I've found that, in some cases, inside a macro_rules matcher (the thing on the left of the =>), $crate is treated as if it's a normal identifier, which seems very weird. I'm not sure if this is intended or a bug.
This weirdness manifests in two different ways:
- A
$crate on its own in the matcher is treated as literally matching $crate, as opposed to being a metavariable or erroring:
macro_rules! foo {
($crate and $crate:tt) => {
"wtf"
}
}
macro_rules! bar {
() => {
foo!($crate and $crate:tt)
}
}
fn main() {
println!("{}", bar!()); // prints "wtf"
}
- I can create a metavariable (e.g., the metavariable
$x:tt has the name x) such that the metavariable's name is $crate:
macro_rules! bar {
($dol:tt) => {
macro_rules! foo {
($dol $crate : expr) => { $dol $crate }
}
}
}
bar!($);
fn main() {
println!("{}", foo!(1 + 2)); // prints 3
}
See also #146967 and #146968 and #146114 for weirdness with $crate.
Meta
Reproducible on the playground with version 1.92.0-nightly (2025-09-24 caccb4d0368bd918ef66)
I've found that, in some cases, inside a
macro_rulesmatcher (the thing on the left of the=>),$crateis treated as if it's a normal identifier, which seems very weird. I'm not sure if this is intended or a bug.This weirdness manifests in two different ways:
$crateon its own in the matcher is treated as literally matching$crate, as opposed to being a metavariable or erroring:$x:tthas the namex) such that the metavariable's name is$crate:See also #146967 and #146968 and #146114 for weirdness with
$crate.Meta
Reproducible on the playground with version
1.92.0-nightly (2025-09-24 caccb4d0368bd918ef66)