regression: overflow evaluating requirement NonNull<Vec<u8>>: Unpin

[Mark-Simulacrum]

https://crater-reports.s3.amazonaws.com/beta-1.74-4/beta-2023-10-21/reg/zvariant-3.15.0/log.txt

[INFO] [stdout] error[E0275]: overflow evaluating the requirement `NonNull<Vec<u8>>: Unpin`
[INFO] [stdout]   |
[INFO] [stdout]   = help: consider increasing the recursion limit by adding a `#![recursion_limit = "256"]` attribute to your crate (`zvariant`)

This seems to recurse into more nested vecs -- not clear why, but doesn't seem like the right behavior.

https://crater-reports.s3.amazonaws.com/beta-1.74-4/beta-2023-10-21/gh/lcnr.binary/log.txt is potentially another example... @lcnr, maybe this means that this is a known bug?

[lcnr]

the root goal seems to be

Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<Vec<u8>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>: Unpin

which feels like it should be incredibly close to the recursion limit. Considering that it hit the recursion limit at NonNull<Vec<u8>>: Unpin it looks like we now increment the recursion depth a bit more somewhere.

The same for https://github.com/lcnr/binary/blob/master/src/main.rs, I apparently set the recursion limit as low as possible, incrementing it to 540 or sth makes this compile again.

#![recursion_limit = "513"]
// We need O(n) levels of recursion, where n is the maximum amount of bits,
// and as we multiply two 256 bit numbers, this is the lowest possible value.

I personally think it's fine to break these crates (maybe opening a Pr to zvariant increasing the recursion limit), but it would still be good to know the root cause.

I wasn't able to test zvariant myself but bisecting https://github.com/lcnr/binary/blob/master/src/main.rs should work 🤔

[Mark-Simulacrum]

Ah, I missed that these lower the limit. Maybe we can add a note about that to these errors?

[apiraino]

WG-prioritization assigning priority (Zulip discussion).

@rustbot label -I-prioritize +P-medium