Merge lp://qastaging/~alan-griffiths/mir/spike-nested-input into lp://qastaging/~mir-team/mir/trunk

FAILED: Continuous integration, rev:1057
http://jenkins.qa.ubuntu.com/job/mir-ci/1445/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-saucy-i386-build/1950
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-saucy-amd64-build/1835
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/683/console

Click here to trigger a rebuild:
http://s-jenkins:8080/job/mir-ci/1445/rebuild

PASSED: Continuous integration, rev:1057
http://jenkins.qa.ubuntu.com/job/mir-ci/1447/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-saucy-i386-build/1954
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-saucy-amd64-build/1839
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/685
        deb: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/685/artifact/work/output/*zip*/output.zip

Click here to trigger a rebuild:
http://s-jenkins:8080/job/mir-ci/1447/rebuild

FAILED: Continuous integration, rev:1059
http://jenkins.qa.ubuntu.com/job/mir-ci/1450/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-saucy-i386-build/1958
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-saucy-amd64-build/1843
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/688/console

Click here to trigger a rebuild:
http://s-jenkins:8080/job/mir-ci/1450/rebuild

PASSED: Continuous integration, rev:1059
http://jenkins.qa.ubuntu.com/job/mir-ci/1451/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-saucy-i386-build/1959
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-saucy-amd64-build/1844
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/689
        deb: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/689/artifact/work/output/*zip*/output.zip

Click here to trigger a rebuild:
http://s-jenkins:8080/job/mir-ci/1451/rebuild

291 + std::shared_ptr<input::EventFilter> const& event_handler) :
event_filter vs. event_handler (pre-existing, and is common naming problem in the code elsewhere)

517 + // TODO can reviewers look carefully at these?
my best guess is security reasons, although i'm not very sure either

all in all, looks good

1. I'm concerned about this:
337 + case mir_event_type_surface:
338 + // TODO Surface attributes shouldn't be changing. So what do we do if they do?
339 + break;
Not sure of the context though. Seems strange and dangerous to drop a whole family of event types. And if it's only that we want to translate motion events, why do we handle mir_event_type_key still?

2. mi::NestedInputRelay::handle() has a scary degree of coupling. Can we not copy the event data more atomically?

> 2. mi::NestedInputRelay::handle() has a scary degree of coupling. Can we not
> copy the event data more atomically?

I wish! The types we adopted from the android input stack are difficult to work with (there's similar code doing the converse transformation) - hopefully we can do some refactoring once we are feature complete.

PASSED: Continuous integration, rev:1061
http://jenkins.qa.ubuntu.com/job/mir-ci/1462/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-saucy-i386-build/1974
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-saucy-amd64-build/1859
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/700
        deb: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/700/artifact/work/output/*zip*/output.zip

Click here to trigger a rebuild:
http://s-jenkins:8080/job/mir-ci/1462/rebuild

524 + static auto const policy_flags = ::android::POLICY_FLAG_INJECTED || ::android::POLICY_FLAG_TRUSTED;

We should be using bitwise OR (|).

::android::POLICY_FLAG_INJECTED is automatically added by injectInputEvent(), so there is no need to add it.

::android::POLICY_FLAG_TRUSTED is also added by injectInputEvent() because uid == 0, but I guess it's better to be explicit about it.

I am not sure if we also need to use ::android::POLICY_FLAG_PASS_TO_USER.

> 524 + static auto const policy_flags = ::android::POLICY_FLAG_INJECTED ||
> ::android::POLICY_FLAG_TRUSTED;
>
> We should be using bitwise OR (|).

Good catch, thanks!

> I am not sure if we also need to use ::android::POLICY_FLAG_PASS_TO_USER.

I think that is correctly done by mia::EventFilterDispatcherPolicy

Looks good.

PASSED: Continuous integration, rev:1062
http://jenkins.qa.ubuntu.com/job/mir-ci/1464/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-saucy-i386-build/1977
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-saucy-amd64-build/1862
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/702
        deb: http://jenkins.qa.ubuntu.com/job/mir-saucy-amd64-ci/702/artifact/work/output/*zip*/output.zip

Click here to trigger a rebuild:
http://s-jenkins:8080/job/mir-ci/1464/rebuild

Top approving.

This code can't be tested on GBM at the moment. (We end up crashing egl when we try for a nested configuration - Alexandros is working on it.)

On android, it doesn't work (but doesn't crash). I'm continuing to work on this.

But I think the code is better off living on trunk while these problems are resolved.