How to idiomatically convert between u32 and usize?
We now have a pretty different answer when we try to compile your code, replacing the number 1
with a variable of type i32
:
error[E0308]: mismatched types
--> src/main.rs:5:28
|
5 | let ch = s.chars().nth(n).unwrap();
| ^ expected usize, found i32
help: you can convert an `i32` to `usize` and panic if the converted value wouldn't fit
|
5 | let ch = s.chars().nth(n.try_into().unwrap()).unwrap();
|
It means that now the compiler recommends you to use n.try_into().unwrap()
that makes use of the trait TryInto
which in turn relies on TryFrom
and returns a Result<T, T::Error>
. That's why we need to extract the result with a .unwrap()
TryInto
documentation
The as
operator works for all number types:
let ch = s.chars().nth(n as usize).unwrap();
Rust forces you to cast integers to make sure you're aware of signedness or overflows.
Integer constants can have a type suffix:
let n = 1u32;
However, note that negative constants, such as -1i32
is internally -
1i32
.
Integer variables declared without an explicit type specification are shown as {integer}
and will be properly inferred from one of the method calls.
The most cautious thing you can do is to use TryFrom
and panic when the value cannot fit within a usize
:
use std::convert::TryFrom;
fn main() {
let s = "abc";
let n: u32 = 1;
let n_us = usize::try_from(n).unwrap();
let ch = s.chars().nth(n_us).unwrap();
println!("{}", ch);
}
By blindly using as
, your code will fail in mysterious ways when run on a platform where usize
is smaller than 32-bits. For example, some microcontrollers use 16-bit integers as the native size:
fn main() {
let n: u32 = 0x1_FF_FF;
// Pretend that `usize` is 16-bit
let n_us: u16 = n as u16;
println!("{}, {}", n, n_us); // 131071, 65535
}
For broader types of numeric conversion beyond u32
<-> usize
, refer to How do I convert between numeric types safely and idiomatically?.
See also:
- Why is type conversion from u64 to usize allowed using `as` but not `From`?