Misleading error messages

[TheButlah]

The error messages cmd_lib gives don't seem as intuitive as bash. Consider the following code:

fn main() -> color_eyre::Result<()> {
    color_eyre::install()?;

    let log_to_find = "evilevil";
    cmd_lib::run_cmd! {
        info "this is a log message";
        pwd;
        cat log.txt | grep $log_to_find;
    }?;

    Ok(())
}

This produces this error:

❯ cargo run
   Compiling foo v0.0.0 (/Users/ryan.butler/P/scratch/foo)
    Finished dev [unoptimized + debuginfo] target(s) in 0.21s
     Running `target/debug/foo`
[INFO ] this is a log message
/Users/ryan.butler/P/scratch
[INFO ] cat: log.txt: No such file or directory
Error:
   0: Running ["grep" "evilevil"] exited with error; status code: 1 at foo/src/main.rs:5

Location:
   foo/src/main.rs:5

But in bash, I get:

❯ cat log.txt | grep evilevil
cat: log.txt: No such file or directory

[tao-guo]

The truth is that: the error was indeed from grep, even in bash (w/ or w/o pipefail):

$ ls /xx | grep y
ls: cannot access '/xx': No such file or directory
$ echo $?
1
$ ls /xx
ls: cannot access '/xx': No such file or directory
$ echo $?
2
$ echo /xx | grep y
$ echo $?
1
$ set -o pipefail
$ ls /xx | grep y
ls: cannot access '/xx': No such file or directory
$ echo $?
1

We have also logged the stderr output, so you can find out why it caused the grep to fail. Note ideally the log level should be WARN/ERROR, however the library only knows there was information from stderr and they are OK in most of the cases.