Struct Observer 

pub struct Observer<R> { /* private fields */ }

Available on crate feature client only.

Observes a resource and triggers re-renders only on status changes.

In v2, the observer only calls cx.notify() when the status actually changes, preventing excessive re-renders from intermediate state updates like retry count increments.

This is a single generic implementation shared by every resource kind (audit L8). Use the QueryObserver / InfiniteQueryObserver / MutationObserver type aliases for the concrete kinds.

This is also the fix for audit findings #1/#11: the raw cx.observe in use_mutation unconditionally called cx.notify() on every entity mutation, causing 2-3 re-renders per retry attempt. By tracking the last status and only notifying on change, increment_retry() / prepare_retry() calls (which don’t change status — it stays Loading) no longer trigger re-renders.

Implementations

impl<R: ObservableResource + 'static> Observer<R>

pub fn new(entity: &Entity<R>) -> Self

Create a new observer for the given entity.

pub fn with_config(self, config: ObserverConfig) -> Self

Set the observer configuration.

pub fn observe<W: 'static>( &self, cx: &mut Context<'_, W>, ) -> Option<Subscription>

Start observing the entity. Returns None if the entity was already dropped.

v2 fix: Returns Option<Subscription> instead of panicking.

Audit #71: takes &self (was &mut self) — the body only reads the weak entity handle and the Copy config flag, so no interior mutation is required. &mut callers coerce to & with no ripple.