Re: [dev] A replacement for at.

From: Mattias Andrée <maandree_AT_kth.se>
Date: Fri, 1 Jan 2016 19:14:42 +0100

On Fri, 01 Jan 2016 19:05:50 +0100
Kamil Cholewiński <harry666t_AT_gmail.com> wrote:

> > sat will create a PID file in $XDG_RUNTIME_DIR.
>
> PID files are a flawed concept, race conditions and
> everything.

Well. But they are standard. satd does not use the
PID file to determine whether it is running, it uses
flock. There is still a possible, less important, race
condition, but this can be fixed if one thought it was
worth it.

>
> > How could supervision be made mandatory without making
> > it tied into one solution.
>
> Simple: just don't fork, don't write PID files, don't do
> anything. You can turn /bin/cat into a daemon under
> supervisord, runit, svscan, systemd, nohup &, or any
> other of the million solutions, as long as you don't try
> to outsmart it.

Would this not require that the user configures satd to
be started by the supervisor? I would not want this because
I would almost never it.

'satd -f' will cause it not to fork, or do other things
that a traditional daemon would do.

>
> K.


Received on Fri Jan 01 2016 - 19:14:42 CET

This archive was generated by hypermail 2.3.0 : Fri Jan 01 2016 - 19:24:09 CET