Age | Commit message (Collapse) | Author |
|
This change moves the bundled library dependencies
from being defined on each module to the plugin definition.
Additionally this change introduces the same check for libraries
than the one we have for modules: If a module needs a library
dependency, then plugin layout is used to verify that the dependency
is satisfied in production.
Test: Added check at build time
Bug: 179064916
Change-Id: I93ecc0691138275b10a7062455c0586bfdec0972
|
|
Test: existing
Bug: 156497102
Change-Id: I99865482f4080546f2c07d83f4fe19b93dd5dba9
|
|
Test: existing
Bug: 156497102
Change-Id: Iaf4aef8660033bf586b04e4709da285d034a1b56
|
|
We are now getting these dependencies from the
prebuilt platform.
Test: existing
Bug: N/A
Change-Id: I5b34558dc0708d4204975ae22b747c32592258b2
|
|
Test: existing
Bug: N/A
Change-Id: I86c4a70ca70879854c152e7d12910a749deed527
|
|
Test: N/A
Bug: N/A
Change-Id: I1f2adc64eebc127c409c9c268d175ede2e9771c3
|
|
Test: N/A
Bug: N/A
Change-Id: I2289165336dd68f8e8d572ef157f1441562c5f02
|
|
These tests were never run; a first attempt of running them from the
IDE revealed a missing dependency and a plethora of failures.
Change I9f363e11c65 fixed ToolkitStub compilability against 11, but
unfortunately the Ant build scripts still use JDK 8 so they broke after
the update.
With this change we remove the tests from the project model and keep
backwards compatibility.
Bug: 163474252
Test: N/A
Change-Id: I3b17a4e0073f8a977cd9b8d2f48e005a346ec8d3
|
|
This reverts commit aebb54ddf6158722ce939f014fed9db54bdc0fc6.
Bug: 163474252
Test: N/A
Change-Id: Ib79d86d762b9ae4d369d157672f75bf42a908118
|
|
We're running UI tests on JDK 11, and ToolkitStub has been adapted to
the new API.
Bug: 135535322
Test: N/A
Change-Id: I699fa53773dee46d79ff1bb19346dbb2994a14ef
|
|
This reverts commit 85f6b523afe1361cadf65e405844668dee5a6604.
Reason for revert: breaks studio_rbe
Change-Id: I47de78fccf56f33c50e43e7d630ef3308d0253b8
|
|
Bug: N/A
Test: compilation in IDE completes
Change-Id: I3c1744d9506503b3c19bdf4608d294bb0b8bcf25
|
|
Bug: 135535322
Test: N/A
Change-Id: I9f363e11c656d0b3fc9239048481f02a8fc48d0a
|
|
These CLs add the libraries that point to prebuilts, and
add the right dependencies to all modules.
Plugins should not be part of the SDK
Test: N/A
Bug: 156497102
Change-Id: I037959e05030d1adeab5cb23eeba804a96b08c02
|
|
Test: existing
Bug: 156497102
Change-Id: Idd54c5c70405045e7d9ccd66ecb3a592bfbcc32d
|
|
Test: N/A
Bug: 156497102
Change-Id: I0c2b3e6f25cd72ce711f6d33c3aa56cb6f208b4e
|
|
These are the modules for the tools/adt/idea project.
They are all tagged as manual for now, see the README in
tools/adt/idea/studio for more information
Test: The new project creation will have its own tests
Bug: 156497102
Change-Id: I422acb6e0c974ee8e43bf0025c136bdaebdd0520
|
|
When building the release bundle with UI tests,
LayoutBuilder.getLibraryName fails while processing the module libraries
in fest-swing, as they have multiple roots. This change names the
library so LayoutBuilder does not have to attempt to infer it from the
names of the jars therein.
Bug: 109812409
Change-Id: I8c471cecf5ecdbcaa612e2aba5f44637d5002fc7
|
|
com.intellij.ui.tree.ui.DispatchThreadValidator does not allow
background threads to handle Swing components, so move the getRowCount()
to EDT instead.
Bug: 138841377
Test: LayoutInspectorTest
Change-Id: I925d0b0fc64d239246837ded0f8335a67e8d3781
|
|
Test code elsewhere would like to extend fixture methods in order
to insert calls to waitForIdle(). Methods on JTextComponentFixture
are fine, but the AbstractComponentFixture for some reason defines
its methods as final.
This CL makes the minimal change for ag/7088761 to work, but if this
CL is right in principle then many more of the methods should probably
have their `final' modifier removed
Test: extension of test framework, used in tests
Bug: 127954783
Change-Id: I7ef82e19b523a944f001d19fb4a65118b69c650c
|
|
Bug: n/a
Test: n/a
Change-Id: I77ff177d0a404b34eb1e8e65b5e1100a12eef982
|
|
… following Change I53c4cf17.
Change-Id: I9bca5939102bb06109d9d8574a99e6ba5eb54df9
|
|
...in the interest of simplifying and consolidating robot text entry.
Change-Id: Ia8b79f66403acb3909b602911d14d3ba2c5dbf68
|
|
|
|
Entering long strings by typing can be slow, usually for no gain.
This change introduces support for pasting text, by setting the contents
of the system clipboard and then using a keystroke to paste it. This
allows speedy entry of long strings in a UI-driven manner.
Bug: 74814367
Test: UI tests
Change-Id: Ifef6f2e27e27a01bd847ee6b7402a1fb01d2c3a5
|
|
Change-Id: I8adb4921071c6046b6f17bd492eae856ec316af5
|
|
… following Change I760e17e6.
Change-Id: I94e49236b4812560b49522d8faf67fa524ac67b6
|
|
The previous implementation was clicking on (0, height/2) which could
lead to clicking outside the cell. To avoid this situation, the click
location has been slightly offset to the right by 5 pixels.
Test: existing test are passing
Change-Id: Ifc1a34a7e18825cd3e5f9b059ddd53f4c5fdecd6
|
|
This reverts commit 9704fe6dc40289ba4cc8fe7e59b9f0f53d13041a.
The CL being reverted breaks some tests relying on incomplete waitForIdle()
and unfortunately it does not fully solves the problem.
The proper solution should probably involve SunToolskit.realSync.
Bug: 71560076
Test: n/a
|
|
* there are at least two EventQueue's (processed on the same EDT)
* an event from the second queue is being processed and the processing posts
a new event onto the first queue (for example SwingUtilities.invokeLater()))
* at the moment when the new event is being posted onto the second queue the
waitForIdle() method has already completed a call to the internal
waitForIdle(*firstQueue*)
==>
when the processing of the first event completes waitForIdle exits since it has
successfully completed waiting for both queues, BUT there is an unprocessed message
in the first queue.
ALSO: Remove special case handling for the case of a single EventQueue since we always
have at least two queues and the multi-queue code also works for a single queue.
Bug: 71058346
Test: n/a (Flaky tests failing because of the race condition do not longer fail)
Change-Id: I95a7399942665826ec00542434e6e4cb113c05de
|
|
Test: Existing tests should pass.
Change-Id: I2570b4616ef7d8143c21b245e9c3fdde4e985464
|
|
Currently, ContextMonitor.processEvent can cause many
WindowVisibilityMonitor instances to be added to a given window. This
is generally wasteful, and the worst case can cause window events to
overflow the stack (when AWTEventMulticaster has a recursive implementation).
Test: Locally. Before this change, I can see many
AWTEventMulticaster.windowDeactivated frames (10 out of 12 include
WindowVisibilityMonitor). Including this change there is only one.
Change-Id: I29b042e0d693c999f8d2fb89ffb9458cb3544fd5
|
|
|
|
Test: fix to test framework
Change-Id: I27bd6b1d510d53e5fcddab361c849347e59cacd3
|
|
|
|
Ensures the event is treated before continuing.
Test: FEST tests
Change-Id: Idb5cfee6ea57e0f800568ebcc72f84049a4c2e91
|
|
Clicking in the middle can cause problems if the node component is very
long. Instead, click on the left.
Test: FEST tests
Change-Id: Ic1974010323219267f99a2309bc49dfb2a1823aa
|
|
After each test is run, check that all Frames were closed.
If a frame is found, close it and fail the test
Test: Run all unit tests. All tests modified on this CL were leaking
a Window/Dialog
Change-Id: I38c8f3856598776e1d002ddeea85a728e1a04399
|
|
If the name of the created Dialog/FrameFixture can't be found,
it throws a ComponentLookupException, and the tests that
run after that will dead lock.
1 - All constructors that call robotWithCurrentAwtHierarchy()
have the potential to dead lock, if sometime later during the object
construction there is some exception.
2 - There is an implicit (hidden) dependency between this constructors
and BasicRobot, but there is a chance that we want to create a
specialised version of the robot.
3 - Is very easy to create and instance of BasicRobot.
Test: Run DialogFixture_constructor_withName_Test in "repeat mode"
Change-Id: Ie9f6947e7f46b22e26aeb94966f519b4f1c3e8b7
|
|
JComboBoxDriver@pressAndReleaseKeys was not overwriting the base
class method.
Test: Found by running Bug279_firstCharInJComboBoxMissing_Test
Change-Id: I092342a604ec798abc816b66a92e1f6c0d2828e2
|
|
studio-master-dev
|
|
Need to press on the menu, for its menu item to become visible
Test: this is a test
Change-Id: I6c0d7478e105938ee8a4466cd9a43effb469950a
|
|
Tests have a menu structure of: File > New > Cut, and the test was
expecting to be able to click "Cut" without pressing all the sequence
Test: this is a test
Change-Id: I9762f3adf10e83aa02cd2b67d3f0d000bfaaad5f
|
|
Window.getOwnedWindows() returns windows that it keeps inside a list
of weak references. Windows that were never set visible or are already
disposed should not be accounted for filtering.
Test: Run WindowFilter_ignore_Test in a loop
Change-Id: I6ffd2bd7066a52cf4db76f50acddc3b7913b9b1b
|
|
|
|
Test: Run touched tests
Change-Id: I10460b7e445758fa5334f9ac6b446c2486adc6ad
|
|
Test: Run touched tests
Change-Id: Id653d02fe4dfab3baec845c989fb103cb476915e
|
|
|
|
Tests extending BasicRobot_TestCase and implementing
beforeShowingWindow() are throwing NPE as they need to access window()
for example calling: window().add(button);
This happens because the window() method asserts that the window is
initialized.
Also, tests that fail due to this NPE, are leaving a dead window
behind (window.destroy() not called)
Test: BasicRobot_requireNoJOptionPaneIsShowing_Test,
BasicRobot_focus_Test and BasicRobot_rotateMouseWheelTest
Change-Id: I7c08a70c2803e89ca7f5a11dc26065caedc4c548
|
|
Added a new Robot@enterText method that sends typing events directly
to the Java event queue, without using the Java Robot.
This new method doesn't need to know about keyboard key-codes and
should be more reliable with all Operating Systems.
Test: Mac, any UI test that types
Change-Id: Ib029af8b5801b08f663ac68aba0ac5730a2c3d48
|