2012-03-30 Daniel d'Andrada <email address hidden>
Release version 3.0.4
2012-03-30 Chase Douglas <email address hidden>
Merge switch from XSync awaits to alarms and other fixes
2012-03-30 Chase Douglas <email address hidden>
Don't try to set the alarm state, it won't work
Created alarms will be active, but with a trigger value of 0. It would
take 584,942,417 years to wrap around to 0, though, so I think we're
safe.
2012-03-30 Chase Douglas <email address hidden>
Switch from XSync awaits to alarms
* Fix X11 test fixture, it was querying the time instead of using the
event
* Switch from awaits to alarms because awaits inhibit other X events
until they trigger
* Create only one alarm and change it as needed
2012-03-29 Chase Douglas <email address hidden>
Merge tool man pages
2012-03-29 Stephen M. Webb <email address hidden>
fixed wording in grail-test-3-1 manpage
2012-03-29 Stephen M. Webb <email address hidden>
Added manpages for all binaries in utouch-grail-tools (lp: #955609).
2012-03-29 Chase Douglas <email address hidden>
Merge removal of the --with-xi option, it is now automatically checked
2012-03-29 Chase Douglas <email address hidden>
Check for XInput 2.2 automatically, no need for --with-xi
This fixes make distcheck so it builds the grail v3 bits too.
2012-03-28 Chase Douglas <email address hidden>
Merge fix for daily build FTBFS
2012-03-27 Chase Douglas <email address hidden>
Add xorg autoconf macro scripts to fix daily build FTBFS
2012-03-27 Chase Douglas <email address hidden>
Default to overwriting autotools files from system installed files
This ensures the build system always has the latest macros and scripts
available.
2012-03-27 Chase Douglas <email address hidden>
Merge touch state rework, fixing potential memory growth issues
2012-03-27 Chase Douglas <email address hidden>
A few miscellaneous fixes that do not affect the build
2012-03-27 Chase Douglas <email address hidden>
Always process frame events for touch changes
Otherwise, we might miss touch begins and ends, which would corrupt the
recognizer state.
This code path shouldn't even be invoked since recognizers are
dynamically created/destroyed as subscriptions are activated/deactivated.
2012-03-27 Chase Douglas <email address hidden>
Rename and recomment recognizer touch processing methods due to changes
2012-03-26 Chase Douglas <email address hidden>
Delete all ended touches after processing frame event
This allows us to remove the circular touch accounting between the
recognizer and the touch class.
2012-03-26 Chase Douglas <email address hidden>
Switch to LGPLv3 for touch class
2012-03-26 Chase Douglas <email address hidden>
Use grail internal touch objects for maintaining touch states
The current touch state accounting is haphazard. Touch state is
determined by which lists a touch is in. While we still need lists for
touch state that is recognizer-specific, like the free touches list, we
can do better for general touch state like acceptance, ending, and
ownership.
This change adds a grail touch state class. There is only one instance
for each touch per recognizer; however, the touch state is shared between
the recognizer, gestures, and slices. When the recognizer receives an
event for the touch, its state is updated and available for use
everywhere in grail.
This fixes issues where the touch ending was not being handled properly.
When a touch ended, it needed to be handled by the gesture, but then
removed from the recognizers' lists. This is now taken care of properly.
Note that the recognizer needs to know about all touches, including ended
touches, so it maintains a map of all the known touches. The map includes
weak references to the touches, so it does not prevent a touch from being
deleted.
2012-03-27 Chase Douglas <email address hidden>
Merge indentation fixes, no code changes
2012-03-27 Chase Douglas <email address hidden>
Fix improper indentation in the recognizers
2012-03-26 Chase Douglas <email address hidden>
Merge code cleanups and change to not keep track of ended accepted
gestures
2012-03-26 Daniel d'Andrada <email address hidden>
v3/Recognizer: Don't add an ended gesture to accepted_gestures_ list.
Nothing else will come out of an ended gesture. Thus no point in still
keeping it around.
2012-03-26 Daniel d'Andrada <email address hidden>
v3/Gesture: Call End() instead of replicating its code
2012-03-26 Daniel d'Andrada <email address hidden>
AtomicRecognizer: s/FindGestureToAccept/FindGesturesToAccept
That method can accept multiple gestures in one run. Thus name it
appropriately.
2012-03-26 Daniel d'Andrada <email address hidden>
AtomicRecognizer: accept a gesture by passing its pointer instead of its id
That way we avoid an unnecessary search for a gesture given its id.
2012-03-26 Chase Douglas <email address hidden>
Merge test fixes
* Don't delete subscription until after unreferencing all gesture slices
* Accept non-atomic gesture in NoTapAfterDrag test
2012-03-26 Chase Douglas <email address hidden>
Accept non-atomic gesture in NoTapAfterDrag test
2012-03-26 Chase Douglas <email address hidden>
Don't delete subscription until after unreferencing all gesture slices
This doesn't cause a bug today, but might in the future if slice deletion
requires references to the subscription to be valid.
2012-03-26 Chase Douglas <email address hidden>
Merge integration test configure flag
2012-03-23 Chase Douglas <email address hidden>
Remove leftover gtest autoconf substitutions
The substitutions are now handled in xorg-gtest.m4.
2012-03-23 Chase Douglas <email address hidden>
Add configure option for enabling/disabling integration testing
2012-03-26 Daniel d'Andrada <email address hidden>
Break reference loop between Slice and Gesture classes
Break it by removing the Gesture pointer variable from the Slice class.
Slice doesn't need to keep a pointer to the Gesture that originated it.
2012-03-26 Stephen M. Webb <email address hidden>
Change licensing from GPL-3+ to LGPL-3 (lp: #963070)
2012-03-26 Stephen M. Webb <email address hidden>
Change library license to LGPL-3 (lp: 963070)
2012-03-23 Chase Douglas <email address hidden>
Merge timeout handling fix
2012-03-22 Chase Douglas <email address hidden>
Update grail time using X sync event time instead of new time request
There may be sufficient enough skew between the event generation time and
the returned time from the server request to cause tap gestures to
timeout erroneously.
2012-03-23 Daniel d'Andrada <email address hidden>
AtomicRecognizer::FindGestureToAccept - handle negative time deltas
It's possible that a touch start time is bigger than the event time.
2012-03-23 Daniel d'Andrada <email address hidden>
Merge fixes to Recognizer::kCompositionTime
- Turn it into a real constant (add const keyword)
- Make it follow latest version of Unity's C++ Style Guide:
http://unity.ubuntu.com/wp-content/uploads/2012/03/cppguide.html#Constant_Names
2012-03-23 Daniel d'Andrada <email address hidden>
s/kCompositionTime/COMPOTISION_TIME
Following latest version of Unity's C++ Style Guide:
http://unity.ubuntu.com/wp-content/uploads/2012/03/cppguide.html#Constant_Names
2012-03-23 Daniel d'Andrada <email address hidden>
Turn Recognizer::kCompositionTime into a real constant
2012-03-23 Chase Douglas <email address hidden>
Merge fix for linking tests against appropriate libtool files
2012-03-23 Jussi Pakkanen <email address hidden>
Do not link against libtool hidden stuff.
2012-03-23 Jussi Pakkanen <email address hidden>
Merged out-of-source build fix.
2012-03-23 Jussi Pakkanen <email address hidden>
Tests can be run from a build directory.
2012-03-23 Jussi Pakkanen <email address hidden>
Merged uninitialization fix.
2012-03-23 Jussi Pakkanen <email address hidden>
Initialize a previously unintialized object variable.
2012-03-23 Jussi Pakkanen <email address hidden>
Merged virtual destructor fix.
2012-03-23 Jussi Pakkanen <email address hidden>
Added virtual destructors to slices.
2012-03-22 Daniel d'Andrada <email address hidden>
Remove threaded recording playback from parallel-atomic-gestures test as well
This has been done already for the other tests by revision 200.4.4