Crates.io | match_type |
lib.rs | match_type |
version | 0.1.5 |
source | src |
created_at | 2023-07-31 15:44:16.14179 |
updated_at | 2023-08-07 15:38:04.92678 |
description | Match on the type of an expression at compile time. |
homepage | |
repository | https://github.com/Garbaz/match_type |
max_upload_size | |
id | 930995 |
size | 32,784 |
‖ Docs.rs ‖ Lib.rs ‖ Crates.io ‖
Match on the type of an expression at compile time.
For example the following macro will expand to a different String
expression depending on what the type of the input expression is:
struct A(i64,i32,i16,i8);
struct B(f64,f32,bool);
macro_rules! as_string {
($e:expr) => {
match_type!(
$e {
A => String: "It's an A :O".into(),
B => String: "B it is ^^".into(),
<T: Display> T => String: format!("{}", self),
<T: Debug> T => String: format!("{:?}", self),
_ => String: "<<Sad Monkey :(>>".into(),
}
)
}
}
Or this macro will give us the "opposite" of a value, using a different operation depending on which is supported (with -
taking precedence in case both !
and -
are):
macro_rules! opposite {
($e:expr) => {
match_type!(
$e {
<T: Neg> T => <T as Neg>::Output: -self,
<T: Not> T => <T as Not>::Output: !self,
}
)
};
}
Note that the right hand sides of the different match arms do not have to be of the same type.
With opt-level
of at least 1
, e.g. in the default release
profile, the whole expansion of match_type
should be inlined away to the point that the assembly output is the same as if you had written the right hand side of the matching arm directly. I.e. match_type
should have zero production runtime overhead.
See HOW.md