Use ::core qualified imports instead of ::std inside tokio::test macro

[complexspaces]

Motivation

At work, we have a crate in our monorepo which is #![no_std] for build size reasons. However, we use tokio to test that its async functionality is functional in CI for simplicity and leave the other executor for the "real" usage of the library.

In a recent tokio update, we started to see this error when compiling its tests:

error[E0433]: failed to resolve: could not find `std` in the list of imported crates
   --> crypto/crate-name/src/lib.rs:556:5
    |
556 |     #[tokio::test]
    |     ^^^^^^^^^^^^^^ could not find `std` in the list of imported crates
    |
    = note: this error originates in the attribute macro `tokio::test` (in Nightly builds, run with -Z macro-backtrace for more info)

I was not immediately able to find which commit caused this, but you can observe the cause here:

tokio/tokio-macros/src/entry.rs

Lines 402 to 406 in fd7d0ad

	quote! {
	let body = async #body;
	#crate_path::pin!(body);
	let body: ::std::pin::Pin<&mut dyn ::std::future::Future<Output = #output_type>> = body;
	}

You can reproduce the issue with this project:

[package]
name = "reproducer"
version = "0.1.0"
edition = "2021"

[dev-dependencies]
tokio = { version = "1", features = ["rt", "macros"] }

#![no_std]

async fn future_to_test() -> i32 {
    1
}

#[cfg(test)]
mod tests {
    use super::*;

    #[tokio::test]
    async fn test_future() {
        assert_eq!(future_to_test().await, 1);
    }
}

Solution

Given that that std::future and std::pin are re-exports of their core equivalents, I propose to qualify these imports inside the tokio::test macro with ::core instead of ::std for better compatibility. AFAIK this should not have any negative impacts for other users.

By making this change, a #![no_std] crate can continue to smoothly use tokio as a dev-dependency without any workarounds needed in the rest of the crate.

[taiki-e]

Thanks!