Allow transmuting generic pattern types to and from their base

[oli-obk]

Pattern types always have the same size as their base type, so we can just ignore the pattern and look at the base type for figuring out whether transmuting is possible.

[rustbot]

r? @BoxyUwU

rustbot has assigned @BoxyUwU.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[BoxyUwU]

How does this handle types containing pattern types, e.g.

#![feature(pattern_types, pattern_type_macro)]

use std::pat::pattern_type;

fn foo<const N: u32>() {
    std::mem::transmute::<Option<u32>, Option<pattern_type!(u32 is N..)>>(loop {});
}

Which should not compile as those two options have different sizes when N > 0

[oli-obk]

These are still rejected accidentally. So good and bad. I think we can land this because there's no code path that could accidentally accept anything here (we explicitly check for pointers)

[BoxyUwU]

So I guess basically all the complex SizeSkeleton stuff exists just to handle pointers in transmute checks more cleverly, everything else is just handled by layout_of or simply =='ing the source/target types in the transmute. So maybe the thing that's actually wrong would be:

#![feature(pattern_types, pattern_type_macro)]

use std::pat::pattern_type;

struct Pointee<T>(T);

fn foo<T, U>() {
    std::mem::transmute::<
        Option<*mut Pointee<T>>,
        Option<
            pattern_type!(*mut Pointee<U> is /* nonnull */>
            ),
        >,
    >(loop {});
}

fn main() {}

Which would just be a matter of checking if the pattern type is a nonnull pattern refinement and changing the SizeSkeleton::Pointer to have nonzero set, similar to how the SizeSkeleton for newtype adts will check for layout_scalar_valid_range... 🤔

Correct me if I'm wrong but I don't think you even support pointer refinement in pattern types right now so it seems like kind of moot to try and handle this correctly rn.

[BoxyUwU]

@bors r+ rollup

[bors]

📌 Commit a639e27 has been approved by BoxyUwU

It is now in the queue for this repository.

[oli-obk]

Correct me if I'm wrong but I don't think you even support pointer refinement in pattern types right now so it seems like kind of moot to try and handle this correctly rn.

Yes, we'll have to look into that when we get there.