Mirror of forum.nim-lang.org

1949 :: How to send SIGKILL in Nim?

• luntik2012 (orginal) [2016-01-18T14:37:16+01:00] view original

  Documentation here http://nim-lang.org/docs/osproc.html says that

  proc terminate(p: Process)
  Run

  sends SIGTERM

  and

  proc kill(p: Process)
  Run

  sends SIGKILL.

  But in sources I see

  
  proc kill(p: Process) =
      terminate(p)
  Run

  There is no differences when I run them too, so how to correctly send SIGKILL to a process?

  Nim Compiler Version 0.12.1 (2016-01-18) [Linux: amd64], Ubuntu 15.10

---

• luntik2012 (orginal) [2016-01-18T14:41:24+01:00] view original

  @luntik2012, you have to call waitForExit() after terminate() and before close() :) I'm so sorry

---

• Jehan (orginal) [2016-01-18T18:40:45+01:00] view original

  There are two definitions of kill() in osproc.nim; you're looking at the one for Windows. The second sends SIGKILL.

---

• luntik2012 (orginal) [2016-01-18T21:51:28+01:00] view original

  @Jehan, Thank you. But I still have zombie processes when I use kill() without waitForExit() on Linux, and they disappear only after my program's death.

  I don't know whether this is normal behavior. It's not a problem, but I wonder why

---

• Jehan (orginal) [2016-01-19T00:49:13+01:00] view original

  Zombie processes are a completely unrelated issue. Basically, because the kernel cannot tell the future, it doesn't know whether you may need the result (remember that wait() and friends return information) or may be using waitpid(), so it can't throw that data away or reuse the process id. The zombie process is basically a placeholder for that data and occupies a process slot until the kernel can positively know that you are done.

---