foolip, could you help us triage this? Have we imported this test? What does the flakiness dashboard say about it?

This was a test that was flaky when running on a Chrome release binary as part of the Travis jobs on web-platform-tests. The tests in question:
http://w3c-test.org/html/browsers/the-window-object/apis-for-creating-and-navigating-browsing-contexts-by-name/

open-features-non-integer-height.html, open-features-non-integer-innerheight.html and others can be seen to have been flaky in https://github.com/w3c/web-platform-tests/pull/5390#issuecomment-291969797

Like Simon, I can't reproduce on macOS. The failure was on Linux, but I also can't produce on Linux, using 59.0.3067, exactly the version that was run on Travis.

The imported tests are here:
https://cs.chromium.org/chromium/src/third_party/WebKit/LayoutTests/external/wpt/html/browsers/the-window-object/apis-for-creating-and-navigating-browsing-contexts-by-name/

Nothing interesting shows up on the flakiess dashboard:
https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=webkit_tests&tests=apis-for-creating-and-navigating-browsing-contexts-by-name

The errors reported were "expected 405 but got 366" and similar, and don't look like failures to parse the string correctly, but perhaps the window manager changing the size of the window, or the size incorrectly including/excluding the UI bits.

Simon, without repro steps I don't think much can be done here, and none of us can repro. If you want to dig deeper, maybe create dummy PRs for wpt that change whitespace in these files and see if you can get another error. If you can, ask @bobholt and @jugglinmike if they can tell what's actually going wrong.

For now, I'll close this as can't reproduce, which is WontFix in Monorail.

@bobholt could reproduce flakiness for open-features-non-integer-height.html; from #testing


zcorpan
bobholt: FYI https://bugs.chromium.org/p/chromium/issues/detail?id=710869#c2 - this test showed instability in wpt stability checker but appears stable per flakiness dashboard. I'll not investigate further myself but I guess if this sort of thing happens again we can try to figure out what's going on...
bobholt
zcorpan: not sure. i'm running those tests locally - the ones in the bug report aren't flaky, but others are
bobholt
all the `/404`, `_404`, `L404` ones are flakey on my linux Chrome unstable 59.0.3067.0
bobholt
but the 405s pass
zcorpan
bobholt: interesting. are those asserting top/left ?
bobholt
assert_equals: "height=/404 begins with an invalid character and should be ignored" expected 1035 but got 999
bobholt
they're the definition of flakey - sometimes they all pass, sometimes they all fail, and sometimes it's 1 or 2 of the 3
bobholt
zcorpan: note i only looked at /html/browsers/the-window-object/apis-for-creating-and-navigating-browsing-contexts-by-name/open-features-non-integer-height.html, not the other files
zcorpan
bobholt: OK. Good I guess that you can reproduce it...