`const_caller_location` should use real Span instead of `DUMMY_SP`

[cachebag]

Background:
I was reading through the codebase and in caller_location.rs, I noticed a FIXME about the usage of DUMMY_SP but I couldn't find or see any existing tracking issue for it.

I tried this code:

// internal API, not user-facing
// fn span_as_caller_location(self, span: Span) -> ConstValue

I expected to see this happen: A tracking issue or documentation for why Span is used directly here, or any design notes.

Instead, this happened: The function passes a placeholder span, but there is no associated issue or explanation of next steps (that I could find at least, I could very well be wrong about this).

Meta

rustc --version --verbose:

rustc 1.88.0 (6b00bc388 2025-06-23)
binary: rustc
commit-hash: 6b00bc3880198600130e1cf62b8f8a93494488cc
commit-date: 2025-06-23
host: x86_64-unknown-linux-gnu
release: 1.88.0
LLVM version: 20.1.5

Backtrace

N/A

[RalfJung]

Thanks for trying to improve Rust! However, picking a random FIXME and making an issue for it isn't going to be very helpful, I am afraid. If you are looking for things to work on, look for issues labeled E-easy or E-mentor.

We don't have tracking issues for each FIXME. It's not worth it. Almost all of them are minor.

FIXMEs are notes by rustc devs for rustc devs. They are not user-facing, and it's okay if they can only be understood with enough background in the area of the compiler that they appear in. They differ wildly in priority. Many are outdated. They are not meant for new contributors, and you should consult with an experienced contributor before trying to work on one.

[cachebag]

@RalfJung

Thanks for the tips, sorry about that. I'll read the guide more thoroughly.