Fix TS definition of publish & multicast

[hermanbanken]

Description:
Change TS definition of publish & multicast to support selectors changing the type.

Related issue (if exists):
#1905

[hermanbanken]

Currently working around like this...

declare module "rxjs/operator/publish" {
	export type selectors<T,R> = (source: Observable<T>) => Observable<R>;
	export function publish<T,R>(this: Observable<T>, selector: selectors<T,R>): Observable<R>;
}

[hermanbanken]

Checks are failing due to some conflicts in the Travis <-> Github API communication after I fixed the commit message and force pushed... /cc @kwonoj

[coveralls]

Coverage remained the same at 97.735% when pulling 9538f62 on hermanbanken:patch-1 into da96b6a on ReactiveX:master.

[kwonoj]

I'm feeling this might be legit case of <T, R=T> with #2614 maybe.

[kwonoj]

And second thought, isn't this PR causes breaking to existing consumers uses publish<T>?

[hermanbanken]

@kwonoj the following two types exist:

export function publish<T>(this: Observable<T>): ConnectableObservable<T>;
export function publish<T, R>(this: Observable<T>, selector: selector<T, R>): Observable<R>;

So:

• users of publish<T>() should not be affected. This will be the largest use case (e.g. publish & connect).
• users of publish<T>(selector) with a selector that does not change the type of the output would not be affected if they did not explicitly write .publish<T>( in their code. Typescript would automatically infer the new R type from the selector. If they did explicitly write <T> they will need to change that to <T,R>.
• users of publish<T>(selector) with a selector that does change the type can compile without errors/warnings, only by using my PR. Those people can remove any workarounds (e.g. as any).

[benlesh]

I think this just needs refactored for v6

[cartant]

@benlesh To me, it looks like this PR doesn't need to be merged. It adds overload signatures to multicast and to publish for selectors that change the observable's type. However, the current v6 codebase already includes said signatures - they appear to have been added in #2891.

[benlesh]

Ugh! Oh no! Am I closing 2 of @hermanbanken's PRs in one day? 😿

So sorry, @hermanbanken! 🙏

@cartant is right here, this work is already done.

[hermanbanken]

I’m just sad we didn’t get this in Rx 5 @benlesh, but since it’s in 6, closing this is just fine 😉 Op do 31 mei 2018 om 20:34 schreef Ben Lesh <notifications@github.com>

…

Closed #2616 <#2616>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#2616 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAwSlX4Q_AT2PLQj8lv9cwJ7OSY7xzkpks5t4DelgaJpZM4Nm71l> .