Ensure focusin/focusout fires with .focus()/.blur()

[trueadm]

This PR alters how HTMLOrSVGElement-impl so that the focus and blur methods also dispatch the relevant focusin and focusout events in the bubble phase, as per how they occur on modern browsers. I wasn't sure where to add any tests for this change specifically, if anyone has any pointers, that would be great.

[eps1lon]

I wasn't sure where to add any tests for this change specifically, if anyone has any pointers, that would be great.

diff --git a/test/web-platform-tests/to-run.yaml b/test/web-platform-tests/to-run.yaml
index c2cff296..3f08f075 100644
--- a/test/web-platform-tests/to-run.yaml
+++ b/test/web-platform-tests/to-run.yaml
@@ -1006,7 +1006,7 @@ idlharness.window.html: [fail, Depends on fetch]
 legacy-domevents-tests/approved/ProcessingInstruction.DOMCharacterDataModified.html: [timeout, Mutation Events not implemented]
 legacy-domevents-tests/approved/domnodeinserted.html: [timeout, Mutation Events not implemented]
 mouse/**: [timeout, Uses testdriver.js]
-order-of-events/**: [timeout, Unknown]
+order-of-events/mouse-events/*: [timeout, Unknown]
 
 ---
 

which cause yarn test-wpt to test test/web-platform-tests/tests/uievents/order-of-events/focus-events/focus-automated-blink-webkit.html. Though this is failing on this branch with Maximum callstack size exceeded.

I guess you could add a smaller test to test/web-platform-tests/to-upstream/uievents to check (with yarn test-tuwpt if the test would require more work or if the current implementation is problematic.

Also: This PR closes #2708

[trueadm]

Thanks for the help @eps1lon. I found focusing-focused-noop.html and updated that to include the logic from this test as it seemed the most reasonable approach.

[domenic]

Can we make sure that this doesn't cause infinite recursion, by enabling the test as @eps1lon suggests and then making it pass, instead of making it fail with Maximum callstack size exceeded?

[trueadm]

@domenic Yep, I plan on looking into that later once I get some time to. :)

I was looking into why my changes caused the build to fail with:

/home/travis/build/jsdom/jsdom/test/web-platform-tests/tuwpt-manifest.jso

[trueadm]

I've fixed the failing test. I ended up spending quite a bit longer on it, but using a build of Chrome and the test to cover the same sequence of how activeElement and relatedTarget get set, especially around iframes.

[trueadm]

@domenic Are you happy with this PR or is there something you'd like me to look into still?

[domenic]

Can you enable the tests @eps1lon suggests?

[trueadm]

@domenic Ah sorry, I thought that mention @eps1lon was in reference to enabling specific tests to run, not exclude. I've commited that change to this PR now too.

[trueadm]

The latest failures seem to be due to #2951 being merged on master.

Looking into why focus-management-expectations.html is failing, it seems to be with some other focus related logic, rather than something introduced in this PR.

[trueadm]

@domenic Would it be possible to get a new release of JSDOM? We want to incorporate this change into the next version of React and thus ensure our Jest test suite is able to pass with the changes to React's event system. We can upgrade Jest once JSDOM releases a new version. :)

[domenic]

For sure; I'll definitely do a release by this upcoming weekend, but hopefully can get it out sooner.

[trueadm]

Thank you so much! I really appreciate it :)

[domenic]

FYI I published v16.3.0 today.

[rickhanlonii]

Thanks @domenic!