In order to use t.step_func in Promise chains it needs to pass-on the return value of its function argument. The current version of testharness.js in dom/imptests however doesn't do that. Now that bug 1089079 has been fixed I wonder if we can just update testharness.js and the related files using updateTestharness.py

I'll wait to run this through try before possibly putting it up for review (try is currently closed)

Try run: https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=afe87da8fcd6

/tests/dom/events/test/test_error_events.html is failing :(

Looks like that test used to stomp on testharness.js's window.onerror handler; now it uses addEventListener, you'll need to ensure the test sets the allow_uncaught_exception property (see <http://testthewebforward.org/docs/testharness-library.html>).

Comment on attachment 8528132 [details] [diff] [review] part 2 - Update dom/imptests/testharness.js and friends This patch is simply the result of running dom/imptests/updateTestharness.py roughly 24 hours ago. That, I believe, corresponds to this revision: https://github.com/w3c/testharness.js/tree/eeea5b74d5ea8530e8474511768872d1b70bb251 For my work, I currently only need the fixes to step_func and perhaps the cleanup_function facility too. However, it seems best to just update everything at once so that we don't have to go through this process for each and every feature one-by-one and merge them as needed. Currently running through try: https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=32e671024897

Comment on attachment 8528132 [details] [diff] [review] part 2 - Update dom/imptests/testharness.js and friends Review of attachment 8528132 [details] [diff] [review]: ----------------------------------------------------------------- rs=me assuming a plain pull from upstream and try green.

Unfortunately this turns bug 886301 into near perma-orange. Not sure why yet. Perhaps our explicit_timeout in testharnessreport.js is not being respected, or perhaps the updated testharness.js is slower and more likely to trigger our own timeout threshold.

Prior to updating testharness.js etc. this test would only intermittently timeout: https://tbpl.mozilla.org/?tree=Try&rev=e2ff8f27398e However, with the changes to testharness.js, it now fails almost permanently: https://tbpl.mozilla.org/?tree=Try&rev=079e87f71ad6 By extending the timeout for this test, it seems to run to completion: https://tbpl.mozilla.org/?tree=Try&rev=67e5b7837b82

Comment on attachment 8530063 [details] [diff] [review] part 3 - Extend the timeout of dom/imptests/editing/conformancetest/test_runtest.html Review of attachment 8530063 [details] [diff] [review]: ----------------------------------------------------------------- rs=me

https://hg.mozilla.org/integration/mozilla-inbound/rev/f8c5d91b1b3a https://hg.mozilla.org/integration/mozilla-inbound/rev/9ac3a114cc4e https://hg.mozilla.org/integration/mozilla-inbound/rev/9ce25fd6315b

https://hg.mozilla.org/mozilla-central/rev/f8c5d91b1b3a https://hg.mozilla.org/mozilla-central/rev/9ac3a114cc4e https://hg.mozilla.org/mozilla-central/rev/9ce25fd6315b