I'm trying detect and fix potential memory leaks in my angular/spartacus application. So basically I'm looking for any "subscribe(..." calls in my code base and evaluate what I can do. I know that simplest form is simply to use observables directly in html and read them by using the async-pipe instead of calling the subscribe method and assigning the value to a component-class-attribute. But sometimes it's not that simple. Sometimes the data for the observable has to be prepared somehow. My question is if these approaches are just as valid? Is it a good Idea to assign ANY "subscribe()" call to a Subscription object, which I will eventually unsubscribe from in an "ngOnDestroy"-method Why can't I instead of using a Subscription object call instead the "unsubscribe"-method right after the "subscribe" call? MOST IMPORTANT: Is there any benefit in using the "pipe()" operator, to receive data, manipulate it with SwitchMap and then return it to the "obs$"-object, that I can read in html via async pipe. Are there any hidden Subscriptions tied to a "pipe(...)"-process that I have to manually unsubscribe from, or is it indeed always the preferable approach compared to explicitly subscribing and then assigning the observed object to a component-class-attribute? Example: instead of this mergedReferences$: Observable<Observable<Product>[]>; subscription = new Subscription(); ngOnInit() { this.subscription.add(this.cart$.subscribe(c => { if (c.code) { this.mergedReferences$ = this.getMergedProducts(c.code); } })); } ngOnDestroy(): void { this.subscription?.unsubscribe(); } doing simply this mergedReferences$: Observable<Observable<Product>[]>; ngOnInit() { this.mergedReferences$ = this.cart$.pipe(switchMap((c: Cart) => this.getMergedProducts(c.code!))) } Continue reading...
