Daniel Urban

> The undecedable operation is really only safe in the context of an effect that is at most Concurrent+Sync, and not Temporal or Async. But we don't have a great...

Based on [this part](https://zio.dev/reference/stm/stm.md/#composing-alternatives) of the documentation, it probably _should_ work like haskell. And note, that for `orTry` it _does_. So this seems like a bug in `orElse`.

We could "fix" this, but then we'd have the opposite problem: the original `Future` would be `isCompleted` _strictly before_ the cancel `Future`. I wonder which is more intuitive? The "most...

Uhh, that definitely shouldn't hang. It's fine that the result is sometimes `Left`/`Right` (`race` is nondeterministic), but if it hangs, that's a bug.

@mperucca Could you provide the complete code that you see hanging? I tried, but couldn't get it to hang with the snippet above...

Thanks for trying it again. I was thinking about the original issue too, and I tend to agree, that the behavior you expected is the "more intuitive" one. Although, I'd...

It seems there are multiple things going on here. The essential thing is that when a _fatal_ exception is detected (which `InterruptedException` is), we're shutting down the whole world (meaning...

> Not entirely sure why that's not working in this instance. It's working. The second iteration of the while loop executes the IO on a different IORuntime. But then the...

> There are incorrect things I see in the behavior of CE with the code above: > - The IORuntime doesn't seem to shut down cleanly. This seems to be...

#### 1. I realize there is a bigger picture here, but I tried to make your first example (with the `server.accept()`) non-leaky. This is what I came up with. Does...