impl<T> From<Option<&T>> for *const T

[kupiakos]

And same for *mut T. I wasn't sure whether it was worth defining the trait as const for the mut version, considering that's a separate feature for mut refs in const. I also considered having the implementation be the transmute this optimizes to, but decided against it.

This also puts off defining the impl for T: ?Sized since the validity of creating a null *const dyn Trait is not settled.

This is my first contribution to the repo - please let me know if I missed something obvious :)

Zulip thread

[rust-highfive]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @kennytm (or someone else) soon.

Please see the contribution instructions for more information.

[kupiakos]

Is there anything else I should be doing for this?

[bors]

☔ The latest upstream changes (presumably #95581) made this pull request unmergeable. Please resolve the merge conflicts.

[Mark-Simulacrum]

Hi! Apologies for the long delay. I've re-assigned this PR to @joshtriplett since it'll need an FCP to be merged (adding new trait impls).

[camelid]

r? rust-lang/libs-api

[dtolnay]

Reassigning BurntSushi's reviews because based on git log -i --grep=r=burntsushi the last time they did rust-lang/rust reviews was over 3 years ago.

r? rust-lang/libs-api

[pitaj]

Ping from triage. @kupiakos FYI you still have merge conflicts. Please @rustbot review when you've resolved them

@rustbot author

[kupiakos]

Looks like the story around const traits has changed. In order to allow .into() in const, I would have to add #[const_trait] on both From and Into, which they don't do at the moment. That change seemed out-of-scope for this PR, and so I no longer have the From<&T> for *const T conversion be const-capable.

@rustbot review

[kupiakos]

Ah, I sent for review too soon. It should be right now. The failing CI and subsequent fix actually shows this is technically a breaking change:

<*const _>::from(x) where x: &T is...weird, but did compile. That line causes the from to resolve to the blanket impl From<T> for T with T = *const Enum. Now, it's ambiguous and results in an error. There has never been a impl<T> From<&T> for *const T which that line implies exists. This being a stable breaking change might be enough reason to reject the PR entirely.

[Dylan-DPC]

Closing this as it is a breaking change and the author agrees it is better to close it