Microbenchmark harness comparing targets doing the same work, e.g. agplatform wrappers against the std functions they wrap. Each benchmark has a setup creating its fixture, its targets are measured in turn on the fixture and reported relative to the first one (the baseline), then the fixture is dropped.
// <crate>/benches/fs.rs, with `[[bench]] name = "fs" harness = false`
use agbench::Bench;
use agplatform::{Fs, Platform};
fn main() -> std::process::ExitCode {
agbench::main(&[read])
}
fn read(bench: &mut Bench) {
bench
.benchmark("fs/read", agplatform::platform)
.target("std", |_| std::fs::read("Cargo.toml").unwrap())
.target("agplatform", |p| p.fs().read("Cargo.toml").unwrap())
.run();
}Run
cargo bench [-p <crate>] -- [--filter PATTERN]... [--filter-target PATTERN]... [--samples N] [--sample-ms MS] [--iters N] [--list] [--check]
Possible output:
benchmark target iters time overhead IQR
fs/read/64B std 8192 927.0 ns
agplatform 1.87 µs +100.4% [+67.6% .. +106.5%]
Platform providing env, fs, request and process infrastructure for production and testing. Abstracted over a Platform trait it allows easy swapping production and test implementations.
fn read_file(p: impl agplatform::Platform) {
let content = p.fs().read_to_string()?;
}
read_file(agplatform::platform()); // production implementation using std, tokio, reqwest
read_file(agplatform::test_platfrom()); // test in-memory implementation with abilities to simulate failures, manipulate time etc.Cargo subcommand checking the structure of Rust tests binding the tests code in the production structure and enforcing the rules. See the [test rules](agtestcheck/rust-test-rules.md) for details.
// source.rs
fn foo() -> i32 { 1 }
struct S;
impl S { fn bar() -> Self { S } }
#[cfg(test)]
mod tests {
use super::*;
mod foo {
#[test]
fn return_1() { assert_eq(foo(), 1); }
}
mod s {
mod bar {
#[test]
fn bar_should_crate_s() { /* testing stuff */}
}
}
#[test]
fn ranom_arbitrary_test() {}
}Run
# cargo install agtestcheck
cargo test-check [<dir|file.rs>]
Possible output:
source.rs:12: test `return_1` should be named `foo_should_<behaviour>` or `foo_<args>_should_<behaviour>`
source.rs:18: comment "testing stuff" should be deleted, start with WHY or GIVEN, WHEN, THEN, AND
source.rs:23: test `ranom_arbitrary_test` should be in the module of its type, trait, function or method