2024: Reserved ## diagnostic is confusing

[ehuss]

Code

// Edition 2024
macro_rules! m {
    ($a:tt $b:tt) => {};
}

m!{##}

Current output

error: invalid string literal
 --> src/lib.rs:5:4
  |
5 | m!{##}
  |    ^^
  |
  = note: unprefixed guarded string literals are reserved for future use since Rust 2024
help: consider inserting whitespace here
  |
5 | m!{# #}
  |     +

error: unexpected end of macro invocation
 --> src/lib.rs:5:6
  |
1 | macro_rules! m {
  | -------------- when calling this macro
...
5 | m!{##}
  |      ^ missing tokens in macro arguments
  |
note: while trying to match meta-variable `$b:tt`
 --> src/lib.rs:2:12
  |
2 |     ($a:tt $b:tt) => {};
  |            ^^^^^

Desired output

error: reserved multi-hash token is not allowed
 --> src/lib.rs:5:4
  |
5 | m!{##}
  |    ^^
  |
  = note: a sequence of two or more # is reserved for future use since Rust 2024
help: consider inserting whitespace here
  |
5 | m!{# #}
  |     +

error: unexpected end of macro invocation
 --> src/lib.rs:5:6
  |
1 | macro_rules! m {
  | -------------- when calling this macro
...
5 | m!{##}
  |      ^ missing tokens in macro arguments
  |
note: while trying to match meta-variable `$b:tt`
 --> src/lib.rs:2:12
  |
2 |     ($a:tt $b:tt) => {};
  |            ^^^^^

Rationale and extra context

The existing message referring to a "string literal" is confusing to me since there are no strings anywhere in the source. I don't think it will be obvious to users to make the connection between ## and "that might come before some reserved string thing that doesn't yet exist". I would suggest avoiding the terminology of "string literal" if there is no string literal following the last #.

Other cases

No response

Rust Version

rustc 1.83.0-nightly (1bc403d 2024-10-11)
binary: rustc
commit-hash: 1bc403d
commit-date: 2024-10-11
host: aarch64-apple-darwin
release: 1.83.0-nightly
LLVM version: 19.1.1

Anything else?

No response

[traviscross]

cc @pitaj