Stack overflow with struct update syntax

[crisidev]

This problem happens when using Struct update syntax with Default implementation (https://doc.rust-lang.org/book/ch05-01-defining-structs.html#creating-instances-from-other-instances-with-struct-update-syntax).

I have found this bug in a more complicated piece of code, but I was able to create a simpler reproduction.

#[derive(Debug)]
pub struct Router<B> {
    routes: Vec<(B, i32)>,
}

impl<B> Default for Router<B>
where
    B: Send + 'static,
{
    fn default() -> Self {
        Self::new()
    }
}

impl<B> Router<B>
where
    B: Send + 'static,
{
    pub fn new() -> Self {
        Self {
            ..Default::default()
        }
    }
}

fn main() {
    let r: Router<String> = Router::new();
    println!("Hello, world!");
}

I expected this to just print "Hello, world!", but instead it panics with a stack overflow:

    Finished dev [unoptimized + debuginfo] target(s) in 0.00s
     Running `target/debug/repro`

thread 'main' has overflowed its stack
fatal runtime error: stack overflow
zsh: abort      RUST_BACKTRACE=1 cargo run

Changing the Struct update syntax to explicit parameters resolves the issue:

    pub fn new() -> Self {
        Self {
            routes: Default::default()
        }
    }

    Finished dev [unoptimized + debuginfo] target(s) in 0.82s
     Running `target/debug/repro`
Hello, world!

Meta

The issue is reproducible with both stable and nightly compilers and adding RUST_BACKTRACE=1 or RUST_BACKTRACE=full does not produce a stack trace.

The reproduction runs also on playground: https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=e5dd848f3af5f5f2b1f13c1d70163f5b

[robojumper]

Router::<_>::new and <Router<_> as Default>::default are mutually recursive. #57965 tracks detecting unconditional recursion across function calls, which would at least emit a compiler warning in this case.

[inquisitivecrystal]

Thanks for the report! Since this appears not to actually be a problem with rustc, and there's already an issue tracking the diagnostic, I'm going to close this issue. Anyone should feel free to reopen if there's something I'm missing.

[crisidev]

Of course! I'll keep a close eye to the diagnostic issue. Thanks for the help!