On Sun, 3 Jul 2016 17:55:42 -0300 Felipe Sateler <fsateler@debian.org> wrote: > On 3 July 2016 at 16:44, Samuel Thibault <sthibault@debian.org> wrote: > > Felipe Sateler, on Sun 03 Jul 2016 17:38:58 -0300, wrote: > >> On 3 July 2016 at 16:33, Samuel Thibault <sthibault@debian.org> wrote: > >> > Felipe Sateler, on Sun 03 Jul 2016 17:31:14 -0300, wrote: > >> >> I think the daemonization needs to move later as well: > >> > > >> > Why? > >> > >> Because the usual protocol is "fork when ready". > > > > ? I would rather expect it to be "parent exits when ready". > > Indeed. But using daemon(3), "fork" and "parent exit" time are the same. Right. I think this is the problem. Afair, systemd will try to read the pid file as soon as the parent exits. That should happen *after* the forked daemon process is ready and has written the pid file. See man daemon(7). This looks like a genuine bug in espeakup to me, and not something which we can workaround in systemd. Note also this comment in man daemon(7): "The BSD daemon() function should not be used, as it implements only a subset of these steps." -- Why is it that all of the instruments seeking intelligent life in the universe are pointed away from Earth?
Attachment:
signature.asc
Description: OpenPGP digital signature