On Thu, Apr 13, 2017 at 11:18 PM, Adam Williamson
On Thu, 2017-04-13 at 22:23 -0400, Nico Kadel-Garcia wrote:
> On Thu, Apr 13, 2017 at 3:17 AM, Pierre-Yves Chibon <pingou(a)pingoured.fr>
> > On Thu, Apr 13, 2017 at 07:04:57AM -0000, Farhad Mohammadi Majd wrote:
> > > > On Thu, 2017-04-13 at 04:05 +0000, Farhad Mohammadi Majd wrote:
> > > > Because we think it's fundamentally a wrong approach.
> > > > ....
> > > > It's not really a question of resources, but of not thinking this
> > > > the correct approach.
> > >
> > > What is the correct approach?
> > You seem to have missed this in Adam's email so here it is:
> > i) just getting things right so we don't need a giant pile of
> > configuration tools
> > ii) tools written at more appropriate layers, mainly desktop
> > environments
> This is how you wind up with systemd.
Um. What? systemd has nothing to do with this, and...
Ignoring this approach is how you wind up with systemd. I'm sorry it
was unclear that I was agreeing with *you. It's the idea that one tool
should provide control of all subsystems and manage all subcomponents
throughout the operating system. We're seeing complications of it
where that new tool takes on managing new components with new
approaches, and becomes very difficult to prevent from managing or
making mistakes with things other developers, more involved with those
other components, didn't ask for.
> One overarching tool
...I didn't say anything about 'one overarching tool'.
> It was tried for
> Fedora years ago, and discarded with a passion.
What 'was tried for Fedora'? What are you referring to?
Linuxconf, which was much like YaST. It was in some of Red Hat's 8 and
a few other pre-RHEL releases, it didn't work very well, and was
maintained for some time in a number of third-party Fedora
repositories. From a casual archive search, it doesn't seem to have
ever made it into the actual Fedora releases.
> > Now if you think this is wrong, maybe you could give a few
examples and bug
> > reports so that there is something tangible to discuss. Otherwise we're
> > discussing around opinions and I doubt that it would lead to anything
> > productive.
> YaST DNS management, limited, painful, and difficult to tune for valid
> site specific configuration such as using the same .zone file for
> multiple domains.
> YaST package management, which attempts to incorporate management of
> non-RPM proprietary tools alongside RPM package management, mishandles
> the proprietary tools, and doesn't report conflicts among them though
> it's alleging to manage both.
> YaST mishandling of timezone confurations. (They may have fixed that one.)
> YaST printer configuration. Mind you, that one's always been painful.
Um. You realize I was saying that we *don't* have anything like YAST
and we explicitly chose not to, right? You seem to be confused about
who's arguing what.
Sorry for confusion. I lost track with the indents on that one. I'm
pretty passionate that Fedora is using a better approach. YaST was
*not* my friend when I had to deal with it, nor was the old linuxconf
tool. Do not get me *going* on what its X configuration tools did with
proprietary NVidia drivers, trying to intermingle them with RPM based
X updates and ignoring the chaos.