Ewan Le Bideau-Canévet [Tue, 13 May 2014 08:46:43 +0000 (10:46 +0200)]
wayland-fits initial packaging
Change-Id: I7056a167bae30611e1c1859d6b8b4b117c4e0c69
Signed-off-by: Ewan Le Bideau-Canévet <ewan.lebideau-canevet@open.eurogiciel.org>
U. Artie Eoff [Tue, 16 Sep 2014 20:03:15 +0000 (13:03 -0700)]
weston-wfits: check output->make equal weston-X11 too
Weston changed the output->make for the x11-compositor from
xwayland to weston-X11:
http://cgit.freedesktop.org/wayland/weston/commit/?id=
22e16b656f67ee577e32a7af91f443840fc3aee1
For uinput emulation on x11 backend, we need to check whether
output->make is either weston-X11 or xwayland (for backwards
compat) now.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 9 Sep 2014 17:55:17 +0000 (10:55 -0700)]
README: contributing
Use github's pull requests for contributing
U. Artie Eoff [Thu, 21 Aug 2014 20:05:58 +0000 (13:05 -0700)]
test/core: disable wl_display bind and data tests
wl_display is no longer exposed as a global since v1.6 so disable it.
See http://cgit.freedesktop.org/wayland/wayland/commit/?id=
78d8047407
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 15 Jul 2014 17:32:32 +0000 (10:32 -0700)]
efl/ListUserMouseScrollTest: move mouse the length of surface
With some elm profiles, the list items are taller/wider. The list
only scrolls the distance of the pointer relative to the list item.
With larger sized list items, the previous absolute hard-coded delta
only generated small list movements resulting in the test taking
longer to scroll the list to its edges; ultimately the test timeout
wins before the test completes and the test fails.
Now, move the pointer over the entire length/height of the window
to ensure larger list scrolls.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 15 Jul 2014 16:41:50 +0000 (09:41 -0700)]
efl/ListUserMouseScrollTest: don't let the list bounce at edges
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 4 Jun 2014 23:01:54 +0000 (16:01 -0700)]
efl/test_actionslider: compat for elm 1.10.99 (1.11)
Actionslider data interface changes once again.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 4 Jun 2014 21:21:17 +0000 (14:21 -0700)]
weston-wfits-input: sync wfits pointer with the global pointer
Every time a client binds to the wfits input interface, attempt to
synchronize the wfits pointer position (particularly for the uinput
case) with Weston's internal pointer position.
Fixes #7
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Thu, 1 May 2014 19:37:00 +0000 (12:37 -0700)]
testmain: add xml-wrap-results option
When --xml option is used, allow the --xml-wrap-results
option specify an additional test suite path to prepend/wrap
the test results output.
Normally, the xml output is like:
<TestLog>
<TestSuite name="Wayland Functional Integration Test Suite">
<TestSuite name="core">
...
With --xml-wrap-results specified, say as "my/custom/path", then
the xml output would look like:
<TestLog>
<TestSuite name="Wayland Functional Integration Test Suite">
<TestSuite name="my">
<TestSuite name="custom">
<TestSuite name="path">
<TestSuite name="core">
...
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Mon, 28 Apr 2014 15:18:05 +0000 (08:18 -0700)]
test/efl: use the correct efl version macro
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Mar 2014 21:21:18 +0000 (14:21 -0700)]
test/efl: make actionslider test compatible with 1.9.99
Getting access to the underlying drag_button_base data
has changed in Elementary 1.9.99 (i.e. 1.10).
https://git.enlightenment.org/core/elementary.git/commit/?id=
2a52eec7f
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 12 Mar 2014 19:58:55 +0000 (12:58 -0700)]
harness: move pointer to origin before each test begins
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 12 Mar 2014 19:14:09 +0000 (12:14 -0700)]
Make efl tests work with g-main-loop enabled efl
This is a major refactor of the efl test framework
to get the efl tests to work with a g-main-loop
enabled efl.
The efl test framework now uses a multi-threaded
model (as opposed to the recursive iterate model) to
drive the tests. That is, the efl main loop executes
in the main thread and the tests execute in a separate
thread. All calls into the efl api's are synchronized
with the main loop by employing the efl threading
synchronization api's.
Fixes #6
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Mar 2014 21:38:05 +0000 (13:38 -0800)]
configure: require boost_thread
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Feb 2014 21:34:35 +0000 (13:34 -0800)]
core: add output wrapper test
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Feb 2014 21:33:00 +0000 (13:33 -0800)]
core/dnd: ensure surfaces aren't overlapped
Overlapped surfaces interfere with the test case and likely
generate a test timeout.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Feb 2014 17:18:44 +0000 (09:18 -0800)]
test/efl: disable tests for unsupported features
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Feb 2014 17:00:35 +0000 (09:00 -0800)]
Fix some more corner-case test failures
Occassionally, the weston shell randomly places client surfaces
in a non-ideal position for some input-based tests. Thus, use
a more determinate, explicit, and known position for client surfaces
before attempting the input testing.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 18 Feb 2014 23:26:56 +0000 (15:26 -0800)]
core: fix MultiResourceTest intermittent failure
The MultiResourceTest would fail occassionally when the test
surfaces got placed near the edges of the compositor output
or when the surfaces were overlapped. Weston's shell places
surfaces in random positions, thus to avoid hitting the above
corner-cases we just position them explicitly with the wfits_manip
test extension.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 18 Feb 2014 23:05:02 +0000 (15:05 -0800)]
Add move_surface wfits test extension.
This adds a new wfits interface called "wfits_manip" which will
allow tests to manipulate certain server-side state. Currently
this interface defines one method to move surfaces to a specific
position on the output. This will allow tests to ensure surfaces
are in deterministic positions so they can avoid various corner-case
surface position selections by the compositor. For example, avoid
two surfaces from being overlapped or avoid surfaces mapped near
the edge of the compositor output geometry; both which can
interfere with certain test situations.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 18 Feb 2014 19:20:41 +0000 (11:20 -0800)]
Only use ecore_app_no_system_modules if ecore >= 1.8
Fixes #5
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Mon, 27 Jan 2014 18:21:44 +0000 (10:21 -0800)]
weston-wfits: use our own Weston SDK version macro
The WESTON_VERSION_AT_LEAST macro logic does not do what one
would expect if taking its name literally. Upstream concensus
is divided about what it should be (i.e. naming vs. logic).
Therefore, we're not even going to use it... we'll define our
own means for supporting multiple Weston SDK versions including
unstable/development branches.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 14 Jan 2014 22:08:55 +0000 (14:08 -0800)]
efl/harness: don't use a static function variable for the client
When the program exits, there seems to be a race between when
wl_display_disconnect is called by ecore_wl_shutdown (i.e. when the
Application instance is destructed) and when the static client instance
is destructed. The client instance destroys some bound interface
objects that require a valid display connection which may be gone
by the time the client instance is destroyed. Hence, resulting
in a segfault.
We still need to create the client instance on first use since
ecore_wl_display_get() might not be available during the harness
instantiation. Thus, just manage the client_ instance as a member
pointer in the ElmTestHarness instance and delete it when the
harness destructs... which will be before the Application instance
is destroyed.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 14 Jan 2014 20:57:20 +0000 (12:57 -0800)]
test/efl/application: don't load efl system modules
We don't want to load any of the efl system modules that
monitor things like power-mgmt, time, etc... since each test
generally executes and finishes very quickly and can result
in erroneous output from those modules, for example:
<ERR> _on_low_battery_get_cb() Message error
org.enlightenment.DBus.Canceled - Canceled by user.
In general, these modules are useless in the context of this
test framework.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 4 Dec 2013 20:23:41 +0000 (12:23 -0800)]
weston-wfits: surface width/height data moved again in weston SDK
The width and height properties are now tracked in the
weston_surface data structure.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Thu, 14 Nov 2013 20:36:31 +0000 (12:36 -0800)]
config: require weston >= 1.3 for WESTON_VERSION_AT_LEAST macro
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Mon, 4 Nov 2013 17:11:03 +0000 (09:11 -0800)]
test/efl: Ecore_Wl_Window struct internals are private in 1.8 now
Use the new ecore_wl_window_surface_get accessor in ECORE 1.8 since
the Ecore_Wl_Window struct internals are private now. Also maintain
backwards compatibility with ECORE 1.7 too.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 23 Oct 2013 18:22:18 +0000 (11:22 -0700)]
weston-wfits.so: support 1.4 weston sdk
Weston surface geometry is now tracked by weston views. Update
the wfits query interface to support the new structure. For now,
only support surfaces with a single view.
The "focus" flag is no longer a member of the weston compositor as
it is not needed anymore. This was originally used to ensure notify
api's always worked when used on x11-backend if the compositor window
didn't have X focus.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 16 Oct 2013 21:55:52 +0000 (14:55 -0700)]
.gitignore: ignore autogenerated test-driver
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 16 Oct 2013 19:28:14 +0000 (12:28 -0700)]
weston-wfits.so: support non-X environments
This also fixes the dependency on libxcb and libX11 symbols that
were previously and implicitly loaded during Weston's runtime.
Weston no longer links directly to X libraries (and appropriately so)
and thus no longer makes the symbols available to the weston-wfits.so
module when it's loaded.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 15 Oct 2013 22:10:12 +0000 (15:10 -0700)]
weston-wfits: support transformed output
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Sat, 5 Oct 2013 15:48:37 +0000 (08:48 -0700)]
don't use wayland-scanner.m4
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
Neil Roberts [Thu, 26 Sep 2013 19:18:34 +0000 (12:18 -0700)]
core: Add a test for multiple pointer and keyboard resources
This adds a test which creates multiple pointer and keyboard resources
for the same client and verifies that they all receive events. It also
tests various combiniations of pointer and keyboard focus and ensures
that for example a keyboard created while the surface already has
focus will correctly get updated about the state.
Neil Roberts [Thu, 26 Sep 2013 19:18:29 +0000 (12:18 -0700)]
core: Add a wrapper for the keyboard
This adds a wrapper for a wl_keyboard in a similar way to the pointer
wrapper. It keeps track of the keys that are pressed so that they can
be quickly verified. wayland-fits now depends on libxkbcommon so that
the keyboard wrapper can pass the keymap to it and get the modifier
indices.
Neil Roberts [Thu, 26 Sep 2013 19:18:22 +0000 (12:18 -0700)]
core/pointer: Keep track of the focus serial
The focus serial is needed to correctly send a cursor so it is useful
to be able to verify that this is working correctly.
U. Artie Eoff [Wed, 25 Sep 2013 16:29:51 +0000 (09:29 -0700)]
README: Describe how to execute in non-forked mode
non-forked mode is useful for gdb debugging.
U. Artie Eoff [Wed, 21 Aug 2013 02:00:39 +0000 (19:00 -0700)]
README: WESTON_WFITS_INPUT_EMULATOR and dependency versions
Describe WESTON_WFITS_INPUT_EMULATOR environment variable.
Update minimum required dependency versions.
U. Artie Eoff [Tue, 13 Aug 2013 19:30:48 +0000 (12:30 -0700)]
configure: don't clobber environment LDFLAGS
configure.ac was explicitly setting LDFLAGS causing the users
environment supplied LDFLAGS to have no effect. This resulted in
some strange linking issues, especially when the developer relies
on the environment LDFLAGS to specify custom library paths.
Append the environment LDFLAGS to the configure.ac LDFLAGS.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 9 Aug 2013 23:20:27 +0000 (16:20 -0700)]
Drop support for Wayland/Weston < 1.2 and update to 1.2 APIs
Dropped usage of wayland server 1.2 deprecated APIs and replaced
with new APIs.
This also drops support for Wayland and Weston SDK < 1.2.
Incremented project version.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 9 Aug 2013 21:31:36 +0000 (14:31 -0700)]
Check for boost libs and cleanup Makefiles and some warnings.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 7 Aug 2013 19:11:47 +0000 (12:11 -0700)]
weston-wfits: workaround to enable deprecated wl api's for now.
compositor.h does "#define WL_HIDE_DEPRECATED" before including
wayland-server.h and therefore doesn't allow for the deprecated
wayland API's to be used. To workaround this, we include
wayland-server.h before including compositor.h for now.
Soon, however, we need to update to the new wayland stable API
to be compliant. Will likely drop support for < 1.2 at the same
time to avoid compile time #ifdef version checks in the weston
extension.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Thu, 1 Aug 2013 20:09:06 +0000 (13:09 -0700)]
efl/test_actionslider: fix undeclared variable
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Thu, 1 Aug 2013 20:06:52 +0000 (13:06 -0700)]
efl/test_actionslider: include the config.h
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Thu, 1 Aug 2013 19:46:46 +0000 (12:46 -0700)]
test/efl: use static_cast for actionslider's smart data
The pragma GCC diagnostic doesn't work properly on gcc >= 4.7.1
Seems to be related to this bug:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55881
Thus, we have to avoid the actionslider macros :-/.
Since Elementary 1.7.x uses a different accessor than 1.8+ to get
to the underlying smart data, added a configure check for the
elementary version (HAVE_ELM_1_7) in order to select which accessor
needs to be used.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 31 Jul 2013 22:45:40 +0000 (15:45 -0700)]
Revert "efl: pragma GCC diagnostic doesn't work properly on gcc >= 4.7.1"
This reverts commit
cee47dd374985a9727fc9e874e25230e7aebf37d.
This change is not compatible with EFL stable 1.7.x release... need
to apply a better solution.
U. Artie Eoff [Wed, 31 Jul 2013 20:36:28 +0000 (13:36 -0700)]
efl: pragma GCC diagnostic doesn't work properly on gcc >= 4.7.1
Seems to be related to this bug:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55881
As a workaround, copied the efl macro body and modified it to apply
a static_cast explicitly.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Mon, 29 Jul 2013 20:01:56 +0000 (13:01 -0700)]
efl: EFL_BETA_API_SUPPORT required to use EO now
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 3 Jul 2013 16:39:56 +0000 (09:39 -0700)]
extensions/weston: Don't delete wl_resource anymore
wl_resource's are now opaque in Wayland 1.2 and lifetime managed by
the wayland server library.
However, for Wayland < 1.2 (1.1.90) we should delete them... but for
now just let it leak until we have time to fix this (i.e. do a
compile time version check) :-/.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 28 Jun 2013 15:23:50 +0000 (08:23 -0700)]
efl: fix actionslider test
Must be explicit, now, about EO usage.
See EFL commit
383592b44402e8552a85b8dc5d152717871d48dd
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Jun 2013 16:54:01 +0000 (09:54 -0700)]
core/test_dnd: test more of the dnd offer states
Add checks for offer positions and offer enter/leave.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Jun 2013 16:51:57 +0000 (09:51 -0700)]
core: notify data target when data device leaves
This helps for testing dnd event flow.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Jun 2013 16:24:09 +0000 (09:24 -0700)]
core/data_device: update offer position on motion
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 19 Jun 2013 16:15:00 +0000 (09:15 -0700)]
core/dnd: Add simple drag and drop test case and wrappers
Added wrapper classes for the wayland dnd interfaces with
sanity assert checks and a simple dnd test case.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 18 Jun 2013 23:05:23 +0000 (16:05 -0700)]
core/pointer: Add button notify signal and use wl_surface type for focus
Add a button notify signal that tests can listen for. This allows tests
to do more things (e.g. start a drag op) when button events occur.
Also, use the wl_surface* type for focus_ instead of Surface* type since
the wl_surface_data_get cannot be guaranteed to be a Surface* type.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 18 Jun 2013 22:48:08 +0000 (15:48 -0700)]
core/harness: don't use local static variable in instance method
The use of a local static variable in a non-static class method is
only instantiated once and used by all instances of that object.
In core::test::Harness::client() we were statically creating the
result (client) with the current instances' display_ member. Hence,
if a second core::test::Harness instance was created using a
different display_, the client() return value for the second instance
would be that of the first instance (i.e. it uses the first instances'
display_). This is not what we'd expect/want as an API consumer.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Mon, 17 Jun 2013 18:52:15 +0000 (11:52 -0700)]
test::Harness: getSurfaceGeometry const correction
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 12 Jun 2013 21:55:32 +0000 (14:55 -0700)]
weston-wfits: use unique timestamp for notify*
Use weston_compositor_get_time() instead of a hard-coded value of
100 for the timestamp in each call to a Weston notify method. This
ensures the probability that each notify event has a unique
timestamp. Some clients may opt to ignore events when it has
already processed a similar event with the same timestamp. Thus,
hardcoding the time value is not a good idea.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 5 Jun 2013 20:09:05 +0000 (13:09 -0700)]
core: explicitly bind to version 1 of wl_seat
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 5 Jun 2013 19:57:23 +0000 (12:57 -0700)]
Revert "core: update Seat to wl_seat_listener version 2"
This reverts commit
2789da9476ec39bbdd5dca1684ce5f2976505357.
Compilation is not backwards compatible with 1.0 and 1.1.
U. Artie Eoff [Wed, 5 Jun 2013 18:25:27 +0000 (11:25 -0700)]
core: update Seat to wl_seat_listener version 2
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 5 Jun 2013 03:01:33 +0000 (20:01 -0700)]
efl: test deselection in DayselectorUserTest, too
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 5 Jun 2013 02:43:05 +0000 (19:43 -0700)]
efl: streamline DayselectorUserTest
Move the SUNDAY special case into the loop by using the modulo
operator to choose correct day selection and label.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Jun 2013 23:45:44 +0000 (16:45 -0700)]
configure: relax gtk+ version req to latest stable release
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Jun 2013 19:15:47 +0000 (12:15 -0700)]
efl: use correct callback signature
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Jun 2013 17:41:16 +0000 (11:41 -0600)]
Be explicit about MIT license in README
U. Artie Eoff [Tue, 4 Jun 2013 17:35:13 +0000 (10:35 -0700)]
weston-wfits: print log message when notify is constructed
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
Brian Lovin [Tue, 4 Jun 2013 17:14:34 +0000 (10:14 -0700)]
test_dayselector: Add user test
Signed-off-by: Brian Lovin <brian.j.lovin@intel.com>
Tested-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Jun 2013 16:04:04 +0000 (09:04 -0700)]
weston-wfits: pointer/keyboard are not pointers in older Weston SDK's
For backwards compatibility we can't do a pointer NULL check on
seat->keyboard or seat->pointer since those members were not pointers
in older Weston SDK's. We only initialize them primarily to make
the extension work on the headless backend (which doesn't init a
pointer/keyboard), thus use isHeadless() as the condition for
initialization.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Jun 2013 14:51:43 +0000 (07:51 -0700)]
efl: remove elementary map test - don't support it
The elementary map widget depends on libcurl which has license
issues. Hence, don't write tests for elementary map.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Jun 2013 05:21:04 +0000 (22:21 -0700)]
weston-wfits: Allow user to configure the input emulator
The available emulators are "uinput" and "notify". To choose
which emulator to use, define the environment variable:
WESTON_WFITS_INPUT_EMULATOR
equal to the name of the desired emulator before loading the
weston-wfits.so module into Weston.
For example:
$ export WESTON_WFITS_INPUT_EMULATOR="notify"
$ weston --modules=weston-wfits.so
If WESTON_WFITS_INPUT_EMULATOR is not defined in the environment
and Weston is using the headless-backend then weston-wfits.so will
choose "notify" by default. Otherwise, it chooses "uinput" by
default.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Jun 2013 04:59:45 +0000 (21:59 -0700)]
weston-wfits: rename input-emulator-server to input-emulator-notify
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 4 Jun 2013 03:58:07 +0000 (20:58 -0700)]
weston-wfits: split logic into separate files
Split the logic for the different wfits interfaces into
separate files. Used classes to encapsulate logic but was forced
to use primarily static members/functions since Weston has a
bug that results in a detrimental side-effect of destructing
objects that are loaded when it fails to fork/exec client
processes (e.g. weston-keyboard, weston-screensaver, etc.). That
is, when Weston fails to fork/exec it calls exit(-1) instead
of _exit(-1) and hence invokes atexit and destructors for the
cloned objects in its forked process. Those destructors might
unload/uninitialize things that effect the main process (e.g.
sockets, ioctls, etc...).
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Mon, 3 Jun 2013 16:49:07 +0000 (09:49 -0700)]
weston-wfits: move destroy listener to init file
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 31 May 2013 17:53:48 +0000 (10:53 -0700)]
weston-wfits: Move module_init to its own file
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 31 May 2013 15:55:39 +0000 (08:55 -0700)]
weston-wfits: Separate different input modes into classes
This encapsulates the logic in specialized classes and will
make it easier to add new input modes. It also enables us to
add functionality to allow runtime/dynamic selection of the input
mode.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
Brian Lovin [Fri, 31 May 2013 00:31:29 +0000 (17:31 -0700)]
gtk: Add test_button.cpp and gtk/ButtonClickTest
Added test_button.cpp and the gtk+ test case
ButtonClickTest
Signed-off-by: Brian Lovin <brian.j.lovin@intel.com>
Tested-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Thu, 30 May 2013 22:06:27 +0000 (15:06 -0700)]
efl: force tests to abort on ERR messages
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 29 May 2013 15:47:41 +0000 (08:47 -0700)]
gtk+: add fullscreen/unfullscreen test
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 28 May 2013 21:21:20 +0000 (14:21 -0700)]
util: allow geometry values to be initialized at construct
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
Sam Spilsbury [Wed, 22 May 2013 14:26:34 +0000 (22:26 +0800)]
Don't ignore the result of write ()
Fixes #3
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
Tested-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Mon, 20 May 2013 22:24:47 +0000 (15:24 -0700)]
efl: keep the wl_display assert but remove the init
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Mon, 20 May 2013 19:42:19 +0000 (12:42 -0700)]
Merge pull request #2 from uartie/master
efl: short-circuit test if wl_display connection fails
U. Artie Eoff [Mon, 20 May 2013 19:35:10 +0000 (12:35 -0700)]
efl: short-circuit test if wl_display connection fails
Force the test to abort early if ecore_wl_init fails to make
a connection to the wl_display. For example, if a test causes
the compositor to crash, then this short-circuit prevents other
tests from continuing and potentially getting stuck waiting for
a test timeout to occur.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 17 May 2013 20:36:05 +0000 (14:36 -0600)]
Update README
U. Artie Eoff [Fri, 17 May 2013 20:23:39 +0000 (14:23 -0600)]
Update README
U. Artie Eoff [Fri, 17 May 2013 19:57:01 +0000 (13:57 -0600)]
configure.ac: correct project info
Add the correct issue tracker and website
U. Artie Eoff [Fri, 17 May 2013 18:09:42 +0000 (11:09 -0700)]
README: contributing
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 17 May 2013 18:03:48 +0000 (11:03 -0700)]
README: change bug url
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 15 May 2013 20:05:34 +0000 (13:05 -0700)]
more README and TODO updates
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 15 May 2013 19:25:14 +0000 (12:25 -0700)]
more additions to README and TODO
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 15 May 2013 19:10:47 +0000 (12:10 -0700)]
README: formatting
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 15 May 2013 19:06:24 +0000 (12:06 -0700)]
add TODO, update README, initialize configure.ac info
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Wed, 15 May 2013 16:35:52 +0000 (09:35 -0700)]
weston-wfits: ensure module_init C exported. Code formatting/comments
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 14 May 2013 17:36:36 +0000 (10:36 -0700)]
gtk+: add WindowGeometryTest
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 14 May 2013 16:24:44 +0000 (09:24 -0700)]
gtk+: add test harness plus basic tests
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Tue, 14 May 2013 16:23:59 +0000 (09:23 -0700)]
efl: add some basic harness tests
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 10 May 2013 21:21:37 +0000 (14:21 -0700)]
efl: use lipsum instead of dickens
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
U. Artie Eoff [Fri, 10 May 2013 21:21:20 +0000 (14:21 -0700)]
remove empty makefile
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
Brian Lovin [Fri, 10 May 2013 19:18:29 +0000 (12:18 -0700)]
weston-wfits: Subtract 1 from absmax
There seems to be a bug on my Ubuntu 13.04 system where the pointer
is moved to a location that is -1px off on the x and y direction
unless these max values are made smaller. Shouldn't effect tests
on other systems - was tested.
Signed-off-by: Brian Lovin <brian.j.lovin@intel.com>
U. Artie Eoff [Wed, 8 May 2013 20:34:00 +0000 (13:34 -0700)]
tools: the test fullname needs to be static
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>