Eliza Weisman

With that said, we probably _could_ fix this by having `with_current_subscriber` check if the current subscriber is `NoSubscriber`, and if it is, having its future do nothing, rather than setting...

Hi @kaffarell, thanks for the fix! I believe that this issue also exists in the documentation on the `master` branch. Would you mind changing this PR to be based on...

Hi @mladedav , thanks for the fix! Would you mind checking whether these tests are also ignored on the`master` branch, and if so, changing this PR to be based on...

yeah, this is a known issue and...it's really unfortunate. we're just not going to add new levels ever, i guess. which, honestly, we probably weren't going to do in the...

Yeah, we should definitely add an implementation for types implementing `IntoFuture`; the only reason we don't have that already is historical: `tracing::Instrument` predates `IntoFuture` being stable in the standard library.

Your program appears to be spawning a very large number of threads at the same time. Can you confirm that there are 4096 threads concurrently active in your program?

Thanks for working on this, this is something I had been planning on doing for a long time but never had the time for. I'm really excited to review this...

I would really like something like this to exist. However, I don't really think the proposed API is actually possible. With a `Future`, an `Instrumented` can implement the `Future` trait....

The [clippy failure](https://github.com/tokio-rs/tracing/pull/2812/checks?check_run_id=18853223091) is unrelated to this change, so don't worry about that.

I'm fine with making functions that borrow both the inner future and the span as a tuple public, the motivation definitely makes sense to me! Personally, I don't love the...