Crates.io | dbg_if |
lib.rs | dbg_if |
version | 0.1.0 |
source | src |
created_at | 2024-05-16 01:40:49.753833 |
updated_at | 2024-05-16 01:40:49.753833 |
description | Use dbg! in the loop without terminal woes. |
homepage | |
repository | https://github.com/shanecelis/dbg_if |
max_upload_size | |
id | 1241684 |
size | 42,330 |
dbg!
in the loop without terminal woes.
The macro [dbg_once!
] only prints its value the first time.
use dbg_if::dbg_once;
for i in 0..10 {
dbg_once!(i); // Outputs: [src/lib.rs:9:9] x = 0
}
The macro [dbg_if_ne!
] only prints changed values.
use dbg_if::dbg_if_ne;
fn f(x: u8) -> u8 {
dbg_if_ne!(x, u8)
}
f(1); // Outputs: [src/lib.rs:58:9] x = 1
f(1); // No output.
f(2); // Outputs: [src/lib.rs:58:9] x = 2
The macro [dbg_if_hash_ne!
] only prints on changed hash values.
use dbg_if::dbg_if_hash_ne;
let mut s: String = "hello".into();
fn f(x: &str) -> &str {
dbg_if_hash_ne!(x)
}
f(&s); // Outputs: [src/lib.rs:37:9] x = "hello"
f(&s); // No output.
s.push('!');
f(&s); // Outputs: [src/lib.rs:37:9] x = "hello!"
The sister macros [once!
], [was_ne!
], and [was_hash_ne!
] return true instead
of printing.
Finally the macro [dbg_if
] provides a kind of drop-in replacment for
dbg
if that is your preference.
use dbg_if::dbg_if as dbg;
let mut x: u8 = 0;
fn f(x: u8) -> u8 {
dbg!(x + 1);
dbg!(x + 2, Once);
dbg!(x + 3, IfNe, u8);
dbg!(x + 4, IfHashNe)
}
x = f(x);
// Outputs:
// [src/lib.rs:10:9] x + 1 = 1
// [src/lib.rs:11:9] x + 2 = 2
// [src/lib.rs:12:9] x + 3 = 3
// [src/lib.rs:13:9] x + 4 = 4
x = f(x);
// Outputs:
// [src/lib.rs:10:9] x + 1 = 1
If the feature "float" is enabled, these macros are available:
abs_diff_ne_args!
] accepts epsilon
argument,relative_ne_args!
] accepts epsilon
and relative_max
arguments,ulps_ne_args!
] accepts epsilon
and ulps_max
arguments.These can be given as the third argument to [was_ne!
] or [dbg_if_ne!
]. See
the [approx
] crate for more details.
#[cfg(feature = "float")]
{
use dbg_if::{dbg_if_ne, abs_diff_ne_args};
fn f(x: f32) -> f32 {
dbg_if_ne!(x, f32, abs_diff_ne_args!(epsilon = 1.0))
}
f(1.0); // Outputs: [src/lib.rs:42:9] x = 1.0
f(1.5); // No output.
f(2.0); // No output.
f(2.1); // Outputs: [src/lib.rs:42:9] x = 2.1
}
fn f(x: u8) -> u8 {
dbg!(x) + 1
}
assert_eq!(f(1), 2);
The dbg!
macro is great. It's like being able to add a probe right into your
code without disturbing everything since it works on expressions and lets them
"pass thru." For straight shot code, it is perfect.
fn f(x: u8) -> u8 {
let mut accum = 0;
for i in 0..100 {
accum += dbg!(x);
}
accum
}
[src/main.rs:59:18] x = 1
[src/main.rs:59:18] x = 1
[src/main.rs:59:18] x = 1
...^C
For code in tight loops, however, dbg!
leaves something to be desired. The
terminal screams, "x = 1" again and again. There has got to be a better way.
Yes! Let's take note of the value at the call site—with a static atomic
variable—and instead of spamming the terminal with the same information, let's
only emit information when it has changed with [dbg_if_ne!
].
use dbg_if::dbg_if_ne;
fn f(x: u8) -> u8 {
let mut accum = 0;
for i in 0..5 {
accum += dbg_if_ne!(x, u8);
}
accum
}
f(1); // Outputs: [src/main.rs:59:18] x = 1
That's fine. Can they be hashed? Because hashes can be stored in an AtomicU64
at the call site. Just use [dbg_if_hash_ne!
].
Some tests require a particular setup in order to run successfully. A couple of
aliases have been placed in .cargo/config.toml
to run these tests.
cargo test
runs the was*
tests.cargo test-output
runs above and the dbg*
tests which verify its output on
stdout.cargo test-all
runs above and the float features.If you see errors that say, "Redirect already exists," that's because some tests
check the stdout and cannot be run multi-threaded. Use cargo test-output
to
run them with the right arguments.
This crate is licensed, at your option, under either
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.
Thank you to Philipp Oppermann for his crate
once
. I initially thought I'd only
write dbg_once!
and submit a PR. But once I got going I realized dbg_if_ne!
would be useful too and these are all require std
; once
is a no_std
crate.
So dbg_if
is inspired and informed by once
but it actually doesn't share
any code with once
.