How do I scale a large number of UiAutomator tests?

I have generated about 60 Android UI tests using UiAutomator, but I'm wondering if there may be a better way to implement them. Most of the answers I find (below) are simple one-test examples that don't translate to large-scale testing suites.

Examples:

• Vogella


• Everybody Tests // UiWatchers

The way I have been organizing my tests up to this point has been to each piece individually, depending on the screen they are on (Home, vs. a List, vs. a Report). Then, I create functions depending on that screen, and finally generate tests / workflows based on those functions. For example:

A public object:

public class ExampleObjects {
     public static UiObject BODY = new UiObject(new UiSelector.resourceId("android:id/body"));
}

A public function (which applies to a test):

public class ExampleFunctions {
     public static boolean isReportDeleted(String reportID) {
            // do stuff
            ExampleObjects.BODY.click();
            // do more stuff
     }
}

And finally the tests themselves:

public class ExampleTest extends UiAutomatorTestCase {
     public String REPORT_ID = "101";

     public void setUp() {
          // set up stuff
     }

     public void tearDown() {
          // tear down stuff
     }

     public void test_exampleTest() {
          assertTrue(ExampleFunctions.isReportDeleted(REPORT_ID);
     }
}

The tests themselves are JUnit style - one assertion that applies to a function defined in a different class. This makes it so I only need to fix one function if the UI changes, but I still have my tests.

My Questions:

1. Given the examples that I've seen, it looks like a lot of testers prefer to create new UiObjects / UiSelectors within the tests, as they write them. Is this common, and a better approach than what I have right now?


2. I haven't had any luck with using the UiWatchers, and I was wondering if my separation of objects / functions / tests means that the UiWatchers are ineffective. Should the watchers be defined within the tests themselves?

Thank you very much for your input.