This makes it the same as default browser settings (for now). Otherwise
the test may crash due to GPU OOM.
This is to be a workaround only, and should be reverted after the bug is
fixed.
Bug: 5382134
Change-Id: I82a8dd83b69c8fceb657af9e32a2c5fa66ead2ce
Some supurious calls to these functions when tests are still
running are observed. Add more logging to see what's the
caller.
Change-Id: I9254c3639bcb8619be39bfbc6981fae5bf5e7efa
there are cases where webkit dump timed out, but a RuntimeException
was not thrown (as it should have) after the timeout limit was
reached. It's probably because the Looper message to invoke time
out is obtained after the message to dump webview. This change
adjusts the order of obtaining the message so hopefully it can make
the timeout work for real.
Change-Id: I4197699f5aad765c88dc81123d052e4c49d7c53d
This avoids problems with LayoutTests fast/frames/lots-of-iframes.html and fast/frames/lots-of-objects.html
Bug: 3440578
Change-Id: Ic045141f986baf779ffc9eec804ba994ce0d0fbe
1. Remove mDeferMultiTouch in WebView, it was for testing only but
we don't need it now since we always pass MultiTouch to WebKit.
2. Remove the use of mDeferMultiTouch in DRT tests.
3. Correct the index of getX(Y) for the second touch point in the
debug string of MotionEvent.
Change-Id: Ib63cfc5935af1a169ed26b2b138f74908492bc18
current timeout mechanism does a text dump and then move to next case.
sometimes the text dump causes WebView to spin, and doing a
stopLoading seems to resolve some of the cases.
Change-Id: Iae08f9c93b2da3e805cacae43f80083fdf1bc5e1
if a webview dump times out, normally it's already in a bad state,
throwing runtime execption will generate a crash on the test case
and make the test move forward.
Change-Id: Id7430700db6b6d90160d60d23debd250b1731789
Needs https://android-git.corp.google.com/g/83647 in external/webkit,
which disallows such cookies by default and provides configuration
methods native side.
Also updates TestShellActivity to allow such cookies, as they are
required for the page cycler.
Bug: 2944150
Change-Id: Idcc7e2812fab33e6a2a8766a2176bc3278e617da
This is required for tests that test PageCache behaviour, eg
- fast/dom/DeviceMotion/no-page-cache.html
- fast/dom/DeviceOrientation/no-page-cache.html
- fast/dom/Geolocation/no-page-cache.html
Change-Id: Ie5862f6c86718142c949be81a85358ca6fac628a
Since the UI uses multi-touch for zooming, the event is passed to webkit
only when the webpage doesn't allow zooming.
1. Updated the data structure TouchEventData in WebViewCore.java
to take multiple points and used it everywhere.
2. Added a passMultiTouchToWebkit() function to do what its name means.
3. Added the multi-point touch support to dumprendertree.
Change-Id: Iafc1aa7b21a587ad26efd2f124b9a66316297ab8
A previous change terminates page load if there's certain JS errors;
the intention was to speed up layout tests such that page accessing
non-existent test controllers/methods will get immediate termination
instead of waiting on timeout. However this causes problem for page
cycler because it may interrupt the test run too early. Also there
was a bug on bracketing on the termination conditions (operator
precendence issue).
Change-Id: I2f19e48fa0061286fddf0f7cbb4953d7f7d88f76
Remove the overridden WebChromeClient.onProgressChanged function. We don't seem to need it and
it appears to be the cause of b/2681943. Without that code we need to tweak the terminating
conditions in onPageFinished and notifyDone.
Change-Id: If2958007d92f2da92ae0de19eb07a0b38258fae0
- Make the layout test counter work when you run tests through the GUI
- Use the progress bar in the applications title bar to display test progress
Change-Id: I3d4b778470e812b80b7a64297b3a64ba6f9d083c
This fixes a bug in commit https://android-git.corp.google.com/g/#change,52518.
Updated index to be zero based when passed around and off by one error on
resume. Note that previous commit changes how DumpRenderTree dumps titles.
This might affect the results of layout tests.
Change-Id: I3d6989d71c336f90168e38c994dd36743bda365c
* reduce timeout limit from 30s to 15s
* terminate a test case under some condition on uncaught JS exception
* minor fixes
Change-Id: Iabc8f214544d2c8c14139756abc049870023fea5
in a consistent, known state. This fixes a bug for example with the appcache
tests, where disable.html turns off appcache support but it is never restored
causing future tests to fail.
Change-Id: I71e2607d29ec6358cfdc6ab37a6861c161e84cd8
Fixes DumpRenderTree to re-use HTTP authentication credentials
This is required for layout test http/tests/appcache/auth.html
Bug: 2098423
Change-Id: Ic9531e3c23a2fa9ebfab70cde3172550f572a404
This change provides the infrastructure for
layoutTestController.overridePreference(). Currently, we only provide an
implementation for the preference 'WebKitOfflineWebApplicationCacheEnabled',
which is required by the layout test http/tests/appcache/disabled.html.
Change-Id: I8552f2f4e23b982db2d067ffa20c052e56d8fb7f
As part of this, make it possible for DRT to configure the timeout threshold between sending touch events to WebCore as the Layout Tests only synthesize single events, not a stream. Because of this, they often get dropped by the WebView for coming too quickly.
Skip the multi touch test as we don't support multi touch in the Browser.
Change-Id: I7b9830f43181fea33206825b49ef2e294269b4dd
Also some logs seem to suggest that in rare occations, a webkit dump after timeout has occured actually never hears any response back from webkit, thus causing the test app to wait forever, because the request webkit data mechanism is not protected by a timeout watchdog. So adding a explicit log print before calling webkit dump for monitoring in future test runs.
