Cleanup Service<> and ServiceFactory<> when clients go away.

Cleanup when clients go away.

We delete Service<> implementations when we get errors from the client
connection. When all Services are deleted we sever the connection from the
ServiceFactory<> to the Shell. That allows it to clean up its reference to
the factory.

BUG=None
TEST=ServiceConnectorTest.ClientError
R=darin@chromium.org, darin

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=249422

[darin (slow to review), 2014-02-06 05:00:46]

LGTM

https://codereview.chromium.org/137623017/diff/1/mojo/public/shell/lib/servic...
File mojo/public/shell/lib/service.cc (right):

https://codereview.chromium.org/137623017/diff/1/mojo/public/shell/lib/servic...
mojo/public/shell/lib/service.cc:16: shell_.reset();
It seems like at this point we should be quiting the RunLoop so that MojoMain
exits.

https://codereview.chromium.org/137623017/diff/1/mojo/public/shell/service.h
File mojo/public/shell/service.h (right):

https://codereview.chromium.org/137623017/diff/1/mojo/public/shell/service.h#...
mojo/public/shell/service.h:93: if (services_.size() == 0)
nit: "if (services_.empty())"

https://codereview.chromium.org/137623017/diff/1/mojo/shell/service_connector.h
File mojo/shell/service_connector.h (right):

https://codereview.chromium.org/137623017/diff/1/mojo/shell/service_connector...
mojo/shell/service_connector.h:45: bool HasFactoryForURL(const GURL& url) const;
nit: the service factory is a private implementation detail. it seems strange to
expose it here. maybe HasConnectionForURL or HasActiveService would be a better
name?

EDIT: but I see you are doing so for testing purposes. Maybe use a name like
HasServiceFactoryForTesting?

[DaveMoore, 2014-02-06 15:52:08]

Address review nits

[DaveMoore, 2014-02-06 15:55:57]

The CQ bit was checked by davemoore@chromium.org

[commit-bot: I haz the power, 2014-02-06 15:56:15]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/davemoore@chromium.org/137623017/70001

[commit-bot: I haz the power, 2014-02-06 16:45:30]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-02-06 16:45:30]

Retried try job too often on android_aosp for step(s) compile
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=android_ao...

[DaveMoore, 2014-02-06 17:35:12]

Committed patchset #2 manually as r249422 (presubmit successful).