Enable all tests in CopylessPasteTest

chrome/android/javatests/src/org/chromium/chrome/browser/CopylessPasteTest.java

Index: chrome/android/javatests/src/org/chromium/chrome/browser/CopylessPasteTest.java
diff --git a/chrome/android/javatests/src/org/chromium/chrome/browser/CopylessPasteTest.java b/chrome/android/javatests/src/org/chromium/chrome/browser/CopylessPasteTest.java
index 59d08fcb1632d6603b3ca1b85dd0b5e5165be52e..755941d141fd54f5a2efa51d1037bd502b690fda 100644
--- a/chrome/android/javatests/src/org/chromium/chrome/browser/CopylessPasteTest.java
+++ b/chrome/android/javatests/src/org/chromium/chrome/browser/CopylessPasteTest.java
@@ -10,7 +10,6 @@ import org.chromium.base.Callback;
 import org.chromium.base.ThreadUtils;
 import org.chromium.base.test.util.CallbackHelper;
 import org.chromium.base.test.util.CommandLineFlags;
-import org.chromium.base.test.util.DisabledTest;
 import org.chromium.base.test.util.Feature;
 import org.chromium.base.test.util.Restriction;
 import org.chromium.blink.mojom.document_metadata.Entity;
@@ -166,21 +165,15 @@ public class CopylessPasteTest extends ChromeTabbedActivityTestBase {
      */
     @LargeTest
     @Feature({"CopylessPaste"})
-    @DisabledTest(message = "Flaky: crbug.com/713172")
     public void testCache() throws InterruptedException, TimeoutException {
-        // The URLs used here should be unique in CopylessPasteTest.
-        String uniqueTag = "#123";
-        // NODATA_PAGE doesn't contain desired metadata.
-        loadUrl(mTestServer.getURL(NODATA_PAGE + uniqueTag));
-        mCallbackHelper.waitForCallback(0);
-        // DATA_PAGE contains desired metadata.
-        loadUrl(mTestServer.getURL(DATA_PAGE + uniqueTag));
+        loadUrl(mTestServer.getURL(NODATA_PAGE));
+        loadUrl(mTestServer.getURL(DATA_PAGE));

[Maria, 2017/04/20 21:34:30]

From CL description, it sounds like you think that two waits are what's causing
flakiness (I find this a really odd explanation). But here you also remove the
unique tag from the test. Do you think that has an effect too?

I am a bit concerned that we are fixing the test by removing an expectation.

[wychen, 2017/04/20 22:00:54]

The flakiness of testNoMeta and testValid is around half of testCache, and all
the failures are about wait timeout, so I think this is the most reasonable
explanation.

It turned out that the uniqueTag is not needed since each test begins with a
clean cache. It shouldn't affect the flakiness.

The waits here are a bit redundant, since these expectations were checked in
other tests (testNoMeta and testValid respectively).
The cache-specific expectation is that repeated URLs don't get callback.
I keep one wait for readability.

         mCallbackHelper.waitForCallback(1);
 
         // Cache hit without entities. Shouldn't parse again.
-        loadUrl(mTestServer.getURL(NODATA_PAGE + uniqueTag));
+        loadUrl(mTestServer.getURL(NODATA_PAGE));
         // Cache hit with entities. Shouldn't parse again.
-        loadUrl(mTestServer.getURL(DATA_PAGE + uniqueTag));
+        loadUrl(mTestServer.getURL(DATA_PAGE));
         assertEquals(2, mCallbackHelper.getCallCount());
     }
 }