fix(subscriber): strict type signature for next method

[raveclassic]

Description:

This PR fixes type signature of Subscriber.prototype.next method that was previously taking optional argument which was violating generic constraint T.

BREAKING CHANGE: Subscriber.prototype.next now takes strict argument.

new Observable(subscriber => {
  subscriber.next() // throws in compile time (expected, caused by this PR)
})

new Observable<string>(subscriber => {
  subscriber.next() // throws in compile time (expected, caused by this PR)
})

new Observable<void>(subscriber => {
  subscriber.next() // ok
})

Related issue (if exists): #2852

Additional changes:

• update package version in lockfile to 7.8.0 (running npm i)

[benlesh]

Yay! I almost forgot we were waiting for v8 for this one.

[raveclassic]

Awesome! Thanks for approving!
Seems I have to fix some things, missed that. Will come back to this PR when I return back to my laptop in a few days 🌴

[benlesh]

Uh oh. We have broken tests.

[raveclassic]

@benlesh tests are fixed 🤌

[raveclassic]

@benlesh could you review, please?

[raveclassic]

@benlesh just a friendly reminder ;) would love this to land in rxjs@7 as now our codebase heavily relies on new Observable(subscriber => {}) notation and we'd love to have this fixed 🙏

[raveclassic]

Folks? :D @benlesh @LcsGa

[LcsGa]

Hello! Sorry, I can't do more I'm not in the team 😅

[jakovljevic-mladen]

@benlesh, is this merged to 7.x by mistake or on purpose?

[raveclassic]

@jakovljevic-mladen @benlesh I expected this to land on 7.x 🤔

[benlesh]

@jakovljevic-mladen honestly, I didn't notice, but I'm generally okay with this as a "fix" in 7.x. I'll go back and have a look to double check, but I think this is a valid bug fix, as people should not be calling next() on a Subscriber<number> or something. Type safety is important.