如何在Windows上调试Rust单元测试？

Question

I'm developing code for the Codingame problems using VS Code on Windows with Rust and the Visual Studio toolchain.

I have found multiple guides explaining how to debug the executable generated by cargo build , the best being Debug Rust on Windows with Visual Studio Code and the MSVC Debugger .

However, when I face problems, I tend to write unit tests (I've done that in Java, JavaScript, Ruby, ...), which I then debug. Unfortunately, I can't find any way to do that in Rust. How do I configure my environment to debug my tests?

I'm not talking about adding println! statements in my tests, as I already know how to do that. I'm also not talking about adding new assertions, because those reside in the test, not in the tested code.

What I want is to use the VS Code Debugger on the code called by my test.

Answer 1

Rust unit tests are compiled as separate binaries, which means you debug them exactly the same as any other binary . Once compiled, they are located at ./target/debug/$name-$hash .

Visual Studio Code

Here's modified versions of the VS Code configuration files that allow me to debug a unit test.

tasks.json

{
    "type": "shell",
    "label": "cargo test build",
    "command": "cargo",
    "args": [
        "test", "--no-run"
    ],
    "problemMatcher": [
        "$rustc"
    ]
}

launch.json

{
    "name": "Run Test Debugger",
    "type": "cppvsdbg",
    "request": "launch",
    "program": "${workspaceFolder}/target/debug/buggin-70708b3916187eeb.exe",
    "args": [],
    "stopAtEntry": false,
    "cwd": "${workspaceFolder}",
    "environment": [],
    "externalConsole": true,
    "preLaunchTask": "cargo test build",
}

Working

Windbg

Build your tests:

cargo test --no-run

Open the built executable in Windbg and open the source file.

---

Finding the hash is the most annoying aspect. The best solution I know of is to write a small script that builds the tests and then finds the test executable based on which is newest. My Powershell skills are not adequate to the task, nor do I know how to directly integrate this with VS Code or Windbg.

There are open issues for Cargo to help with identifying the file:

• produce deterministic filenames for build --test and test --no-run (#1924)
• Make it possible to run binaries produced by cargo directly (#3670)

Answer 2

You could either use the println! debugging

#[test]
fn test_false() {
    println!("This test is not OK");
    assert!(1 == 2);
}

Or use the assert macro to provide custom messages.

#[test]
fn test_complex() {
    let a = vec![1, 2];
    let b = vec![3, 4, 5];
    assert!(a.len() == b.len(), "Mismatch between vec lenghts! ({} - {})", a.len(), b.len());
}

This will result in

cargo test
    Finished dev [unoptimized + debuginfo] target(s) in 0.00s
     Running target/debug/deps/foo-3cf890b7dd22d797

running 3 tests
test test_complex ... FAILED
test test_false ... FAILED
test test_true ... ok

failures:

---- test_complex stdout ----
thread 'test_complex' panicked at 'Mismatch between vec lenghts! (2 - 3)', src/main.rs:21:5
note: Run with `RUST_BACKTRACE=1` for a backtrace.

---- test_false stdout ----
This test is not OK
thread 'test_false' panicked at 'assertion failed: 1 == 2', src/main.rs:14:5


failures:
    test_complex
    test_false

test result: FAILED. 1 passed; 2 failed; 0 ignored; 0 measured; 0 filtered out

error: test failed, to rerun pass '--bin foo'

Note: The third, good case has also a println! in it, but it won't show up, because the test is successful.

In general, your test cases should always be that easy, that there should no need to debug your test cases, but your code. But that's theory I guess :)