bug#46403: SIGINT ignored when using system*

["Ludovic Courtès" <ludo@gnu.org>]

Hi!

Ricardo Wurmus <rekado@elephly.net> skribis:

> I execute commands in a loop and wish to be able to interrupt the loop
> with SIGINT.  Here’s the first attempt:
>
>     guile -c \
>       '(for-each (lambda (n)
>                   (display n)
>                   (system* "sleep" "3"))
>                  (list 1 2 3 4))'
>
> At no point will this program be interrupted by SIGINT.  Strace reveals
> that SIGINT is in fact received and the sleep is interrupted, but the
> wait is restarted immediately afterward.

‘system*’ explicitly ignores SIGINT:

--8<---------------cut here---------------start------------->8---
  /* Make sure the child can't kill us (as per normal system call).  */
  scm_dynwind_sigaction (SIGINT,
                         scm_from_uintptr_t ((uintptr_t) SIG_IGN),
                         SCM_UNDEFINED);
--8<---------------cut here---------------end--------------->8---

Indeed, system(3) says:

   During  execution  of  the command, SIGCHLD will be blocked, and SIGINT
   and SIGQUIT will be  ignored,  in  the  process  that  calls  system().
   (These  signals  will be handled according to their defaults inside the
   child process that executes command.)

(I did some archaeology and found it has always been this way since Rob
introduced ‘system*’ in 0db17ef9abd59da51ebc30d90fb2dc482b02a4a1.)

> Is this by design?

I think so!

Ludo’.