Add feature gates for for and ? in consts

[jonas-schievink]

These operations seems relatively straightforward to support, and only seem to be blocked on impl const Trait.

I have included a working test for const_try, but const_for is currently unusable without reimplementing every single defaulted Iterator method, so I didn't do that.

(both features still need tracking issues before this is merged)

[oli-obk]

but const_for is currently unusable without reimplementing every single defaulted Iterator method, so I didn't do that.

maybe #86857 helps here?

[jonas-schievink]

In theory, yeah, this works:

#![feature(const_refs_to_cell)]
#![feature(const_mut_refs)]
#![feature(const_fn_trait_bound)]
#![feature(const_trait_impl)]

trait Iter: Copy + Sized {
    type Item: Copy;
    
    fn next(&mut self) -> Option<Self::Item>;
    
    #[default_method_body_is_const]
    fn count(mut self) -> usize {
        let mut count = 0;
        while self.next().is_some() {
            count += 1;
        }
        count
    }
}

#[derive(Copy, Clone)]
struct ConstIter {
    next: Option<u8>,
}

impl const Iter for ConstIter {
    type Item = u8;
    
    fn next(&mut self) -> Option<u8> {
        match self.next {
            Some(val) => {
                self.next = val.checked_add(1);
                Some(val)
            }
            None => None,
        }
    }
}

const _: () = {
    let c = ConstIter { next: Some(0) }.count();
    if c != 256 {
        [()][4];
    }
};

...but it requires a lot of feature gates and Copy bounds (since everything is generic) that wouldn't work on Iterator. Seems like there's still a bunch of work to do until iterators work in consts.

[oli-obk]

Then I would say we don't add a const_for feature gate, we can start figuring out const iterators without it.

[jonas-schievink]

But we'll need one eventually, right? Why not add it now?

[oli-obk]

Yea. Just right now it doesn't really get us anything actionable. I guess as long as we have a canary test showing the behaviour of the feature gate it should be ok to have an unusable feature gate. Just thought it unusual.

[bors]

☔ The latest upstream changes (presumably #83484) made this pull request unmergeable. Please resolve the merge conflicts.

[bors]

☔ The latest upstream changes (presumably #86998) made this pull request unmergeable. Please resolve the merge conflicts.

[usbalbin]

#86853 is a PR which might have some overlap with the try-operator part of this. My goal there was to try and constify the most trivial conversion impls to enable basic usage with Option and Result (however that turned out to be blocked on some more work on const_trait_impl)

[usbalbin]

Btw is #87576 a duplicate of #74935? :)

[jonas-schievink]

Yes, these tracking issues are pretty hard to discover

[oli-obk]

I guess we should start printing the function's name in the diagnostic. Very uninformative this error XD

[oli-obk]

@bors r+

[bors]

📌 Commit c5a29f9 has been approved by oli-obk

[bors]

⌛ Testing commit c5a29f9 with merge 87dc824...

[bors]

☀️ Test successful - checks-actions
Approved by: oli-obk
Pushing 87dc824 to master...