fix(db): coerce async snapshot subscriptions with debounce operator

[byrondover]

Checklist

• Issue number for this PR: [email protected] Change Detection Irregularities #1583
• Docs included?: n/a
• Test units included?: yes
• In a clean directory, yarn install, yarn test run successfully? yes

Description

Swaps delay(0) in favor of debounceTime(0) to coerce async fromRef() database snapshot subscriptions. This prevents async listChanges() scans from burst emitting multiple copies of the same ref['on']('value') events.

@jamesdaniels Is this an acceptable / desirable change? It works well for me, but may have implications I'm not thinking of outside of standard RTDB use cases.

An alternative might be to revisit how listChanges() is implemented, perhaps by augmenting distinctUntilChanged() to debounce objects / arrays as well as immutable primitives.

Code sample

src/database/observable/fromRef.ts

export function fromRef(ref: DatabaseQuery, event: ListenEvent, listenType = 'on'): Observable<AngularFireAction<DatabaseSnapshot>> {
   ...
   // Ensures subscribe on observable is async. This handles
   // a quirk in the SDK where on/once callbacks can happen
   // synchronously.
-  .delay(0)
+  .debounceTime(0)
   .share();
 }

[jamesdaniels]

I have a suspicion on both this and the root cause of your slow down now; I'm going to close this while I think on it more + come up with an ultimate fix.

I think the .share got moved to a different place, leading to multiple listeners being created w/your usage patterns. While it's not changing the data it's leading to unnecessary change detection since we're slicing the array. distinctUntilChanged() is thinking in the right direction, but I think ultimately the correct thing to do is be more careful about using slice and creating a new array (which trips change detection, since === is false) when the snapshotChange is a no-op.

[tayambamwanza]

@byrondover That change fixed everything for me, I'll use it until another official fix is developed. Thank you so much you saved me a lot of hours 🤣 💐 !