The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/020d79def4fe5affa527ef9113dc9680519d5448

commit 020d79def4fe5affa527ef9113dc9680519d5448
Author: twellington <twellington@chromium.org>
Date: Thu Oct 06 22:57:56 2016

Annotate WebappSplashScreenIconTest with @RetryOnFailure

BUG= 653701 
TBR=mlamouri@chromium.org

Review-Url: https://codereview.chromium.org/2396223002
Cr-Commit-Position: refs/heads/master@{#423719}

[modify] https://crrev.com/020d79def4fe5affa527ef9113dc9680519d5448/chrome/android/javatests/src/org/chromium/chrome/browser/webapps/WebappSplashScreenIconTest.java

benwells@ team is working on web apps. Assigning to him for triage.

Oops missed this. Passing on to Dom who's looking after the web app UI stuff.

Also removing the view restriction.

I think I've already de-flaked WebappSplashScreenIconTest#testUmaCustomIcon. The only non-results are on the Lollipop Tablet tester now, and those seem to be a result of the tablet disconnecting / some unrelated issue.

WebappSplashScreenIconTest#testShowSplashIcon is pretty sad on the Lollipop Low-end Tester It occasionally take forever to complete. But I wonder whether that's also a bot issue as well since there's a bimodal distribution in how long a test run takes (passes are < 7 seconds, but failures > 10 seconds).

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/020d79def4fe5affa527ef9113dc9680519d5448

commit 020d79def4fe5affa527ef9113dc9680519d5448
Author: twellington <twellington@chromium.org>
Date: Thu Oct 06 22:57:56 2016

Annotate WebappSplashScreenIconTest with @RetryOnFailure

BUG= 653701 
TBR=mlamouri@chromium.org

Review-Url: https://codereview.chromium.org/2396223002
Cr-Commit-Position: refs/heads/master@{#423719}

[modify] https://crrev.com/020d79def4fe5affa527ef9113dc9680519d5448/chrome/android/javatests/src/org/chromium/chrome/browser/webapps/WebappSplashScreenIconTest.java

[Automated comment] removing mislabelled merge-merged-2840

Log results: https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Fchromium.swarm%2FAndroid_N5X_Swarm%2F2174%2F%2B%2Frecipes%2Fsteps%2Fchrome_public_test_apk_on_Android%2F0%2Fstdout

I  570.612s run_tests_on_device(00d7b59e51c8b063)  detected failure in org.chromium.chrome.browser.webapps.WebappSplashScreenIconTest#testShowSplashIcon. raw output:
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_STATUS: numtests=1
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_STATUS: stream=
I  570.613s run_tests_on_device(00d7b59e51c8b063)    org.chromium.chrome.browser.webapps.WebappSplashScreenIconTest:
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_STATUS: id=AndroidJUnitRunner
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_STATUS: test=testShowSplashIcon
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_STATUS: class=org.chromium.chrome.browser.webapps.WebappSplashScreenIconTest
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_STATUS: current=1
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_STATUS_CODE: 1
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_RESULT: shortMsg=java.lang.NullPointerException
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_RESULT: longMsg=java.lang.NullPointerException: Attempt to read from field 'java.lang.String org.chromium.chrome.browser.download.DownloadInfo.mUserAgent' on a null object reference
I  570.613s run_tests_on_device(00d7b59e51c8b063)    INSTRUMENTATION_CODE: 0

What, DownloadInfo is a null pointer? That doesn't seem to make any sense, this test shouldn't be downloading anything....

+qinmin

Any idea why this test would be failing with a NullPointerException on calling DownloadInfo#mUserAgent()? I've had a poke through the callers of that method but can't really figure out any time that object could be null.

I'm also having difficulty understanding why this method is being called in the execution of a webapp test.

https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Fchromium.android%2FMarshmallow_Tablet_Tester%2F9751%2F%2B%2Frecipes%2Fsteps%2Fchrome_public_test_apk_on_Android%2F0%2Fstdout

I have no idea what the Webapp tests is trying to test, and whether the logs are correct

checking the code, the only place we need DownloadInfo.mUserAgent is when Chrome send a download to Android DownloadManager.
Since Chrome now handles almost all downloads,  so the only remaining download to be handled is OMA download. But I don't see a null DownloadInfo in all code paths.

Test now deflaked as of crrev.com/c/611660 and a few other changes.