swc-project/swc · error

failed to execute transfomred code

Error message

failed to execute transfomred code

What it means

In the babel_like harness, execution tests that compare stdout first run the TRANSFORMED code through node via `stdout_of(&code)` (which delegates to `exec_node_js`). The `expect("failed to execute transfomred code")` (typo intentional, it is in the source) fires when running the transformed output fails to produce stdout: node cannot be spawned or the transformed code crashes/errors when executed.

Source

Thrown at crates/swc_ecma_transforms_testing/src/babel_like.rs:202

                Color::Green.paint("Code"),
                code_without_helper
            );

            if let Some(output_path) = output_path {
                // Fixture test

                if !self.allow_error && handler.has_errors() {
                    return Err(());
                }

                NormalizedOutput::from(code)
                    .compare_to_file(output_path)
                    .unwrap();
            } else if compare_stdout {
                // Execution test, but compare stdout

                let actual_stdout: String =
                    stdout_of(&code).expect("failed to execute transfomred code");
                let expected_stdout =
                    stdout_of(&fm.src).expect("failed to execute transfomred code");

                testing::assert_eq!(actual_stdout, expected_stdout);
            } else {
                // Execution test

                exec_with_node_test_runner(&format!("// {}\n{code}", self.input.display()))
                    .expect("failed to execute transfomred code");
            }

            Ok(())
        });

        if self.allow_error {
            match err {
                Ok(_) => {}
                Err(err) => {

View on GitHub (pinned to 5176682b65)

Solutions

1. Run the failing fixture manually: the harness prints the transformed code; save it and run `node file.js` to see the real error.
2. Install node and ensure it is on PATH (`node --version` from the same shell that runs cargo).
3. If node is too old for the emitted syntax (e.g. top-level await, class fields), upgrade node.
4. Temporarily skip execution tests with `EXEC=0 cargo test` while diagnosing, then re-enable.

// Skip execution tests when node is unavailable.
fn node_available() -> bool {
    std::process::Command::new("node")
        .arg("--version")
        .output()
        .map(|o| o.status.success())
        .unwrap_or(false)
}

When it happens

Trigger: A `compare_stdout`-style fixture where the transformed code throws at runtime (bad transform output, reference error, syntax the generated code uses but node rejects), or the `node` executable is missing from PATH in the test environment.

Common situations: Running execution tests (`EXEC != 0`) in CI images without node installed; a transform regression that emits broken code; transformed output using syntax newer than the installed node version.

Related errors

• CARGO_MANIFEST_DIR
• test_inlined_transform does not support paths outside of the
• failed to get current directory
• failed to read file
• failed to create parent directory for temp directory

AI-assisted analysis of swc-project/swc@5176682b65 (2026-08-17). Data as JSON: /api/errors/d7127f799fd346af. Report an issue: GitHub.