Fix event path iteration #686
Merged
+41
−36
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
annevk
added
topic: shadow
|
@whatwg/components I think this is what we want here, but careful review appreciated. If I did it correctly this does seem like a much nicer model. |
annevk
commented
Sep 6, 2018
| <a for=Event/path>touch target list</a> is a <a for=/>node</a> and its <a for=tree>root</a> is a | ||
| <a for=/>shadow root</a>, and false otherwise. | ||
|
|
||
| <li><p>Set <var>event</var>'s {{Event/eventPhase}} attribute to {{Event/CAPTURING_PHASE}}. |
There was a problem hiding this comment.
As far as I can tell "legacy-pre-activation behavior" cannot trigger side effects so setting this later is not observable.
There was a problem hiding this comment.
Actually, it looks like we fire change event on radio buttons & checkboxes so the event phase change is observable. Gecko follows the current spec and WebKit & Blink uses the new behavior.
https://gist.github.com/rniwa/6c502dca3e16d5816db7958ce7bab4d7
There was a problem hiding this comment.
Thanks, we should add that test to the test suite. None seems like a reasonable phase to me given that traversal hasn't started yet, WDYT @smaug----?
There was a problem hiding this comment.
Agreed that none seems more reasonable than capturing given the event dispatching hasn't even started yet.
There was a problem hiding this comment.
From API user's point of view Gecko's behavior makes more sense, since for them the event dispatch clearly has started already once dispatchEvent is called.
There was a problem hiding this comment.
I disagree. Just because we're in dispatchEvent, doesn't mean eventPhase needs to be something other than "none".
|
The change looks sensible to me. |
As pointed out in #685 this was a bit confusing.
annevk
force-pushed
the
annevk/listener-calling-order
branch
from
30085f8
to
39b97c3
Compare
kisg
pushed a commit
to paul99/webkit-mips
that referenced
this pull request
Sep 14, 2018
… hosts https://bugs.webkit.org/show_bug.cgi?id=174288 LayoutTests/imported/w3c: Reviewed by Darin Adler. * web-platform-tests/dom/events/Event-dispatch-handlers-changed-expected.txt: Rebaselined. This test's expectation is now wrong because event listner 3 is added after the event listener list is cloned for capturing event listeners but before cloned for bubbling event listeners. As a result, event listener 3 is now invoked. It used to be not called because both bubbling and capturing event listeners are called after cloning the event listner list once, which didn't have event listener 3. * web-platform-tests/dom/events/EventTarget-dispatchEvent-expected.txt: Rebaselined. This test expects event listener 2, which is bubbling, to be called between two capturing event listeners 1 and 3, which is no longer true after this patch. Source/WebCore: <rdar://problem/33530455> Reviewed by Darin Adler. Implemented the new behavior proposed in whatwg/dom#686 [1] to fix the problem that capturing event listeners on a shadow host is invoked during bubbling phase when an event is dispatched within its shadow tree. To see why this is a problem, suppose we fire a composed event at span#target in the following DOM tree: section#hostParent + div#host -- ShadowRoot - p#parent - span#target Then capturing and bubbling event listeners on #target, #parent, #host, and #hostParent are invoked in the following order in WebKit & Chrome right now: 1. #hostParent, capturing, eventPhase: CAPTURING_PHASE 2. #parent, capturing, eventPhase: CAPTURING_PHASE 3. #target, capturing, eventPhase: AT_TARGET 4. #target, non-capturing, eventPhase: AT_TARGET 5. #parent, non-capturing, eventPhase: BUBBLING_PHASE 6. #host, capturing, eventPhase: AT_TARGET 7. #host, non-capturing, eventPhase: AT_TARGET 8. #hostParent, non-capturing, eventPhase: BUBBLING_PHASE This is counter-intuitive because capturing event listeners on #host isn't invoked until bubblign phase started. A more natural ordering would be: 1. #hostParent, capturing, eventPhase: CAPTURING_PHASE 2. #host, capturing, eventPhase: AT_TARGET 3. #parent, capturing, eventPhase: CAPTURING_PHASE 4. #target, capturing, eventPhase: AT_TARGET 5. #target, non-capturing, eventPhase: AT_TARGET 6. #parent, non-capturing, eventPhase: BUBBLING_PHASE 7. #host, non-capturing, eventPhase: AT_TARGET 8. #hostParent, non-capturing, eventPhase: BUBBLING_PHASE This also happens to be the order by which Gecko's current shadow DOM implementation invoke event listners. This patch implements this new behavior using the spec-change proposed in [1]. Note that this patch also impacts the invocation order of event listeners when there is no shadow tree. Namely, before this patch, event listeners on the event's target is invoked in the registration order. After this patch, all capturing event listeners are invoked before bubbling event listeners are invoked. To implement this behavior, this patch introduces EventTarget::EventInvokePhase indicating whether we're in the capturing phase or bubbling phase to EventTarget::fireEventListeners. We can't use Event's eventPhase enum because that's set to Event::Phase::AT_TARGET when we're at a shadow host. Test: fast/shadow-dom/capturing-and-bubbling-event-listeners-across-shadow-trees.html * Modules/modern-media-controls/media/media-controller-support.js: (MediaControllerSupport.prototype.enable): Use capturing event listeners so that we can update the states of media controls before author scripts recieve the event. (MediaControllerSupport.prototype.disable): Ditto. * dom/EventContext.cpp: (WebCore::EventContext::handleLocalEvents const): (WebCore::MouseOrFocusEventContext::handleLocalEvents const): (WebCore::TouchEventContext::handleLocalEvents const): * dom/EventContext.h: * dom/EventDispatcher.cpp: (WebCore::dispatchEventInDOM): Invoke capturing event listners even when target and current target are same. This happens when the current target is a shadow host and event's target is in its shadow tree. Also merged the special code path for the event's target with the code in the bubbling phase. * dom/EventPath.cpp: (WebCore::WindowEventContext::handleLocalEvents const): * dom/EventTarget.cpp: (WebCore::EventTarget::dispatchEvent): Invoke capturing and bubbling event listeners in the order. (WebCore::EventTarget::fireEventListeners): (WebCore::EventTarget::innerInvokeEventListeners): Renamed from fireEventListeners to match the spec. Use EventInvokePhase to filter out event listeners so that we can invoke capturing event listners before bubbling event listeners even when eventPhase is Event::Phase::AT_TARGET. * dom/EventTarget.h: * dom/Node.cpp: (WebCore::Node::handleLocalEvents): * dom/Node.h: * html/HTMLFormElement.cpp: (WebCore::HTMLFormElement::handleLocalEvents): * html/HTMLFormElement.h: * page/DOMWindow.cpp: (WebCore::DOMWindow::dispatchEvent): LayoutTests: Reviewed by Darin Adler. Added a W3C style testharness.js test and rebaselined two tests. See below for rationals of rebaselines. * fast/shadow-dom/capturing-and-bubbling-event-listeners-across-shadow-trees-expected.txt: Added. * fast/shadow-dom/capturing-and-bubbling-event-listeners-across-shadow-trees.html: Added. * media/media-load-event-expected.txt: Rebaselined. The logging of oncanplaythrough event is now happening before canplaythrough() is called because the logging is done by waitForEvent which uses a capturing event listener whereas canplaythrough is called by a event handler, which is non-capturing. * platform/ios-11/imported/w3c/web-platform-tests/dom/events/EventTarget-dispatchEvent-expected.txt: * platform/ios/imported/w3c/web-platform-tests/dom/events/EventTarget-dispatchEvent-expected.txt: git-svn-id: http://svn.webkit.org/repository/webkit/trunk@236002 268f45cc-cd09-0410-ab3c-d52691b4dbfc
annevk
added a commit
to web-platform-tests/wpt
that referenced
this pull request
Sep 17, 2018
annevk
added a commit
to web-platform-tests/wpt
that referenced
this pull request
Sep 17, 2018
This was referenced Sep 17, 2018
annevk
added a commit
to web-platform-tests/wpt
that referenced
this pull request
Mar 1, 2019
annevk
removed
the
needs tests
annevk
added a commit
to web-platform-tests/wpt
that referenced
this pull request
Mar 1, 2019
annevk
force-pushed
the
annevk/listener-calling-order
branch
from
4e0c64d
to
794e934
Compare
Instead of shoehorning all target handling into the bubbling iteration, this separates "capturing" iteration from "bubbling" iteration and the Event object's phase is set to target as appropriate in both. This also invokes the event listeners in a more natural order. Tests: web-platform-tests/wpt#13030 & web-platform-tests/wpt#13031. Fixes #685.
annevk
force-pushed
the
annevk/listener-calling-order
branch
from
794e934
to
7aaff50
Compare
annevk
added a commit
to web-platform-tests/wpt
that referenced
this pull request
Mar 1, 2019
annevk
added a commit
to web-platform-tests/wpt
that referenced
this pull request
Mar 2, 2019
moz-v2v-gh
pushed a commit
to mozilla/gecko-dev
that referenced
this pull request
Apr 1, 2019
…estonly Automatic update from web-platform-tests DOM: test listener invocation order For whatwg/dom#686. -- wpt-commits: 103df1bf782674889deb7136cdf7b1318dadb86b wpt-pr: 13031
moz-v2v-gh
pushed a commit
to mozilla/gecko-dev
that referenced
this pull request
Apr 1, 2019
…tion behavior, a=testonly Automatic update from web-platform-tests DOM: eventPhase during legacy-pre-activation behavior For whatwg/dom#686. -- wpt-commits: 079928e6d608251333281d1a934f50d01ffeb901 wpt-pr: 13030
This was referenced Oct 14, 2019
Closed
This was referenced Oct 23, 2019
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Instead of shoehorning all target handling into the bubbling iteration, this separates "capturing" iteration from "bubbling" iteration and the Event object's phase is set to target as appropriate in both.
This also invokes the event listeners in a more natural order.
Tests: ...
Fixes #685.
Preview | Diff