I think I know what you mean: if it ain't broke, why fix it?
To get some practical experience of systemd, I'm dual-booting Debian "Jessie" with Slack on my Thinkpad T410. Debian, because it's the other distro I'm most familiar with. And I've glanced through the Arch wiki page on systemd. So I'll seem to know what I'm talking about when I oppose it...maybe.
It's Linux, Jim, but not as we know it.

I would think because "if it ain't broke, why fix it" mostly applies to static systems. If your car isn't broke you don't fix it. But the Linux environment, like so many others, is not a static environment, it evolves, even more since the underlying hardware evolves and the stuff we do with it evolves. Sysvinit was not broken, but nonetheless people that wanted more from it saw its limitations and added the sysvrc stuff to it. This was not enough for some people, so they invented stuff like daemontools, Runit, s6, OpenRC, Upstart and systemd.
It is like: horse carts were not broken, but nonetheless we replaced them almost everywhere with alternatives that overcame their limitations. When we ran into the limitations of 16 Bit CPUs we replaced them with 32 Bit CPUs and later with 64 Bit CPUs. That does not mean that 16 bit CPUs were broken, in fact, where those limitations don't matter we still use them (even 8 Bit CPU are still used in many places). No, 16 Bit CPUs weren't broken, but that doesn't mean that we should maintain universal support for 16 Bit CPUs in modern desktop PCs. Cars fit most modern requirements better than horse carts, 64 Bit CPUs fit many modern requirements better than 16 Bit CPUs, systemd fits many modern requirements better then sysvinit/sysvrc or sysvinit with BSD style scripts.
Some day in the future we will replace cars with a better alternative, we will replace 64 Bit CPUs with a better alternatives, we will replace systemd with a better alternative, oh heresy, we might even replace Linux with a better alternative. That doesn't mean that you can't use cars, 64 Bit CPUs or systemd anymore, it just means that you will have to look harder for spare parts (in case of cars or CPUs) or developers that support your use-case. That is what I am calling for: there is nothing wrong with using sysvinit if it fits your use-case, but when more and more developers look at an alternative that they deem to be better (note: it doesn't matter at all if it is actually better) and start to use its functionality than you will have a harder time to get support for sysvinit based systems, which will eventually come to a point where most of the work simply will have to be done by those that want to use it.

3 members found this post helpful.

Oh yeah, I understand the need for progress. But in this particular case, my objection - as I've said before - is that the systemd cabal want to remake Linux. Not just the init system. They want some sort of unified package management, amongst other things. It was all in Poettering's blog that was linked to in several threads. If they feel that Linux is in need of such drastic changes, why not start from scratch with a new OS? Why don't they go and help Microsoft where the need for improvement is greater?

1 members found this post helpful.

Probably because:

• Even RH can't afford that (bear in mind that HURD, in active development since 1990 is not yet production ready as it still miss sound and USB support, according to what I have recently read.
• They do not need to do that as they can use an existing kernel with a well documented API, known from the developers.

The reason they don't care to stay POSIX compliant on the API level is that they just target Linux and thus don't care for portability as LP has clearly stated.

This choice may please neither you nor me, but seems logical and consistent considering their aim. That's their choice anyway and I don't feel that they ought me anything that could forbid them to go their way.

Sorry! Define Linux Operating System, please!

Or, maybe you talk about something like a Linux kernel driving an GNU (BUT remember, GNU is Not Unix!) user-land, or, much better (around 25000% as users), about a Linux kernel driving an Android user-land?

I feel like the guy that finally say the truth about Santa Claus...

Well, buddy, there is no thing such like Linux (Oh, UNIX, yeah, baby!) Operating System. Its just a kernel, and something similar to UNIX userland, provided by GNU. Nothing more.

If you really want UNIX, you should buy UNIX and don't play with something that NOT guarantee to be even SIMILAR.

BTW, after all, GNU have a default INIT system? I do not remember, right now. Just courious if we really we respect in the Taliban Style right now, the The GNU/Linux, err .. UNIX words. Or we are already heretics?

OK, the systemd cabal's plans involve a makeover/takeover of GNU/Linux. But that issue has been debated on already in several threads. I'm as much in favour as you are in using GNU/Linux, but I think we're in a minority. So I just use Linux, except in threads about GNU/Linux vs Linux. OK?
Where did I say I wanted UNIX?
P.S.
Seen this?
http://pedrocr.pt/text/how-much-gnu-in-gnu-linux/

1 members found this post helpful.

'Progress' for progress sake, or not......... hmmm

Something fundamental like changing diesel/petrol cars for e.g. hydrogen fuel or electric where an obvious benefit exists for so may reasons, developments in computers are not. Well, I will stand corrected for bespoke/technically specific cases of computing perhaps that are required for progressing other more worldly important requirements.

Always I keep in mind for initial test, "what or who is it for and why is it of fundamental benefit to those who or what parts".

Most people, the majority (unless gaming and then there are consoles for that, or serious vid. editing) will never use the power of even the most modest chip of today so I am told. 64-bit, 32-bit I'm not sure most people would notice or understand.

As I acknowledged, years ago the 'world of linux' was looking for improvement. Where it is now is pretty good and I find it hard to envisage what more progress there is for the majority experience.

Linux can run on the most modest/heritage ;-) of computers.

As I asked, what is the end-game and when is enough [alleged] 'progress'. 'Chasing the next iphone release' (not literally) just seems to me to be a waste of earthly resources and without serious scientific merit (in my opinion)....so the next iphone provides a few more pixels, a bit more swipe this and that or allows you to talk to it.

Of course the [greedy] profiteers would have you buy the next computer, smart phone, cerebral-chip interface controller thingy.......it makes them rich. But does it really make your 'word processing' better. I'm only using silly examples perhaps.

I just do not see a reason as yet that I understand, to why systemd (in his case) will change **my** experience or allow software developers to create something different with their software packages that it will be a fundamental wow factor, or to meet an obvious beneficial requirement coming in the near or far future.

If there is a strategic point and systemd is fundamental for it that is so better than what is now....what is it?!

(by the way, I am not for or against systemd - systemd just happens to be the contentious subject on this thread as I read it).

On the other hand, pursuing systemd does seem to be meaning wholesale change and cost as a lot see it for little gain.

I don't know. I'm just not coming to this from a non-techie view....more a why waste resource, money and effort point of view if much of everything now works to the satisfaction of a lot of people set against the cost of changing for a real benefit not a lot will/can see.

In the end, what will be will be I suspect.

Thank you for the opportunity to share with you.

1 members found this post helpful.

That link say:

Looks like I do not have the rights to known the True Truth about GNU/Linux... Reserved just for the Stallman friends?

No, they don't want to remake Linux. Linux is just a kernel, not an OS. But they want to follow their vision how a distribution should work, just like the Debian people do it, the Suse people do it, the Gentoo people do it and just how Mr. Volkerding is doing it. Of course their vision implies changes that all not people find good, pretty much the same as we see with the change from Gnome 2 to Gnome 3 or KDE 3 to KDE 4. That is natural, their vision does not have to be your vision.
Have you read that blog post? I found it pretty reasonable, Mr. Poettering shows the existing solutions, points out the problems with those for different use cases and tries to come up with a solution that would fix those short comings. That is what engineers do, problem solving. He just proposed his ideas (or in this case, the ideas of him and his colleagues) and showed how BTRFS could be used to implement them. It should be clear for anyone that this scheme, if would become reality, would automatically enforce the usage of BTRFS, which at the same time disqualifies it for widespread usage. No major distribution would want to tell their users that from now on only BTRFS could be used, so it should be pretty clear that this was rather an academic exercise than an actual proposal what would be implemented in systemd.
But even if this style of package management would be implemented in systemd that doesn't mean that any distro automatically has to accept it. Nothing hinders openSuse to simply use RPM packages instead, nothing prevents Debian from using APT for package management, nothing hinders anyone to adapt it as they see fit. I personally would prefer to use the classical package management with parts of his proposal on top, only for third party packages that are not part of your distribution or not trusted (Steam, Skype, ...), though I would deem overlayfs to be a better fit for this.
Why they don't go to Windows should be pretty clear, Windows is not open source, so they can't just make changes. And why should they start a new OS? There is no need for that, the kernel works fine, the GNU userland works fine, they just replace what they think should be replaced. Isn't that exactly what open source is all about?
I mean the "here do you have the toolchain, there you can find the source code, do with it whatever you want unless you violate the license" stuff. They can do that as everyone else can do that, there is no inbuilt limitation like "you aren't allowed to change these things". They are entitled to their own vision of how a Linux system should work and open source gives them the tools to follow their vision. You don't have to like their vision, not at all, you don't have to follow it, no one has to, nobody can force anyone to do so, but you also can't prevent them from following their vision and you can't prevent others from sharing their vision.
Of course, in case of systemd, their vision is shared by so many people that it became the dominant vision, which makes it harder for others to follow their own vision, but not impossible, only harder. That only means that people with a different vision have to work harder for their vision to be viable, not that other visions are somehow not allowed or less worth or should go away.

5 members found this post helpful.

It shows a pie chart of percentages of total lines of code in a GNU/Linux distro, with the kernel at 9% and GNU at 8%.
But, where would Linux be without GNU?

OK so if im reading that paragraph right you are saying that systemd is of no current benefit, may be of some benefit in the future and if developers develop programs in a systemd environment they will need systemd for their programs to function.

That all makes sense to me and i agree. But i spent some of my life in sales and this "future benefit" stuff smells like a con to me.
Is it not apparent to everyone that this is about a large company trying to obfuscate and control linux for the sake of sales and support revenue?

3 members found this post helpful.

Maybe it's the right time to invest some money in NYSE: RHT then

But if instead for any reason you want their attempt to fail, I am afraid that just posting your thoughts in this thread be not the most effective way...

Yes, I read it when it first appeared, and have skimmed through it again. Here's a link to it for anyone who hasn't read it:
http://0pointer.net/blog/revisiting-...x-systems.html

How a distribution should work shouldn't have to follow one set of guidelines like systemd. That's extremely shortsighted in the fact GNU/Linux has always been about choice, and making choices. That choice has extended for the longest time down to the core tools of what makes a system work, which is a kernel, module loaders, a bootloader, a simple set of init scripts, a shell, and some basic tools to manage things like files, file systems, logs, a file viewer, and an editor. My question is why should that have to change from something so simple in design, to something complex?

The point of revisiting the way we put systems together isn't about that, it's about creating an entirely new userland that's compatible with GNU tools to an extent. Progress for the sake of progress, isn't progress, it's regression.

Sysvinit had some flaws that simply couldn't be addressed like service supervision, but along came add-ons like daemontools and perpetrator that allowed sysvinit to exist as the control matrix, but passed along via execve() to an init-like controller such as daemontools and perpetrator to manage the daemons and other system services and then when completed, execve() back to init to allow the system to be halted or rebooted. And we've had daemontools since the late 1990s so the excuse of not having a better way to manage daemons and boot the system faster using parallelization is falsified on the highest levels. However, things like instances where a service is needed or unneeded have always been subject to other subsystems. For example, Consolekit could always be started either the init system, or it could be triggered via an X session if the desktop supported using it.

This is what I don't understand of how people say "this hasn't existed before" but in reality, it always has, but either went unnoticed or unused, so how can systemd's design be better for creating systems when nothing it does hasn't been done before.

1 members found this post helpful.

A very thoughtful blog post. what's is wrong about it?

BTW. Funny to Die! For connoisseurs, the round image in the left is translated (from Arabian), "The (God/Allah) Messenger".