Looks like it started here:
https://ci.chromium.org/buildbot/chromium.clang/ToTWin64/798https://ci.chromium.org/buildbot/chromium.clang/ToTWin/830
The failures all look ntp-related, e.g.
../../chrome/browser/ui/browser_browsertest.cc(1158): error: Expected equality of these values:
chrome::kChromeUINewTabURL
Which is: 00000001461B5C70 pointing to "chrome://newtab/"
tab_strip_model->GetActiveWebContents()->GetLastCommittedURL().spec()
Which is: "chrome-search://local-ntp/local-ntp.html"
So maybe it's just an official build thing like you guessed off-bug.
blundell's https://chromium-review.googlesource.com/870313 mentions "ntp" -- blundell, any chance that that change changed the ntp url in official builds?
Cc: treib@chromium.org cco3@chromium.org Summary: NTP-related browser tests failing in official builds. (was: browser_tests failing on Clang's ToTWin and ToTWin64 builders)
https://chromium-review.googlesource.com/868994 looks related too.
Here's a list of failing tess:
InputImeApiTest.SendKeyEventsOnNormalPage
StartupBrowserCreatorTest.ProfilesWithoutPagesNotLaunched
ContentScriptApiTests/ContentScriptApiTest.CannotScriptTheNewTabPage/0
ContentScriptApiTests/ContentScriptApiTest.CannotScriptTheNewTabPage/1
BrowserTest.ClearPendingOnFailUnlessNTP
ExtensionApiTabTest.TabsOnUpdated
BrowserTest.NavigateToDefaultNTPPageOnExtensionUnload
AllUrlsApiTest.WhitelistedExtension
ExtensionURLRewriteBrowserTest.NewTabPageURL
PolicyTest.HomepageLocation
BrowserNavigatorTest.NavigateFromNTPToOptionsInSameTab
(see above for an example failure)
It shouldn't be the change to NTP Snippets -- that's just refactoring in the code that fetches content suggestions and displays them on the NTP.
treib@ is the right person to be CC'd here AFAIK :).
Looking at the commits going into
https://ci.chromium.org/buildbot/chromium.clang/ToTWin/830
There's one that mentions NTP. No idea if this is related.. I'm still trying to get the test to run at all on my machine.
---
Replace the NTP Interceptor with Throttle
This change replaces the NewTabPageInterceptorService with a
NewTabPageNavigationThrottle. The throttle reduces complexity, but it
is also neutral regarding the network stack / network service, so it
fixes the NewTabPageInterceptorServiceTests that were broken with the
NetworkService feature enabled. These tests are now renamed
NewTabPageNavigationThrottleTest.
BUG= 802926
Cq-Include-Trybots: master.tryserver.chromium.linux:linux_mojo
Change-Id: Iec0664df33e79d6bcc18f60aecc4017de7f65858
Reviewed-on: https://chromium-review.googlesource.com/868994
Commit-Queue: Conley Owens <cco3@chromium.org>
Reviewed-by: Marc Treib <treib@chromium.org>
Reviewed-by: Matt Menke <mmenke@chromium.org>
Reviewed-by: Reilly Grant <reillyg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#530370}
---
blundell's CL only affects mobile, so that's not the culprit.
The Throttle does look related. What it does is redirect to the local NTP ("chrome-search://local-ntp/local-ntp.html", i.e. the one that shows up in the error messages) on various errors. Maybe the new Throttle is more eager than the old Interceptor was? No idea why that would only happen in official builds though.
Verified locally that reverting the NTP Throttle patch fixes at least the StartupBrowserCreatorTest.ProfilesWithoutPagesNotLaunched test.
Over to cco3.
Since the patch still reverts cleanly, I'd suggest that you revert it before starting to fix.
hans@, You said that you verified reverting locally fixes it. Does that mean you were able to reproduce locally? The tests I'm looking at, it seems that it must have to do with the configuration on the trybots.
It repros in official builds. Set
is_official_build=true
is_chrome_branded=true
in your args.gn and add src-internal to your .gclient. Sync, rebuild, you should be able to repro.
Comment 1 by thakis@chromium.org
, Feb 1 2018