skipWhileRequested: reuse pending atm request

[strigefleur]

Case: you request /endpoint-foo then in a moment or so you manage to repeat same /endpoint-foo request.

What happens then? First request is still pending and is not cached by repo so you can't skipWhile it, thus second request is sent to backend freely; you end up with multiple responses.

What's the idea: do cache request observable for requestKey - and return it instead of new observable if request is pending.

Neither Cashew nor Elf cover this.

[NetanelBasal]

Cashew handles simultaneous requests

[strigefleur]

Yep, that's what I'm planning to do.

As per Cashew - it was not obvious by docs - do I get it right that I can expect to have single observable to subscribe to on 2+ simultaneous requests using Cashew?

[NetanelBasal]

Yes

[strigefleur]

I think it actually takes something like RequestsStore to provide the behavior I think about.

Just skipping similar pending request doesn't help with request tracking if there are more than a single subscriber.

[strigefleur]

Yes

Aha, sounds nice. Going to try now.

[strigefleur]

Well, I'd say it's not really what I tried to achieve:

http.get('...', {
  withCache(
    ...
    returnSource: ...
  ),
}).pipe(
  ...
  skipWhileCached(
    ...
    returnSource: ...
  )
)

where Cashew returnSource expects request object whereas Elf returnSource expects data.

I do understand there are different layers and mixing up is not really a good idea but in my mind it took a combined shape.