| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
| |
Otherwise unchanged intercepted request leads to second call
in the same interceptor but on ui thread after io thread.
Ammends a05bb73747.
Fixes: QTBUG-86267
Change-Id: I4e7c662d24a58be5445f5c8b6d0bf3751f40cc05
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
| |
Do not include certificate error qml test if no ssl.
This change does some copy-paste but this will
be handled in qt6.
Change-Id: I8cc6d37074d78ca9f55333f479fb410ef927385d
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
| |
Binary compatibility files added.
Change-Id: I71c208cf9cb5a79e410416ae769ec48c23a219cc
Reviewed-by: Jani Heikkinen <jani.heikkinen@qt.io>
|
|
|
|
|
|
|
|
|
| |
Makes it more stable.
Change-Id: I2b479330e66e9d87355b09ebb0f4c4064ef96963
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Tamas Zakor <ztamas@inf.u-szeged.hu>
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The function is called for all navigation requests. The original bug
report was most likely due to the use relative links on a page without
a base url, which indeed results in no call to acceptNavigationRequest
being made. But this has nothing to do with local vs remote URLs, it's
rather that relative links, which are relative to nothing, are not
valid links and there thus cannot be any navigation request either.
Task-number: QTBUG-48435
Change-Id: I08bd0c86d67bf1dd1d7662468321777254a7db0b
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
| |
Submodule src/3rdparty c91f4d20..809e16e4:
> Add cookie filter to URL requests cookie headers
> Use audio-manager app-name for MPRIS
Fixes: QTBUG-85526
Change-Id: Icab26cad3cea8e2ee021a3e589f41bf0543d64fa
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
|
|
|
|
|
|
|
|
| |
This preserves unicode characters in the environment variable on
Windows.
Fixes: QTBUG-85054
Change-Id: I3352cc3db4ce8d7b9b352c1634e3239954780618
Reviewed-by: Kirill Burtsev <kirill.burtsev@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Parts of Chromium depends on it for temporary directories it creates.
This also changes the logic of depending on it being empty as indicating
an in memory profile, so it required a bit of cleaning up.
Note this does subtly change the return value of off-the-record profiles.
Task-number: QTBUG-62957
Change-Id: Iec6e24556128c515fce27803171a766b8a9d92e3
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
|
|
|
|
|
| |
Task-number: QTBUG-84011
Change-Id: Ie1eef3bc850e99437bb02bea01e06a49029e0dc4
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
| |
Change-Id: Idee9e3742ea3cb949f786d0e55a2e2f8530c7b1e
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
| |
Use current configure instead of qtHaveModule().
Change-Id: Ifc4500ff5eee4b997a0c02fb387d8918f88d06d1
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
| |
It is necessary after
e8284237 Enable most qmltests without testsupport
Change-Id: Ib6668b852588b0a2afa911144a550b722db7f918
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
| |
Leftover from
cffba828 Test: port tests to QRegularExpression
Task-number: QTBUG-72596
Change-Id: I4009bc00f24b773539a674ad5f2ba0aad1cef04b
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
|
|
|
|
|
| |
Change-Id: I5609440666d41ea273f3be44fd860195750af4dd
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
| |
Change-Id: Ic26d982ef9695705a9f7c757b7d983824b31a070
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
| |
The test has been removed by:
cfef4615 Delete QtWebKit tests from tst_qwebenginepage.cpp
Pick-to: 5.15
Change-Id: I677772f00c99ae3b6e33a2d61c7135d4306d3674
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
(cherry picked from commit 0891ac9c05b96471a0858011ef4f84b353337d16)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
CloseContents() triggers a windowCloseRequested() signal. The handler of
the signal might close the QWebEngineView and it would result the
destruction of the WebContentsDelegateQt. Thus, any operation on
WebContentsDelegateQt is not safe after a
WebContentsAdapterClient::close() call.
For example, calling QWebEngineView::close() on
QWebEnginePage::windowCloseRequested() would result crash because it
sets life cycle state to discarded which destructs web contents.
Fixes: QTBUG-84355
Change-Id: I5da404bc9b1923cc19085ee899b550da49d1416b
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In some cases QUrl::toLocalFile() does wrong path conversion on
Windows. Therefore we have to convert the file URL to absolute path
in another way and use FilePath::IsAbsolute() for validating that.
Also stabilize WebEngineViewSingleFileUpload tests. They didn't wait
for html title change when the expected value was the same as the
default title.
Change-Id: Ica7798a299f9b28657afeeccccfba3fdecc515e2
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since PermissionManagerQt functions independent of webcontents initialization,
permission installment may be done earlier (right after page base constructor).
Practically, this allows to grant/deny permission before first ever navigation
to avoid permission feature request from well-behaving web application.
Unfortunately, this only applies to Web Notifications: there is a way
to check without request: Notification.permission javascript static object.
Other supported features lack similar mechanism: they operate through
success/failure callbacks, which unconditionally invoke permission requests,
and Page/View API doesn't automatically answer with remembered permissions.
Task-number: QTBUG-83476
Change-Id: I63a3cbca25498d5026975073e125f2ab4f9ab2ad
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
| |
Fixes: QTBUG-83732
Change-Id: Ie07a1ecf87b214436ab7cea07003ddd5aeaedda3
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
| |
Make it easier to get right by normalizing the origin and test it.
Task-number: QTBUG-83476
Change-Id: Ie4562e937cbcca5a1cf24e00d340a3ac2e9e2210
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
Reviewed-by: Kirill Burtsev <kirill.burtsev@qt.io>
|
|
|
|
|
|
| |
Task-number: QTBUG-58121
Change-Id: I9953eaec9d4dd971a7f7fb3ed9a156bccffa05d5
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|\
| |
| |
| | |
Change-Id: I916838caf8c981d5dac876631a6e510ed2ffbe0e
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Instead of using QSharedPointer's reference count to communicate
adoption/non-adoption, change adoptNewWindow to return a adapter pointer, with
null meaning non-adoption. Then change QWebEnginePage's implementation to reuse
already existing adapters if possible, restoring previous behavior of
OpenURLFromTab when createWindow returns this.
Task-number: QTBUG-80596
Change-Id: I8ee7c31e4294aabd3207c504cba67d6171c66cb0
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
For a QWebEnginePage subclass where createWindow returns the this-pointer, we
get a crash in WebContentsDelegateQt::AddNewContents because AddNewContents
expects the adapter to be adopted already, but QWebEnginePage delays adoption.
Revert 3855015600 by getting rid of the delayed adoption code path. It's not
needed anymore with the delayed initialization of the WebContentsAdapter.
Revert 8a4091c210 by forcing adoption of the adapter even when it doesn't have
contents. This no longer results in a crash since OpenURLFromTab ensures that
the adapter is initialized before use. However, it does result in a behavior
change since return-this now consistently overrides the adapter, so, e.g.
navigation history is now always cleared whereas previously it was only cleared
by the AddNewContents code path. Fixed with new approach in next patch.
Fixes: QTBUG-80596
Change-Id: I4d2230c1bffcf2d77fa59ded9be51da49a820474
Reviewed-by: Kirill Burtsev <kirill.burtsev@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Resetting the network context can cause some (expected) disconnects, so return
value of HttpServer::stop should be ignored.
Fixes: QTBUG-83670
Change-Id: Ieb17b38c422f0cdaea5423d082fd04c22715cebe
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|/
|
|
|
| |
Change-Id: Ie02da939a37575dd2b0a43440978748fd79874e4
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We use now network service avoid io-ui-io-ui hops,
pipe proxying url loader factory directly to ui thread.
This solves thread safty issues.
Add deprecated request interceptor test cases.
Task-number: QTBUG-83082
Task-number: QTBUG-82999
Change-Id: I38778cf1a70789c5e92e04c93d1c93e2cc4c765a
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Interrupt the waiting on the event loop in CallbackSpy::waitForResult()
to not block calls on the UI thread.
It is necessary because RenderProcessHost creates channel for mojo
communication on the UI thread. QWebEnginePage::runJavaScript() needs
this channel because the JavaScript is executed in the render process.
If evaluateJavaScriptSync() is called before the mentioned channel is created,
the JavaScriptExecuteRequest mojo message might not be sent because the
wait would block the thread.
Change-Id: Ic5bb5a6fde02717cec49dcf9e458f2eaac09eacf
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
| |
As the dependency 6420ad91d3 in declarative for the only present
focusChild is integrated and test doesn't fail according to grafana.
Change-Id: I0d8fbb966548687d3fb92ec123726f110f18d8d5
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The normal WebViewImpl::SetBaseBackgroundColor expects to be called only during
certain lifecycle states, otherwise a DCHECK will be triggered. Whereas
the *Override version forces a lifecycle update.
Needs 3rdparty change to stop RenderWidgetHostViewBase::SetBaseBackgroundColor
from clearing the override color when switching between opaque and transparent
colors though.
Change-Id: If4c1e22886d7ebc7d87ca880dd9c6a659855a931
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
|
|
|
|
|
|
|
|
| |
Move the qmltests that require testsupport to qmltests2, this enables
us to run most of the tests on CI configurations that isn't developer-
builds.
Change-Id: Ie0ba060632b36cbd738d6ede512a6a5f35880ae3
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
|
|
|
|
| |
Change-Id: Icaf68648cbc2c13a61fc3b208ff1e64ca36c90b6
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
|
|
|
|
|
| |
Fixes: QTBUG-82953
Change-Id: I7dfa916ea0ada00ce84a1e9e49de9eea1800a108
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
After tedious investigation the issue boils down to error prone
synchronization of web engine settings. WebEngineSettings are
synchronized between the browser process and the render process.
Moreover in the browser process the sync message is send to the render
with QTimer::singleShot, which can cause race conditions if
for example QWebPage::setUrl was used meanwhile. This makes
current settings not being picked up by the render process and results
in 'titleUpdate' test case flaky. This happens due to the fact that
ShouldDisplayErrorPageForFailedLoad in the render process frame view
could have invalid value.
Try to sync web engine settings on every adapter load, setContent or
reload.
Mark some flaky settings in tests.
Fixes: QTBUG-83078
Change-Id: I5289472f146e104d5cb6c3b9b20b26d3dc42f4b1
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|\ |
|
| |\
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Conflicts:
src/3rdparty
tests/auto/widgets/qwebenginepage/tst_qwebenginepage.cpp
Change-Id: Idbc9d3b06c0052a721c974fbcbfa3164faaead14
|
| | |\
| | | |
| | | |
| | | | |
Change-Id: I5f7aab6031f1c1ca289d5ba408d408684849031c
|
| | | |\
| | | | |
| | | | |
| | | | | |
Change-Id: I50afabc6022fea5b3db2a7dcb8659b654573661f
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The RenderViewObserverQt_SetBackgroundColor message was sent with wrong
Routing ID.
This fix also cleans up RenderWidgetHostViewQt::UpdateBackgroundColor()
method.
Fixes: QTBUG-81781
Change-Id: Ida198fb061715d389859ace17e1f773db491c51d
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
| | | |/
| | |/|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This reverts commit d4022e03ccaeb92e41075f276e4011bd49627165
and partially fb430b4e104dd6313a776980b4798f1333193149.
Reason for revert: initial blacklisting was done because of wrong assumption for test being flaky. But this is not the case. There is a non-zero chance that invalid build for macOS will be generated due to some undiscovered issue in build configuration or CI. Produced binaries are invalid and can't be used. If this first testcase fails then the whole suite will fail for every test. Blacklisting only moves failure further: first time from contextMenuRequested to javaScriptDialogRequested and etc, and second time to next test in whole suite tst_InspectorServer::testPageList. This creates a lot of really invalid failure entries in grafana testresults aggregator. It also creates confusion for the reason behind these kind of test failures and invalid bug report for other test failures.
Task-number: QTBUG-76549
Change-Id: I1dfe850f6a9cabec352c6e2bd5471e7c4f2e99ca
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|/ / /
| | |
| | |
| | |
| | | |
Change-Id: I80c99deb95c06191c4e872e45c56dff0c6c26001
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Avoid having the enum name in the enum values.
Change-Id: Ida5b3405e6922eb1a2a37edc6fb650019254f4d4
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
|
| | |
| | |
| | |
| | |
| | | |
Change-Id: Ic4b70d31c85a31a434644c311eb173cda6a82fbd
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |
| | |
| | |
| | |
| | | |
Change-Id: I23bd548db07c6332170e9d304977e5974c7f049f
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|\ \ \ |
|
| |\| |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Conflicts:
tests/auto/quick/qquickwebengineview/BLACKLIST
Change-Id: I365daa73d2eb422ac916c32b86c5ad0d3c32086d
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Change-Id: I847a1750ce5c9533db43fb60f91b9739c544791a
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |/
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Toggling page's setting is actually first batched and then executed
asynchronously by timer. So javascript code might not necessarily see
this update immediately after change on UI thread.
Change-Id: I1a1f373b5fd0b96c5b937a2dca1ce0ed99364c33
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The replacement of error with networkError was reverted.
Change-Id: Iae618b1e74ac7f67b6c02a809b3177b2183b5783
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
|