Await result of Uni X for the duration of the execution of Uni Y

[technical-debt-collector]

We have an API in which we schedule several unis to execute simultaneously.
Some of these operations can may spend unacceptable durations in our API. In such cases we'd rather ignore their results than wait for them (these occasionally slow operations are unnecessary to the API operation itself, but allows for fail-fast and synchronous rejections whenever they execute fast enough).

Currently, we've solved this by chaining ifNoItem().after(SOME_DURATION).recoverWithUni(SOME_RESOLUTION), but this forces us to assign a fixed duration which is not ideal.

I have looked through existing discussions but haven't really found anything relevant, so I propose adding the functionality to execute a Uni X for as long as the execution duration of uni Y, resolving (or failing) the execution of X with something predefined whenever Y completes first.

This way we'd be able to spend the maximum time possible waiting for a result without ever incurring prolonged response times in our API (hopefully).

[jponge]

Mutiny is not doing the work of a scheduler.

That being said you may explore a combination of:

1. Uni.join().first(...) (the in-flight Uni will receive a cancellation signal when the first one completes)
2. .onCancellation() https://smallrye.io/smallrye-mutiny/2.3.1/tutorials/observing-events/#events to catch these events

[technical-debt-collector]

Revisiting this, I'm not sure if I fully agree with this being a scheduler's work. To clarify, my use of the term "scheduling" comes from various Quarkus documentation such as this (I might've interpreted it more broadly than intended).

What I'm asking for is essentially Uni.join().first() but instead of waiting for the first one to complete, we wait for a designated one and cancel the the other(s) if still in-flight.

I attempted to implement your recommendation but faced some challenges. Any further guidance would be appreciated.