move ._handler off state

[yoshuawuyts]

State should be serializable; by moving ._handler off state we keep state a bit cleaner. Thanks!

[tornqvist]

Looks good! 👍

[bcomnes]

Didn't even know it was there.

[goto-bus-stop]

this makes sense to me but I wonder what @marcbachmann thinks because there might've been a reason for doing it this way in #613

[marcbachmann]

So far the state variable was the only one which was route-specific and mutable, which was the reason I put it there. I'm good with that change. Tests for that would be good as we'll probably run into the same issue again.

[yoshuawuyts]

@marcbachmann just to have some context: what issue were you running into? Agree we should have some tests for this; want to make sure we test for the right things.

[marcbachmann]

what issue were you running into?

I was referencing to the statement in your issue description. (State should be serializable).
I had no idea that we wanted to have that state object serializable. How do people use it and do we depend on that?

[tornqvist]

How do people use it and do we depend on that?

I've been experimenting with persisting state to a service worker (for offline support and shared state between tabs) but if you try to post a message with anything but plain serializable values it'll throw.

But yeah, we should probably document the desire to keep state serializable.

[yoshuawuyts]

But yeah, we should probably document the desire to keep state serializable.

yeah, we should probably write a test for that too. I'll do that in a separate patch.

[yoshuawuyts]

v6.10.3