1 files changed, 32 insertions, 0 deletions

diff --git a/docs/grail-problems.txt b/docs/grail-problems.txt
new file mode 100644
index 0000000..ab95c28
--- /dev/null
+++ b/docs/grail-problems.txt
@@ -0,0 +1,32 @@
+Split gestures into recognition and application
+-----------------------------------------------
+
+Gesture detection is not an exact science, but the algorithms
+inevitably involve decisions influenced by what gestures are available
+to choose from. Therefore, the recognition should be considered an
+inherent property of the device itself, regardless of user interaction
+context. Just imagine having a keyboard that resized slightly
+everytime you switched from chrome to gimp.
+
+
+Tag events as being part of a gesture
+-------------------------------------
+
+One of the problems faced when replacing one set of events with
+another is where and how this filtering should take place. One window
+might listen for spiral gestures, whereas another window is a paint
+program where you actually want to draw a spiral. By introducing a
+relation between a gesture and an event, filtering becomes something
+simple that can be performed close to the window. The relation can be
+implemented as a bitmask representing what gestures the event is part
+of.
+
+
+Help the client to keep track of active gestures
+------------------------------------------------
+
+A gesture starts, continues, and either completes or aborts.
+A client that listens to a subset of gestures will have to keep track.
+Compatible to Qt approach.
+
+Would there be client feedback on confirmation?