Moves BackgroundShellApplicationLoader to shell/android

Moves BackgroundShellApplicationLoader to shell/android

As this is the only place that uses it. Also renamed to
BackgroundApplicationLoader.

BUG=none
TEST=none
R=qsr@chromium.org

Committed: https://chromium.googlesource.com/external/mojo/+/f0ef0825babf1f08a38af4b2ef08d770817c7e30

[sky, 2014-12-16 19:22:19]

https://codereview.chromium.org/788243007/diff/40001/mojo/application_manager...
File mojo/application_manager/application_manager_unittest.cc (left):

https://codereview.chromium.org/788243007/diff/40001/mojo/application_manager...
mojo/application_manager/application_manager_unittest.cc:635:
TEST_F(ApplicationManagerTest, BDeleted) {
Contrary to the description this test was exercising a combination of
BackgroundShellApplicationLoader and the handle watcher code. The reason the
code worked before was resetting the loader on 646 (old) triggered shutting down
the background thread and background message loop. Shutting down the MessageLoop
closes the pipes. As A/B are InterfaceImpls they are destroyed when the pipe
closes.

Removing BackgroundShellApplicationLoader from here means resetting the app (646
old) no longer triggers closing pipes and this test fails.

[sky, 2014-12-16 19:22:30]

sky@chromium.org changed reviewers:
+ qsr@chromium.org

[qsr, 2014-12-17 10:34:34]

https://codereview.chromium.org/788243007/diff/40001/mojo/application_manager...
File mojo/application_manager/application_manager_unittest.cc (left):

https://codereview.chromium.org/788243007/diff/40001/mojo/application_manager...
mojo/application_manager/application_manager_unittest.cc:635:
TEST_F(ApplicationManagerTest, BDeleted) {
On 2014/12/16 19:22:19, sky wrote:
> Contrary to the description this test was exercising a combination of
> BackgroundShellApplicationLoader and the handle watcher code. The reason the
> code worked before was resetting the loader on 646 (old) triggered shutting
down
> the background thread and background message loop. Shutting down the
MessageLoop
> closes the pipes. As A/B are InterfaceImpls they are destroyed when the pipe
> closes.
> 
> Removing BackgroundShellApplicationLoader from here means resetting the app
(646
> old) no longer triggers closing pipes and this test fails.

 Can you still make this work by having Tester use WeakBindToPipe instead of
BindToPipe and close all implementation when going away?

 It seems we still want to be able to test this use case.

https://codereview.chromium.org/788243007/diff/40001/shell/BUILD.gn
File shell/BUILD.gn (right):

https://codereview.chromium.org/788243007/diff/40001/shell/BUILD.gn#newcode395
shell/BUILD.gn:395: sources += [
"android/background_application_loader_unittest.cc" ]
This test is not run on android at the moment :( I will add it.

[sky, 2014-12-18 00:57:53]

This is now relative to https://codereview.chromium.org/812083002/ .

[qsr, 2014-12-18 08:59:46]

lgtm

[sky, 2014-12-18 16:55:32]

Committed patchset #6 (id:100001) manually as
f0ef0825babf1f08a38af4b2ef08d770817c7e30 (presubmit successful).