addEventListener - which browsers/events default to passive

[redblobgames]

MDN URL

https://developer.mozilla.org/en-US/docs/Web/API/EventTarget/addEventListener

What specific section or headline is this issue about?

Improving scrolling performance with passive listeners

Documentation of which browsers default to passive

What information was incorrect, unhelpful, or incomplete?

The page says "browsers other than Safari and Internet Explorer" default certain events to passive, suggesting Chrome and Firefox do this. But it seems to be Chrome and Safari that default those events to passive.

This is also mentioned in the description of the passive flag.

What did you expect to see?

I expect it to say "Chrome and Safari change the default value of the passive option to true for {wheel, mousewheel, touchstart, touchmove}".

Do you have any supporting links, references, or citations?

I looked up the history of the touch events defaulting to passive:

1. 2017 https://developer.chrome.com/blog/scrolling-intervention/ Chrome says they will default touchstart, touchmove to passive
2. 2018 Touch listeners defaulting to passive WICG/interventions#18 (comment) Safari defaults touch events to passive

I've also tested the touchstart event myself to see if .cancelable is set, and

• iOS 16.3/Safari: it's false (implying passive)
• Android/Chrome 109: it's false (implying passive)
• Android/Firefox 109: it's true (implying not passive)

Separately, wheel events:

1. 2019 https://developer.chrome.com/blog/scrolling-intervention-2/ Chrome says they will default wheel events to passive, but also says that Firefox implemented the touchstart passive already (which doesn't match my tests)

I tested the wheel event to see if .cancelable is set, and

• Mac/Safari 16.3: it's false (passive)
• Mac/Chrome 109: it's false (passive)
• Mac/Firefox 109: it's true (not passive)

So here too it seems to be Chrome and Safari that default to passive, not "browsers other than Safari and Internet Explorer".

Do you have anything more you want to share?

I'm happy to make a pull request but thought I should file an issue first.

MDN metadata

Page report details

• Folder: en-us/web/api/eventtarget/addeventlistener
• MDN URL: https://developer.mozilla.org/en-US/docs/Web/API/EventTarget/addEventListener
• GitHub URL: https://github.com/mdn/content/blob/main/files/en-us/web/api/eventtarget/addeventlistener/index.md
• Last commit: a122e87
• Document last modified: 2022-09-14T06:08:23.000Z

[wbamberg]

Thanks for filing, @redblobgames , and for your analysis.

It's better for maintainability for us to minimize references to browser support in the main body of the page, and refer to the browser compat tables for that. So as for the prose of this article, I think it would be better to say something like:

To prevent that problem, browsers other than Safari and Internet Explorer have changed the default value of the passive option ... For details of browser support for this behavior, see the browser compatibility table.

But from your description it also sounds like the BCD table is wrong? The table claims everyone except Safari supports this behavior, but your description suggests that it should be everyone except Firefox.

Having said that it does look like this was implemented in Firefox, e.g.:
https://bugzilla.mozilla.org/show_bug.cgi?id=1449268
https://bugzilla.mozilla.org/show_bug.cgi?id=1673278

[redblobgames]

I agree, browser compat tables are a better place for this. I had thought Firefox didn't implement the default passive but I was wrong. I ran some more tests here.

I had been testing a proxy for active/passive — the cancelable property on Event. This "should" tell me if it's passive, but it doesn't. I ran my tests again, this time attempting to .preventDefault() to see if I was able to prevent scrolling. Firefox does not allow me to .preventDefault(), but the event is marked .cancelable = true. Related bug

There's also web platform tests and more but I am not sure how to interpret these

I am thinking we should change the BCD table to mark Safari as supporting "options.passive parameter defaults to true". It looks like every current browser now behaves this way. Internet Explorer does not but it's no longer in the BCD.

Then we could remove the "browsers other than Safari and Internet Explorer have changed the default value…" text to say "browsers have changed the default value…" (or maybe "current browsers"? "modern browsers"? not sure what to say here)

[wbamberg]

I am thinking we should change the BCD table to mark Safari as supporting "options.passive parameter defaults to true". It looks like every current browser now behaves this way. Internet Explorer does not but it's no longer in the BCD.

Then we could remove the "browsers other than Safari and Internet Explorer have changed the default value…" text to say "browsers have changed the default value…" (or maybe "current browsers"? "modern browsers"? not sure what to say here)

We usually say "modern browsers" here. And yes, I think this is right:

• update BCD to fix Safari behavior
• update the page:
  • change "browsers other than Safari and Internet Explorer" to something like "modern browsers set passive to true for the wheel ..."
  • delete the note
  • remove "in all browsers" in the next paragraph.

Do you want to contribute these fixes? If not, no problem, we will take care of it.

[redblobgames]

While I'm happy to contribute a pull request (or maybe two, as the compat tables seem to be in a separate repository), I'm even happier if someone else does it ;-)

[estelle]

This issue seems to have been addressed. @redblobgames @wbamberg Is it OK to close?

[wbamberg]

No, this isn't fixed, it looks like I was going to add it into #29083 but it got merged before I got around to it. But I will take care of it now.

[wbamberg]

Ah, but it looks like the BCD did get updated: mdn/browser-compat-data#21523 . Yay.