The Debian init system general resolution returns
The Debian init system general resolution returns
Posted Oct 20, 2014 16:15 UTC (Mon) by mgb (guest, #3226)In reply to: The Debian init system general resolution returns by nhippi
Parent article: The Debian init system general resolution returns
All the necessary code to support sysvinit already exists in Wheezy. At most it might need to be recompiled.
We do however need a GR to prevent a very small number of people from harming Debian by removing sysvinit support.
Posted Oct 20, 2014 16:31 UTC (Mon)
by johannbg (guest, #65743)
[Link]
No but as has been repeated on numerous occasion is that you do need to write patches for other applications and application stacks for example read this [¹] and answer the question to whom does the burden fall having to come up with, carry and maintain the alternative to systemd logind which is provided to support the $other_initsystems in Debian for this particular application stack?
1. http://blog.martin-graesslin.com/blog/2014/10/libinput-in...
Posted Oct 20, 2014 16:37 UTC (Mon)
by mathstuf (subscriber, #69389)
[Link] (12 responses)
Is there an instance of this that you can link to? Or is it a strawman?
Posted Nov 9, 2014 7:20 UTC (Sun)
by mgb (guest, #3226)
[Link] (11 responses)
Gnome 3 of course needs systemd but Gnome 3 is totally avoidable so it is not a problem.
The most obstructive borged package is probably policykit-1. A separate systemd-policykit-1 could trivially have been created while leaving the working package in Jessie but instead the working package was replaced.
Another egregious borgee is bsdutils. The good version works fine with both syslog and journald but a ridiculous function was added to Jessie which bloats Essential with libsystemd0 and more.
In all we currently pin five Wheezy packages and thereby avoid both systemd and systemd-shim in Jessie. We will have to pin or replace a half dozen more packages in order to avoid libsystemd0. These numbers may change depending on the outcome of the GR between the borg and the free DDs.
Posted Nov 9, 2014 8:10 UTC (Sun)
by Cyberax (✭ supporter ✭, #52523)
[Link]
I'm a rabid systemd fan and I've been biting people all the time, infecting them left and right with systemd-virus.
However, I have a slight problem with sysv-phobia. How can I excise every last trace of it from my systems? I tried removing sysvinit-utils, but it totally b0rked my system by removing everything!!!
Why do you insist on forcing megabytes of dead SysV weight on everyone???
Posted Nov 9, 2014 13:46 UTC (Sun)
by mathstuf (subscriber, #69389)
[Link] (9 responses)
As for polkit1, if parallel packaging efforts would have been possible, why not package it yourselves ("you" being the "we" you, mgb, keep referring to)? Here, you would have forced a rename of a package where the old one may well have been unmaintained /anyways/ and potentially dropped unless someone stepped up. This route seems much more manageable than threatening a fork of the distro where you claim system-wide breakage when the actual defect set is of size 1 and would potentially be solvable with a parallel packaging effort. It also seems small enough to not deal with the inflammatory threads part of the GR[2] so much and burn so much time and energy from the entire DD community.
[1]One example that comes to mind is the set of packages for the current Fedora theme (artwork, icons, and cursors) when I'm completely happy with bluecurve. So I'm forced to have adwaita (at least today) when I don't use it at all with no loss of functionality.
Posted Nov 9, 2014 15:28 UTC (Sun)
by mgb (guest, #3226)
[Link] (8 responses)
libsystemd0 is still a problem. It bloats Essential and is itself deliberately monolithic and therefore a roadblock to F/LOSS progress.
We're evaluating alternatives to decide where to go next. Jessie looks possible but increasingly inconvenient. We need to move before Stretch and my guess is we'll be on a more modern distro before Wheezy EOLs.
Posted Nov 9, 2014 15:39 UTC (Sun)
by Cyberax (✭ supporter ✭, #52523)
[Link]
libsystemd0 is a small 250kb library. Hardly a 'bloat'. And it certainly doesn't interfere with you using upstart, sysv or whatever over crap you so want to use.
Posted Nov 9, 2014 17:16 UTC (Sun)
by jspaleta (subscriber, #50639)
[Link] (6 responses)
And you keep using the royal we... its just a bit creepy.
Posted Nov 9, 2014 17:33 UTC (Sun)
by mgb (guest, #3226)
[Link] (5 responses)
More modern distros may offer systemd as an option but they don't limit themselves to that fossilizing plumbing.
> And you keep using the royal we... its just a bit creepy.
"I" am posting here. "We" are working on this. "You" work alone?
Posted Nov 10, 2014 9:36 UTC (Mon)
by anselm (subscriber, #2796)
[Link] (4 responses)
It would be interesting to find out who will actually be providing such a “more modern distro”. It seems that the current mainstream Linux distributions pretty unanimously won't.
Going with a niche distribution of course carries its own risks, including the one that the distribution in question, if avoiding systemd is not its major selling point, will eventually also include stuff that depends on libsystemd0.
Posted Nov 10, 2014 18:08 UTC (Mon)
by mgb (guest, #3226)
[Link] (3 responses)
Then Microsoft used EEE to seize power. So we moved forward with F/LOSS.
Then systemd borged many of the old distros. "You could try to escape us by forking but haha you couldn't possibly muster the resources." Well that might be true if we limited ourselves to old technology but there is newer and better technology.
Central Command & Control always stagnates. Free distros are the next future.
Posted Nov 10, 2014 18:19 UTC (Mon)
by jspaleta (subscriber, #50639)
[Link] (1 responses)
Can we just shelve this conversation until y'all are ready to present something concrete that uses the newer and better technologies. The coy references to unspecified better technologies is getting sort of tiring, it and sounds a lot like mealy mouthed vaporware hype. I don't even know wtf technologies you are talking about at this point as you haven't actually pointed to candidate technologies. I'm glad your excited by unspecified new and better technologies. I look forward to evaluating these new and better technologies at whatever point in time you are prepared to make them public for critical review. Good day.
Posted Nov 10, 2014 19:45 UTC (Mon)
by mgb (guest, #3226)
[Link]
Emerge, OBS, and decentralized distribution are some of the technologies already providing more freedom than Redhat.
F/LOSS is progressing beyond Redhat-enforced conformity.
Posted Nov 10, 2014 18:57 UTC (Mon)
by bronson (subscriber, #4806)
[Link]
If true, this would require an immense amount of effort. What would their motivation be? And who exactly is "they"? You seem to implying an IBM or Microsoft Borg conspiracy... Is there a shadow group acting behind the scenes? Is there money behind establishing a service manager monopoly?
Also, "central command and control" describes this very resolution, doesn't it? Wouldn't you want to give Debian developers the freedom to make their own decisions?
Genuinely curious.
The Debian init system general resolution returns
The Debian init system general resolution returns
The Debian init system general resolution returns
The Debian init system general resolution returns
The Debian init system general resolution returns
[2]Maybe the GR itself is useful; I make no assertion here.
The Debian init system general resolution returns
The Debian init system general resolution returns
The Debian init system general resolution returns
The Debian init system general resolution returns
The Debian init system general resolution returns
More modern distros may offer systemd as an option but they don't limit themselves to that fossilizing plumbing.
The Debian init system general resolution returns
The Debian init system general resolution returns
The Debian init system general resolution returns
The Debian init system general resolution returns