If you use unload handlers, then you should check out this potential change in Chrome.
Unload handlers are VERY unreliable but on desktop they also prevent the use of the bfcache on Chrome (and Firefox), while on mobile they don't. We want to align it to how it works on mobile.
https://groups.google.com/a/chromium.org/g/blink-dev/c/oU1yt5cdGH8
The bfcache allows for instant page navigations when going back or forwards and is one of our key recommendations to help you pass Core Web Vitals:
https://web.dev/top-cwv-2023/#ensure-pages-are-eligible-for-bfcache
So, aligning this to the mobile implementation will improve web performance for desktop.
The `pagehide` event is basically the same as the `unload` event except it does NOT restrict use of the bfcache.
It's also unreliable, but at least doesn't prevent the bfcache being used.
The proposed change will basically make `unload` act pretty much like `pagehide`.
We've long advised using the `pagehide` event (when the page is navigated away) and `visibilitychange` (when a page is backgrounded) over `unload`:
https://web.dev/bfcache/#never-use-the-unload-event
Many sites and libraries have moved to using those, but we're reaching a long tail of code that hasn't.
Longer term, the Pending Beacon API hopes to add a RELIABLE way of sending a beacon as the page is discarded, rather than depending on unreliable APIs or visibility changes as a proxy for that.
But, back to the `unload` handler proposed change. Do let us know if you see any problems with this proposal, or have any other feedback:
https://groups.google.com/a/chromium.org/g/blink-dev/c/oU1yt5cdGH8
Let me know if you have any questions!
Barry Pollard
Kenji Baheux has moved
