On Sun, Nov 01, 2015 at 09:37:50PM +0100, Vincent Bernat wrote:
>  ???  1 novembre 2015 21:21 +0100, Willy Tarreau <[email protected]> :
> 
> >> > If nobody objects, I'd rather do that so that you don't have to add
> >> > a specific if/elif/endif just for this. What do you think ?
> >> 
> >> On the other hand, it is not really systemd-specific (and could be made
> >> not Linux-specific if it wasn't using /proc/self/exe to find its own
> >> name). It could be useful on other systems as well, as a way to turn
> >> HAProxy into a "regular" daemon. It could also be used to simplify the
> >> current init.d script. I could be renamed to just "haproxy-wrapper".
> >
> > Sure, I've thought the same, but always with the same question in the
> > end : "for what purpose ?". If we provide binaries which provide useless
> > features whose purpose cannot easily be explained to their users, probably
> > the binary in question should not be there. If it provides a real benefit,
> > then that's different, it's a matter of packaging choice and we should not
> > remove it by default. Given that I never needed it and still failed to see
> > a benefit outside of systemd, I'm inclined to think we don't need it, but
> > I could be wrong.
> 
> The current scripts work fine, so, yes, there is little interest in
> adding a binary to do the same thing. It could be useful with other
> supervisors (like supervisor, runit) or inside containers.
> 
> I don't use it outside systemd either, so it was just an idea. It can be
> removed from the default build until someone needs it outside systemd
> (and provide the patches to make it generic).

Jérôme made me realize it was removed already, the issue was the install-bin
target which was still trying to install it even when not built. So that's
a fix instead.

Cheers,
Willy


Reply via email to