1891259 - event.preventDefault() is not respected when called in microtask

Status: REOPENED

(Core :: DOM: Events, defect)

Description

[nick.gard.dev]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:126.0) Gecko/20100101 Firefox/126.0

Steps to reproduce:

There is a minimal test case on this pen: https://codepen.io/NickGard/pen/mdgLoZv?editors=1011

The code is:

<input id="X" is="dumb-thing">
<script>
X.addEventListener("beforeinput", (e) => {
  console.log("beforeinput");
  // when called here, preventDefault() correctly prevents "input" event from dispatching
  // event.preventDefault();

  queueMicrotask(() => {
    // this is too late to call preventDefault() in Firefox.
    // Safari and Chrome still prevent the "input" event from dispatching when called here
    event.preventDefault();
  });
});
X.addEventListener("input", (e) => {
  console.log("input");
});
</script>

Actual results:

The "input" event handler is still triggered, logging "input" in the console.

Expected results:

The "input" event handler should not have been triggered.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

The Bugbug bot thinks this bug should belong to the 'Core::DOM: Events' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

I think the issue is this https://searchfox.org/mozilla-central/rev/d23849dd6d83edbe681d3b4828700256ea34a654/dom/events/EventListenerManager.cpp#1302
We should remove that.

Testing a bit

Comment 3

[nick.gard.dev]

After exploring this a bit more with other events' default behaviors, all of the ones I've tested are correctly prevented from occurring when event.preventDefault() is called in a microtask. For instance, the following code correctly stops characters from being entered into the input and precludes the "beforeinput" and "input" events from dispatching.

<input id="X">
<script>
  X.addEventListener('keydown', event => {
    queueMicrotask(() => {
      event.preventDefault();
    });
  });
</script>

It seems that "beforeinput" is the only one that doesn't work in a microtask.

Comment 4

[Olli Pettay [:smaug][[e-mail]]]

Right, because I assume Gecko dispatches beforeinput inside another (C++) event listener.

Comment 5

[Olli Pettay [:smaug][[e-mail]]]

Confirmed, it is about https://searchfox.org/mozilla-central/rev/d23849dd6d83edbe681d3b4828700256ea34a654/dom/events/EventListenerManager.cpp#1302
Pushing to try to see what removing that legacy code breaks.
https://treeherder.mozilla.org/jobs?repo=try&revision=2178e650959eae709ef532c99c85d36e02382485

Comment 6

[Olli Pettay [:smaug][[e-mail]]]

Attachment: Bug 1891259 - beforeInputEvent.preventDefault() is not respected when called in microtask, r=edgar

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/13c25f58aaca
beforeInputEvent.preventDefault() is not respected when called in microtask, r=edgar

Comment 8

[Aron Cseh]

Backed out for causing xpcshell failures on test_ext_scripting_contentScripts.js

• Backout link
• Push with failures
• Failure Log

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ebcc80d71094
beforeInputEvent.preventDefault() is not respected when called in microtask, r=edgar

Comment 10

[Sandor Molnar[:smolnar]]

https://hg.mozilla.org/mozilla-central/rev/ebcc80d71094

Comment 11

[Andrew Overholt [:overholt]]

Given bug 1893226, should we back this out until we can sort that regression out?

Comment 12

[Sandor Molnar[:smolnar]]

Backed out for breaking handful of things on slack

Backout link: https://hg.mozilla.org/mozilla-central/rev/2d59d3cfc93fa569342ae78098d774b23b75c810

Comment 13

[BugBot [:suhaib / :marco/ :calixte]]

There is an r+ patch which didn't land and no activity in this bug for 2 weeks.
:smaug, could you have a look please?
If you still have some work to do, you can add an action "Plan Changes" in Phabricator.
For more information, please visit BugBot documentation.

Comment 14

[nick.gard.dev]

Has there been any progress on fixing this in a way that doesn't cause a regression like the one in Slack?

Comment 15

[Olli Pettay [:smaug][[e-mail]]]

I've been rather busy with other things, but this is very high up in my todo list.

Comment 16

[Theodore Brown]

Is there any update on fixing this? This has broken oninput state modifications in Svelte 5 for us. Compare https://svelte.dev/playground/bf28bf223e4746a199f15ca80385487d?version=5.39.2 in Firefox vs. Chrome or Edge.