Re: Challenge to you: Voice your concerns regarding systemd upstream
Am Freitag, 26. September 2014, 14:51:01 schrieb Miles Fidelman:
> John Hasler wrote:
> > Miles Fidelman writes:
> >> the technical committee selects takes a vote that essentially imposes
> >> systemd on all of the upstream developers and packagers
> >
> > Where the hell do you get that from?
>
> Isn't that effectively what happened?
>
> If I'm an upstream developer, and I want my stuff to run on Debian, I
> now have to include systemd init scripts (or the packagers do).
>
> Sure, it's "voluntary" - but not really.
Oh, the same way I could say:
I am forced to write init scripts for a package. As I recently just did:
And guess what: On writing a debianized variant of the atopacctd initscript
where the upstream initscript actually caused lintian warnings I clearly
learned about the limitations of it. Look at it [1] and tell me how you like
that it unconditionally kills any process named atopacctd and the PIDFILE
variable is not even used anywhere in the script.
I´d have a clear word for that: crap.
Now you can argue upstream needs to implement PID file handling for a double
forking daemon. But I make is a case now that systemd needs *less* care of
upstream, rather than more. It forces *less* on upstream than sysvinit for
best practice. With systemd upstream doesn´t have to change the actual
implementation of the daemon. And even without a service file it will just use
the initscript anyway, with the same limitation then.
[1] https://github.com/teamix/atop-debian/blob/master/debian/atopacct.init
Ciao,
--
Martin 'Helios' Steigerwald - http://www.Lichtvoll.de
GPG: 03B0 0D6C 0040 0710 4AFA B82F 991B EAAC A599 84C7
Reply to: