This was previously disabled in crbug.com/757165

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5c12b24dc8d0bf3f595f2e37c33a605aff26a094

commit 5c12b24dc8d0bf3f595f2e37c33a605aff26a094
Author: tmartino <tmartino@chromium.org>
Date: Fri Oct 13 23:02:07 2017

Marking snav-container-white-space.html flaky

Bug:  774688 
Change-Id: I9cbeba332838cf54ea76915364a06b743834df1e

TBR=qyearsley@chromium.org,jeffcarp@chromium.org,dpranke@chromium.org

Change-Id: I9cbeba332838cf54ea76915364a06b743834df1e
Reviewed-on: https://chromium-review.googlesource.com/719610
Reviewed-by: Tommy Martino <tmartino@chromium.org>
Commit-Queue: Tommy Martino <tmartino@chromium.org>
Cr-Commit-Position: refs/heads/master@{#508859}
[modify] https://crrev.com/5c12b24dc8d0bf3f595f2e37c33a605aff26a094/third_party/WebKit/LayoutTests/TestExpectations

Naively assigning to xiaochengh@, who has some of the more recent changes in this directory. Hoping you can triage this better than I can :)

Assigning to kochi@ since the test is about spatial navigation.

Hugo, could you take a look?

Looking at the test failure [1], I see that the first Down key is missing! This is similar to Issue 753192.

TEST COMPLETE
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 1. Was start. // Missed the Down to 1.
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 2. Was 1.
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 3. Was 2.
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 4. Was 3.
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 5. Was 4. 
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 4. Was 3.
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 3. Was 2.
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 2. Was 1.
FAIL gFocusedDocument.activeElement.getAttribute("id") should be 1. Was start.
PASS gFocusedDocument.activeElement.getAttribute("id") is "start"
This test tests that a anchor element with in image in it and a space after the image, can still give the correct size in ContainerNode::getRect().

This missing Down-key seems to be the problem on both Mac [1] and Windows [2].

[1] https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Mac10_11__retina_/19070/layout-test-results.zip
[2] https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win7/56989/layout-test-results.zip
From https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=webkit_tests&tests=fast%2Fspatial-navigation%2Fsnav-container-white-space.html 

I'd like to bounce this back to the Mac/Win/test/input developers because I only got a Linux workstation myself and this test does not flake on Linux. Could it simply be that content_shell sometimes is slow to start on Mac/Win?

If I'd have a Mac/Win machine I'd loop this test locally, with some debug prints, to try to find out where the key gets lost. Maybe loosing the first eventSender.keyDown() is general source of LayoutTest flakiness on our Win/Mac bots?