Reactive system panicking at `already borrowed: BorrowMutError` about leptos HOT 10 CLOSED

Hee it is @gbj

use leptos::*;
use leptos_router::*;
use wasm_bindgen::prelude::*;

#[wasm_bindgen(start)]
pub fn start() {
  console_error_panic_hook::set_once();

  mount_to_body(view_fn);
}

fn view_fn(cx: Scope) -> impl Mountable {
  provide_context(cx, RouterIntegrationContext::new(BrowserIntegration {}));

  view! { cx,
    <span>
      <Router>
        <span>  // <-- won't compile whithout these
          <Routes>
            <Route path="/" element=|cx| view! { cx, <Home /> } />
            <Route path="/found-another" element=|cx| view! { cx, <h1>"bug!"</h1> } />
          </Routes>
        </span> // <-- won't compile whithout these
      </Router>
    </span>
  }
}

#[component]
fn Home(cx: Scope) -> Element {
  use_navigate(cx)("/found-another", Default::default()).unwrap();

  view! { cx,
    <h1>"/"</h1>
  }
}

from leptos.

Thanks. To be clear, if you remove that uae_navigate does it still panic?

from leptos.

The workaround I am currently using is to just schedule the navigation in the next microtask using wasm_bindgen_futures::spawn_local, which runs after whatever was borrowed releases the lock.

from leptos.

The use_navigate() does not panic, but calling the fn it returns does.

from leptos.

Thanks. This is the kind of thing the reactive system doesn't like: rendering the app requires reactively reading the current route, which synchronously renders the component, which updates the current route... such that you're mutating the value of the route signal within an effect that depends on the route signal.

I've basically moved your solution into the library, here, using queue_microtask rather than spawn_local for any navigation.

from leptos.

Is this something that should potentially have a disclaimer for authors creating primitive utilities using leptos_reactive? Or is this a situation that the average user might find themselves in while using signals and effects?

I personally ran into it out of trying to create a Redirect component which redirects the user elsewhere if the component renders, which currently does not exist in the leptos_router crate. All other uses of signals and effects I've used haven't triggered this panic yet.

from leptos.

Yeah you know it's actually going to take a more general solution: I actually ran into the same thing in the TodoMVC example this morning. It's pretty easy to accidentally create these loops. I'll try to solve it without the user land/library land hack.

from leptos.

Hi, I just ran into this today, on 0.3.1. Is it supposed to be fixed or is a solution still in the works?

from leptos.

@nguiard The reactive system has been almost completely rewritten since this issue. I'm quite sure the panic in your case was caused by something completely different from this old issue. Could you please open a new issue with a reproduction? Otherwise it won't be possible to help you.

EDIT: Sorry, I may have misunderstood for a second. Did you run into the same issue with use_navigate?

from leptos.

I wasn't specific enough, my bad. You are correct, the issue was caused by something completely different. I'll open an issue to describe it a bit and try to make a minimal reproduction. Thanks.

from leptos.