brush doesn't seem to implement kill with a numeric signal argument

[jmquigs]

brush version

brush 0.4.0 (git:d42bdd2)

bash version (for comparison)

GNU bash, version 5.2.21(1)-release (x86_64-pc-linux-gnu)

Platform / OS

Linux

Steps to reproduce

Under brush just run kill with any numeric argument for the signal on a pid you own, it will display an error.

~$ bash
~$ ps
    PID TTY          TIME CMD
  62862 pts/5    00:00:00 bash
  62889 pts/5    00:00:00 bash
  62901 pts/5    00:00:00 ps
~$ kill -9 62889
Killed
~$ brush
~$ ps
    PID TTY          TIME CMD
  62862 pts/5    00:00:00 bash
  62922 pts/5    00:00:00 brush
  62955 pts/5    00:00:00 ps
~$ kill -9 62922
kill: invalid signal name

Actual behavior (brush)

kill: invalid signal name

FWIW it also didn't kill the process as I expected it would

Expected behavior

Should translate the numeric argument name into the proper signal and send it

Additional context

No response

[jmquigs]

Thanks for fixing this. Note there is still an issue with kill -0 which is not parsed successfully:

~$ bash
~$ ps
    PID TTY          TIME CMD
  40467 pts/3    00:00:00 bash
 154866 pts/3    00:00:00 bash
 154878 pts/3    00:00:00 ps
~$ kill -0 40467
~$ echo $?
0
~$ kill -0 1234567
bash: kill: (1234567) - No such process
~$ echo $?
1
~$ brush
~$ ps
    PID TTY          TIME CMD
  40467 pts/3    00:00:00 bash
 154866 pts/3    00:00:00 bash
 154929 pts/3    00:00:00 brush
 154962 pts/3    00:00:00 ps
~$ kill -0 40467
error: kill: EXIT: invalid signal specification
~$ echo $?
1
~$ kill -0 1234567
error: kill: EXIT: invalid signal specification
~$ echo $?
1

kill -0 acts like an existing check to see if the pid actually exists. It seems trivial, but LLMs do generate bash code that uses it.
One spec describing the behavior is here (https://pubs.opengroup.org/onlinepubs/9699919799/functions/kill.html). That is the kill system call, but I believe the internal bash implementation as well as /usr/bin/kill are both expected to handle that case.

[reubeno]

@jmquigs Thanks for highlighting the special -0 case. We'll work on that separately.

(Contributions always welcome, of course 🙂)