIten is tagged for REVISION 1
so the related signal should be tagged for REVISION 1 as well.
Change-Id: I918de55c81779424011e42c358d5c7aeed2909ee
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes access of label of MediaStream tracks
Also fixes the logged error:
WebContentsDelegate::CheckMediaAccessPermission: Not supported.
Change-Id: I3fee9ccd9e8b2e5cbd6b707336cc61425a44ba31
Task-number: QTBUG-52216
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
We shouldn't special case some platform integration names, instead
check if it supports the proper capabilities
Change-Id: Ib8a28a974a4054cea4fbfb503be7c0e0f797e242
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I644475dc966b9a0a13bba242b653d2bc2357f24b
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I9e1acaa0dbd31978730dae33ebb517f467f8df49
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I1bd0c58a0cee3933370238841c12cf0c1d965c93
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@theqtcompany.com>
|
|
|
|
|
|
| |
Change-Id: I40706d93348c868f7572fa4c0485f0babf98a06e
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The qputenv() call inside gl_surface_qt.cpp, which is executed on a
GpuChildThread, can reallocate the process environment structure,
and it is possible that at the same time the main thread calls getenv,
which will dereference a pointer to the freed environment structure,
essentially causing a use-after-free crash.
Make sure the qputenv() call happens before Chromium initialization
starts, so no thread-race can occur.
Change-Id: I4ecbdc8bf2abbe45f7d6c5d2633dc9fe27f51e66
Task-number: QTBUG-52124
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
(cherry picked from commit 76c61aa1400ef2def204c3732e30e08e40631e8d)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Compiling against NSS 3.23 fails with current Chromium. Also, with NSS
3.21 there are failures connecting to e.g. google.com.
Fix this by adapting the setup endorsed by upstream Chromium: BoringSSL
is always used for cryptography, and NSS only for certificate handlng.
Patches included in 3rdparty update:
0a385bb [backport] Call EnsureNSSHttpIOInit in the chimera build.
0472123 Fix build against newer NSS
90c62c4 <third_party/libpng> [Backport] update to libpng 1.2.56
34857b8 <third_party/libpng> [Backport] Stop large iCCP chunks causing delays and "Aw Snap!"
Task-number: QTBUG-52193
Task-number: QTBUG-51890
Task-number: QTBUG-52068
Change-Id: If8aaed9b9a09475c5ed0dfec64d31f45ce9670f5
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Chromium usually clears the pending entry when it fails,
so that an arbitrary URL is not left visible above a committed page.
However, it does preserve the pending entry in some cases,
such as on the initial navigation of an unmodified blank tab.
So in other cases Chromium clears the pending entry
(if the navigation is not in progress), and it does force the UI (URL bar)
to refresh.
In our case we allow failed URLs to stick around in the URL bar
but only when alternate error pages are used.
Change-Id: I26e81e7a99415aa184880dcb647a06d977584119
Task-number: QTBUG-48995
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
| |
Task-number: QTBUG-52104
Change-Id: Icf8b4eea7fc52498b2ce9a2c04da6a24e8bdb070
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
moc.prf contains magic (translation: weird heuristics) to determine if
the moc command line would exceed the command line length limit on
Windows. If such an excess is detected a response file is created in
MOC_DIR. As our build directory is not qmake's default one, moc calls
will not find the response file.
Set MOC_DIR accordingly to generate the reponse file in the right
location.
Task-number: QTBUG-51847
Change-Id: Id0695451a986cd9514a7c965c6d33c5c3ef800cd
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Chromium insists on getting the command line via the system call
GetCommandLineW. QCoreApplication already did that, and we want to pass
the already split argument vector to init CommandLine.
This way, the user can pass custom arguments to QGuiApplication, and
Chromium will pick those up correctly.
Task-number: QTBUG-51971
Change-Id: I85b2071fcd8e5781b13ec8d22dc00b1a0c71601b
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
This test is failing on 32 bit Windows. Unfortunately, we couldn't
reproduce the failure locally yet. Blacklist the test for now.
Change-Id: I7bdbc52033ccc0ed16ccf934418cc56a7783f88b
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently if a < 3.2 OpenGL Compatibility profile is requested on OSX,
a webengine application would crash saying that the global profile
does not match the default profile. That happens because in the Cocoa
QPA any requested OpenGL Compatibility profile or Core profile with
version smaller than 3.2 gets reset to QSurfaceFormat::NoProfile and
version 2.1.
Fix consists in making sure that the QSurfaceFormat check only
considers Core profile with versions >= 3.2. All other combinations
would result in NoProfile 2.1 and thus not cause any issues for
webengine.
Change-Id: I7c9866d761c052e52389022abe8e213d062db41f
Task-number: QTBUG-51058
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I93f68098718d7f0303accabc9def49ff83e65ca2
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: Ic99ce79856fdc9a528b78f518c27013579bf38cf
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently on OSX, two Info.plist files are generated and copied to the
QtWebEngineCore framework, one into Contents/Info.plist and another
into Resources/Info.plist. The first one is generated by
core_headers.pro and the second by core_module.pro.
OS X frameworks only use the plist found in the Resources directory,
and the Contents one was generated by qmake erroneously, assuming
that core_headers is an application, rather than a framework.
When macdeployqt is used with -always-overwrite option, it copies
Contents/Info.plist (which contained an incorrect bundle id) over to
Resources/Info.plist, and afterwards when an application is executed,
it cannot find the plist file because it contains the the wrong
CFBundleIdentifier, thus not being able to find the QtWebEngineProcess.
Fix consists in making sure code_headers.pro does not generate an
Info.plist, and just use the one generated by core_module.pro.
Change-Id: I6295bde3b1d1771ed45f3a2fdff9fc60577f2ab6
Task-number: QTBUG-51939
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
Subjects each *.png file that matched grep -law "sRGB" to:
pngcrush -ow -brute -rem allb -reduce -force
Change-Id: I0fe8ea5a1c89d5c9c0dce765607e55ae9e9cf9fd
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Remove misleading "singleton, if one has been set" from the accessor's
documentation. Also link to the accessor's from the class documentation,
and remove mentioning of cookie store as something that can be accessed
for an individual page.
Finally, mark QWebEngineProfile::cookieStore as new in Qt 5.6.
Change-Id: Ia20ca0ef45a9a15de0052f7ceb7f59d454c70fdc
Reviewed-by: Florian Bruhin <qt-project.org@the-compiler.org>
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I877bbd8bc5c710370f135a27bcd7f0f7c95a7292
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
Describe the "loading" property first and remove the
"LoadStatus" enumeration.
Change-Id: I2357c604197c305e9cd2d3ff7dd322949b1b9e4b
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
| |
Use CONFIG+=relative_qt_rpath to make sure the web engine process
contains a relative rpath instead of an absolute path rpath value.
Change-Id: I5f3c9fb93273c41bed795aeba112f260382d2bf8
Task-number: QTBUG-50155
Reviewed-by: Jake Petroules <jake.petroules@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
|
|
|
|
|
|
|
| |
For executables, QT and QT_PRIVATE are the same.
Change-Id: I8f8ef29aea6f4ebb3d7b6cd2f3bc6a3cbf83b0e4
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When a QWebEngine link is clicked on, and as a result a modal dialog is
opened, a QEvent::TouchCancel without any touch points is forwarded
to Chromium, which tries to access the first touch point
position, and causes a QList assertion.
Fix consists of two parts:
1) Make sure that no TouchCancel is forwarded, in case if no TouchBegin
or TouchUpdate was issued beforehand.
2) Because QEvent::TouchCancel events might contain an empty touch
point list, and Chromium expects at least one point, make sure
to forward the last saved touch points (saved in previous TouchUpdate)
together with the TouchCancel.
Task-number: QTBUG-48661
Change-Id: I1eeb2980417b1b04e8387dc9f82f935ef2bd8f00
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
In the WebEngineView type and QWebEnginePage and
QWebEngineView docs.
Change-Id: Ibbabcf6cc38a48fb14abb105f5fda4c03a89ccba
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I0050a0090e4e0877dc56b8c529d9cbb77a0cdd33
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
When QtWebEngineProcess bundle is created, the generated Info.plist
CFBundleIdentifier contains the default dummy prefix added by qmake,
namely: "com.yourcompany.QtWebEngineProcess".
Make sure it uses the proper "org.qt-project.Qt" prefix.
Change-Id: I8e060aeeeb8a53980139e790d3e3947752ffa1d3
Task-number: QTBUG-51942
Reviewed-by: Michael Brüning <michael.bruning@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
The Qt::AA_ShareOpenGLContexts attribute can be used to set up
an OpenGL Context, instead of passing it in the constructor.
Task-number: QTBUG-51379
Change-Id: I46bcb07fc69c0b8e7ac9bf632d328c15dca2ea45
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
|