RxJS - 为什么我要使用 throwError 而不是简单地抛出错误？

Question

First of all, I want to make it clear that I know the difference between Rx throwError operator and JS's throw keyword. I just wanted to know why would I use the throwError operator? What am I gaining by creating a new observable that all it does is to throw an error the second anyone subscribes to it?

Long story short, why would I want to do this:

.catchError(err => throw "error!!")

Over this:

.catchError(err => throwError("error!!"))

Thanks!

Answer 1

In this case, the syntax is the only difference. There is no reason to use throw instead of throwError or the other way around. Choose one approach with your team and stick to it.

Proof

// Error using reactive throwError
// logs error 'oops!!'
of(1)
    .pipe(
        switchMap(
           (x) => throwError('oops!!'))
        )
    .subscribe({
        next: console.log,  // not called
        error: console.error, 
        complete: () => console.log('complete!'),  // not called
    }); 

// Error using imperative throw
// logs error 'oops!!'
of(1)
    .pipe(
        switchMap((x) => {
            throw 'oops!!';
        })
    )
    .subscribe({
        next: console.log,  // not called
        error: console.error,
        complete: () => console.log('complete!'), // not called
    }); 

The output is the same in both ways. Note the completion handler which is not called in both scenarios, according to the Observable contract .

Answer 2

May be problematic to use throwError inside pipe operator as throwError observable may be just passed down the pipe depending on subsequent operations. See this article for details: https://medium.com/angular-in-depth/throwerror-is-not-throw-error-ad6c76c53377

Answer 3

If you just throw an error, it will end the pipeline. If you use throwError , it will return a new observable that is going to emit that error, essentially will pass it on down the pipeline. throwError(error) is equivalent to this:

return new Observable(observer=>{
      observer.error(error)
    })