Usage of void and replacement with core::convert::Infallible

[aleokdev]

Hello, I've noticed a crate named void is used for infallible situations. However, since Rust 1.34.0, we have Infallible, which performs this task. Am I missing something, or could the void type the crate offers be replaced by this type?

[Rahix]

The missing pieces is the ResultVoidExt trait which gives us the handy void_unwrap() method.

core already has an equivalent in the making, called into_ok() but unfortunately this is still unstable.

So far, my take on the situation was to wait on stabilization of into_ok() and switch then. That is, unless the use of void proves to be problematic for downstream users in some place. If that is the case, please speak up :)

[Beanow]

See also:

• Solution for asserting infallibility of a Result<T, Infallible> rust-embedded/embedded-hal#329

At least in embedded-hal 0.2 there were some instances of void too. https://github.com/rust-embedded/embedded-hal/pull/150/files so regardless of what void_unwrap alternative to use, I think a migration to embedded-hal 1.0 should drop the void crate too.

[Rahix]

Nice, thanks for digging into this! I wasn't aware of the unwrap-infallible crate until now - I'd be totally on board with replacing void with unwrap-infallible here in avr-hal.

You are right, the best time is probably when adopting e-h 1.0 - until then I think the disruption is not worth it...