)]}'
{
  "log": [
    {
      "commit": "be1aa8250cee7819c49741e819e81659d1d03823",
      "tree": "3041a44459110da304f930b1a3ebcbd042291999",
      "parents": [
        "ad65b46d5f6c569b588556dcf85b64d59600b175"
      ],
      "author": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Wed Jul 27 16:04:54 2011 -0700"
      },
      "committer": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Sun Jul 31 15:38:09 2011 -0700"
      },
      "message": "Refactor input reader to add stylus support.\nBug: 5064702\n\nIntroduced the concept of an InputListener to further decouple\nthe InputReader from the InputDispatcher.  The InputListener\nexposes just the minimum interface that the InputReader needs\nto communicate with the outside world.  The InputReader\npasses arguments to the InputListener by reference, which makes\nit easy to queue them up.\n\nConsolidated all of the InputReader locks into one simple global\nMutex.  The reason this wasn\u0027t done before was due to potential\nre-entrance in outbound calls to the InputDispatcher.  To fix this,\nthe InputReader now queues up all of the events it wants to send\nusing a QueuedInputListener, then flushes them outside of the\ncritical section after all of the event processing is finished.\nRemoving all of the InputMapper locks greatly simplifies the\nimplementation.\n\nAdded tests for new stylus features such as buttons, tool types,\nand hovering.\n\nAdded some helpers to BitSet32 to handle common code patterns\nlike finding the first marked bit and clearing it.\n\nFixed a bug in VelocityTracker where the wrong pointer trace\ncould get cleared when handling ACTION_POINTER_DOWN.  Oops.\n\nChanged PointerCoords so it no longer stores useless zero\naxis values.  Removed editAxisValue because it is not very\nuseful when all zero value axes are absent and therefore\ncannot be edited in place.\n\nAdded dispatch of stylus hover events.\n\nAdded support for distance and tool types.\n\nChange-Id: I4cf14d134fcb1db7d10be5f2af7b37deef8f8468\n"
    },
    {
      "commit": "6894a2947eb1f9d499fd7f1a1ec4e7098e07d25d",
      "tree": "5aaf199228185826ea5deb9200adfc1633519994",
      "parents": [
        "63acea6ad4a9316e1450dea47840f45df4aaed25"
      ],
      "author": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Fri Jul 01 17:59:27 2011 -0700"
      },
      "committer": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Fri Jul 01 19:08:15 2011 -0700"
      },
      "message": "Workaround apps that make assumptions about pointer ids.\n\nModified the touch input mapper to assign pointer ids sequentially\nstarting from 0 instead of using the tracking id or slot index\nsupplied by the driver.  Applications should not depend on this\nordering but some do.  (sigh)\n\nBug: 4980884\nChange-Id: I0dfeb3ac27c57a7102a13c960c760e2a02eb7669\n"
    },
    {
      "commit": "2ed2462aa29c564f5231f317c27b3188da875e52",
      "tree": "63209b0ae028c1353e79d79b5e4000e7c4402ecc",
      "parents": [
        "ace13b17866dc9136aeecf6dfaf7077f37434469"
      ],
      "author": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Mon Mar 14 19:39:54 2011 -0700"
      },
      "committer": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Tue Mar 15 19:59:47 2011 -0700"
      },
      "message": "Improve VelocityTracker numerical stability.\n\nReplaced VelocityTracker with a faster and more accurate\nnative implementation.  This avoids the duplicate maintenance\noverhead of having two implementations.\n\nThe new algorithm requires that the sample duration be at least\n10ms in order to contribute to the velocity calculation.  This\nensures that the velocity is not severely overestimated when\nsamples arrive in bursts.\n\nThe new algorithm computes the exponentially weighted moving\naverage using weights based on the relative duration of successive\nsample periods.\n\nThe new algorithm is also more careful about how it handles\nindividual pointers going down or up and their effects on the\ncollected movement traces.  The intent is to preserve the last\nknown velocity of pointers as they go up while also ensuring\nthat other motion samples do not count twice in that case.\n\nBug: 4086785\nChange-Id: I2632321232c64d6b8faacdb929e33f60e64dcdd3\n"
    },
    {
      "commit": "ace13b17866dc9136aeecf6dfaf7077f37434469",
      "tree": "0ec5d01ef83b4d5148525f5befa28ca026b294c7",
      "parents": [
        "fd10d5cf56e5b1ba7692400e4fe4ae26b61f3285"
      ],
      "author": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Wed Mar 09 17:39:48 2011 -0800"
      },
      "committer": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Mon Mar 14 14:12:03 2011 -0700"
      },
      "message": "Use touch pad gestures to manipulate the pointer.\n\n1. Single finger tap performs a click.\n2. Single finger movement moves the pointer (hovers).\n3. Button press plus movement performs click or drag.\n   While dragging, the pointer follows the finger that is moving\n   fastest.  This is important if there are additional fingers\n   down on the touch pad for the purpose of applying force\n   to an integrated button underneath.\n4. Two fingers near each other moving in the same direction\n   are coalesced as a swipe gesture under the pointer.\n5. Two or more fingers moving in arbitrary directions are\n   transformed into touches in the vicinity of the pointer.\n   This makes scale/zoom and rotate gestures possible.\n\nAdded a native VelocityTracker implementation to enable intelligent\nswitching of the active pointer during drags.\n\nChange-Id: I5ada57e7f2bdb9b0a791843eb354a8c706b365dc\n"
    },
    {
      "commit": "01ce2e9eee41cc0c24b0d16465710a28ea337d5d",
      "tree": "08e45a8ec47e63760655a1770dac367258511538",
      "parents": [
        "37cedce14bc0e68e8bf27c2cecb181728226e6ed"
      ],
      "author": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Sun Sep 26 22:20:12 2010 -0700"
      },
      "committer": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Sun Sep 26 22:20:12 2010 -0700"
      },
      "message": "Add suuport for splitting touch events across windows.\n\nThis feature is currently used to enable dragging the start and end\nselection handles of a TextView at the same time.  Could be used for\nother things later.\n\nDeleted some dead code in ArrowKeyMovementMethod and CursorControllers.\n\nChange-Id: I930accd97ca1ca1917aab8a807db2c950fc7b409\n"
    },
    {
      "commit": "46b9ac0ae2162309774a7478cd9d4e578747bfc2",
      "tree": "46ad021a41e25ca9f1250b709a29b724dc6b504d",
      "parents": [
        "f62c57d684b83df7d2817db976c0afdb500ae92a"
      ],
      "author": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Thu Apr 22 18:58:52 2010 -0700"
      },
      "committer": {
        "name": "Jeff Brown",
        "email": "jeffbrown@google.com",
        "time": "Sun Jun 13 17:42:16 2010 -0700"
      },
      "message": "Native input dispatch rewrite work in progress.\n\nThe old dispatch mechanism has been left in place and continues to\nbe used by default for now.  To enable native input dispatch,\nedit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy.\n\nIncludes part of the new input event NDK API.  Some details TBD.\n\nTo wire up input dispatch, as the ViewRoot adds a window to the\nwindow session it receives an InputChannel object as an output\nargument.  The InputChannel encapsulates the file descriptors for a\nshared memory region and two pipe end-points.  The ViewRoot then\nprovides the InputChannel to the InputQueue.  Behind the\nscenes, InputQueue simply attaches handlers to the native PollLoop object\nthat underlies the MessageQueue.  This way MessageQueue doesn\u0027t need\nto know anything about input dispatch per-se, it just exposes (in native\ncode) a PollLoop that other components can use to monitor file descriptor\nstate changes.\n\nThere can be zero or more targets for any given input event.  Each\ninput target is specified by its input channel and some parameters\nincluding flags, an X/Y coordinate offset, and the dispatch timeout.\nAn input target can request either synchronous dispatch (for foreground apps)\nor asynchronous dispatch (fire-and-forget for wallpapers and \"outside\"\ntargets).  Currently, finding the appropriate input targets for an event\nrequires a call back into the WindowManagerServer from native code.\nIn the future this will be refactored to avoid most of these callbacks\nexcept as required to handle pending focus transitions.\n\nEnd-to-end event dispatch mostly works!\n\nTo do: event injection, rate limiting, ANRs, testing, optimization, etc.\n\nChange-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25\n"
    }
  ]
}