Merge lp:~vanvugt/mir/fix-1506331 into lp:mir

[Daniel van Vugt]

Revert optimization r2782. Although it gave us a nice 5ms reduction in
touch latency, it also reduced Android-input's ability to smooth the curve
of the resulting event stream. And that reduction in smoothness is going
to be visible in some touch scrolling scenarios, just as it is visible in
mir_demo_client_fingerpaint / mir_demo_client_target. (LP: #1506331)

Screenshots of before and after are in bug 1506331.

[Alexandros Frantzis (afrantzis)]

Looks good.

review: Approve

[PS Jenkins bot (ps-jenkins)]

FAILED: Continuous integration, rev:3026
http://jenkins.qa.ubuntu.com/job/mir-ci/5275/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-vivid-i386-build/4457
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-clang-vivid-amd64-build/3364/console
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-vivid-touch/4399
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-wily-touch/357/console
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1429
        deb: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1429/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/357
        deb: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/357/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4400
        deb: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4400/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-runner-touch/7037
    SUCCESS: http://s-jenkins.ubuntu-ci:8080/job/touch-flash-device/24224
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-wily-armhf/358/console

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5275/rebuild

review: Needs Fixing (continuous-integration)

[Daniel van Vugt (vanvugt)]

Hmm failures appear to be unrelated...
[ FAILED ] NestedServer.display_configuration_changes_are_visible_to_client
[ FAILED ] FD leak was detected
[ FAILED ] ThreadedDispatcherSignalTest.keeps_dispatching_after_signal_interruption

[PS Jenkins bot (ps-jenkins)]

FAILED: Continuous integration, rev:3027
http://jenkins.qa.ubuntu.com/job/mir-ci/5294/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-vivid-i386-build/4482
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-clang-vivid-amd64-build/3389/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-vivid-touch/4424/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-wily-touch/382/console
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1448
        deb: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1448/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/382
        deb: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/382/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4425
        deb: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4425/artifact/work/output/*zip*/output.zip
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-runner-touch/7061/console
    SUCCESS: http://s-jenkins.ubuntu-ci:8080/job/touch-flash-device/24269
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-wily-armhf/383/console

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5294/rebuild

review: Needs Fixing (continuous-integration)

[PS Jenkins bot (ps-jenkins)]

PASSED: Continuous integration, rev:3028
http://jenkins.qa.ubuntu.com/job/mir-ci/5320/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-vivid-i386-build/4512
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-vivid-amd64-build/3419
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-vivid-touch/4454
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-wily-touch/412/console
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1474
        deb: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1474/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/412
        deb: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/412/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4455
        deb: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4455/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-runner-touch/7090
    SUCCESS: http://s-jenkins.ubuntu-ci:8080/job/touch-flash-device/24320
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-wily-armhf/413/console

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5320/rebuild

review: Approve (continuous-integration)

[Alan Griffiths (alan-griffiths)]

OK

review: Approve

=== modified file '3rd_party/android-input/android/frameworks/base/services/input/InputTransport.cpp'
--- 3rd_party/android-input/android/frameworks/base/services/input/InputTransport.cpp	2015-08-17 18:33:38 +0000
+++ 3rd_party/android-input/android/frameworks/base/services/input/InputTransport.cpp	2015-10-19 02:38:04 +0000
@@ -43,11 +43,7 @@
 
 // Latency added during resampling.  A few milliseconds doesn't hurt much but
 // reduces the impact of mispredicted touch positions.
-//static constexpr const std::chrono::nanoseconds RESAMPLE_LATENCY = std::chrono::nanoseconds(5 * NANOS_PER_MS);
-
-// Mir modification: No artificial latency please. This seems to provide
-// visibly and measurably lower input latency, with no noticeable down side...
-static constexpr const std::chrono::nanoseconds RESAMPLE_LATENCY(0);
+static constexpr const std::chrono::nanoseconds RESAMPLE_LATENCY = std::chrono::nanoseconds(5 * NANOS_PER_MS);
 
 // Minimum time difference between consecutive samples before attempting to resample.
 static constexpr const std::chrono::nanoseconds RESAMPLE_MIN_DELTA = std::chrono::nanoseconds(2 * NANOS_PER_MS);

=== modified file 'tests/unit-tests/client/input/test_android_input_receiver.cpp'
--- tests/unit-tests/client/input/test_android_input_receiver.cpp	2015-08-17 18:33:38 +0000
+++ tests/unit-tests/client/input/test_android_input_receiver.cpp	2015-10-19 02:38:04 +0000
@@ -222,51 +222,6 @@
     EXPECT_FALSE(mt::fd_is_readable(receiver.watch_fd()));
 }
 
-TEST_F(AndroidInputReceiverSetup, no_artificial_latency_in_resampling)
-{
-    using namespace testing;
-    using namespace std::chrono;
-    using namespace std::literals::chrono_literals;
-
-    auto t = 0ns;
-    MirEvent ev;
-    bool handler_called{false};
-
-    mircva::InputReceiver receiver{
-        client_fd,
-        std::make_shared<mircv::XKBMapper>(),
-        [&ev, &handler_called](MirEvent* event)
-        {
-            ev = *event;
-            handler_called = true;
-        },
-        std::make_shared<mircv::NullInputReceiverReport>(),
-        [&t](int)
-        {
-            return t;
-        }
-    };
-    TestingInputProducer producer(server_fd);
-
-    producer.produce_a_pointer_event(123, 456, t);
-    flush_channels();
-
-    while (!mt::fd_becomes_readable(receiver.watch_fd(), 1ms) && t < 100ms)
-        t += 1ms;
-
-    receiver.dispatch(md::FdEvent::readable);
-    EXPECT_TRUE(handler_called);
-    ASSERT_EQ(mir_event_type_motion, ev.type);
-
-    auto const resample_latency_ms =
-        duration_cast<std::chrono::milliseconds>(t);
-
-    // Check that we're not using the Android-default RESAMPLE_LATENCY of 5ms
-    // which is too high...
-    // Use plain integers so any failures are readable:
-    EXPECT_THAT(resample_latency_ms.count(), Lt(1));
-}
-
 TEST_F(AndroidInputReceiverSetup, slow_raw_input_doesnt_cause_frameskipping)
 {  // Regression test for LP: #1372300
     using namespace testing;
@@ -304,11 +259,14 @@
     EXPECT_TRUE(handler_called);
     ASSERT_EQ(mir_event_type_key, ev.type);
 
+    // The motion is still too new. Won't be reported yet, but is batched.
     auto start = high_resolution_clock::now();
 
     EXPECT_TRUE(mt::fd_becomes_readable(receiver.watch_fd(), 1ms));
     handler_called = false;
     receiver.dispatch(md::FdEvent::readable);
+    // We've processed the data, but no new event has been generated.
+    EXPECT_FALSE(handler_called);
 
     auto end = high_resolution_clock::now();
     auto duration = end - start;
@@ -322,7 +280,7 @@
 
     // But later in a frame or so, the motion will be reported:
     t += 2 * one_frame;  // Account for the slower 59Hz event rate
-    EXPECT_TRUE(handler_called || mt::fd_becomes_readable(receiver.watch_fd(), next_event_timeout));
+    EXPECT_TRUE(mt::fd_becomes_readable(receiver.watch_fd(), next_event_timeout));
     receiver.dispatch(md::FdEvent::readable);
 
     EXPECT_TRUE(handler_called);