Post

The Navigation API is newly baseline! The web now has sensible, low-level routing for navigations.

easrng 

@easrng.net

15 Jan

Is there a polyfill?

Firefox for Web Developers

@webdevs.firefox.com

15 Jan

I'm not aware of a good quality polyfill. I'm not sure it's possible to build one.

Kevin Crawford

@kcrwfrd.bsky.social

8 Feb

I released a polyfill with pretty good good support for navigation history entries. As I start to think about how to tackle navigation transition lifecycle (Navigate event, cancellation, interceptors, etc.) I was wondering if you have any wisdom to share in the parts you think are impossible? 🙏🏼

Firefox for Web Developers

@webdevs.firefox.com

Keeping the browser loading spinner going pops to mind, but there may be more. Things like knowing a navigation is just a `hashChange` seems tricky in the case of traversals. As in, how do you know it's same-document or not?

9/2/26, 10:15 am0 reposts0 likes

easrng 

@easrng.net

9 Feb

>Keeping the browser loading spinner going I think this could be done with serviceworkers/iframes maybe?

Firefox for Web Developers

@webdevs.firefox.com

9 Feb

maybe! It could be that saying this couldn't be polyfilled was a total lack of imagination on my part.

Kevin Crawford

@kcrwfrd.bsky.social

9 Feb

Yeah, hash changes were tricky to handle, but I think I got it down. I was surprised to learn they fire a popstate event! All of my e2e tests around history entries work with both native Navigation or the polyfill so far.

Kevin Crawford

@kcrwfrd.bsky.social

9 Feb

I believe sameDocument is indeed impossible. As well as navigationType: ‘reload’ on entry change event. The polyfill also introduces some restrictions on history API calls since we have to augment the state object with our entry key and ID.

Kevin Crawford

@kcrwfrd.bsky.social

9 Feb

So the browser spinner is shown while interception handler resolves? That’s pretty neat, thankfully I think its omission is some acceptable graceful degradation. What I’m wondering is can we polyfill sufficiently enough that a navigation-powered router for a SPA can be implemented.