TrueOS Making Use Of OpenRC Init System, Faster Boot Times

[Passso]

I really think Linux users should have choice and choosing the init system that works best for them. Why is systemd being rammed down our collective throats? As I use Debian, I feel that Debian devs should build packages that don't strictly have systemd as a dependency. On first install why not allow Debian users to select what init system they want?

Because 99% of users do not care about this subject ?

[MoonMoon]

I'll tell you this. Without users, distro is soon dead. Make your conclusions.

The developers themselves are users, too, and as long as they are interested and willing to maintain/develop a distro it will happen. If there are other users is largely irrelevant to many projects, it certainly helps to have more users (for example for bug reports), but it is by no means mandatory for a distribution to exist.
Of course, there are distros that will die without users, mostly those that need users for commercial viability (RHEL, SLED/S, Ubuntu, Slackware, ...), but there are also many distros that are not dependent on their users.

[aht0]

Because 99% of users do not care about this subject ?

When they run into systemd-related issues (especially if they did not before) they form a certain opinion. It does not even have to be a big issue. For example slight change in fstab would trigger systemd hanging up the boot process. "Average Joe" does not get what's wrong or how to fix it."Average Mary" even less. All he/she knows is - it used to work before this "newfangled crap". Even when there is written instruction on console STARING BACK AT THEM about how to skip and proceed. Which actually does not solve the issue but at least allows user to get out of hung-up boot. It's just single common issue.
I've also hit systemd not properly handling dhcp, not being able to shut down the system, journald eating the max cpu resources and so forth.. why the fuck should I tolerate it for a 10sec faster boot into desktop? It's like getting a race car which is really awesome, when it works.. but put it on the line with average Opel and by the end of 100km trip, only Opel reaches the fucking finish.

You do know about first impression forming definite opinions in inter-personal relationships? It carries over to software as well. You had issue with something, which replaced stuff that "used to work" - you will instinctively feel dislike for it. Older you get, more intense the feeling becomes.

The developers themselves are users, too, and as long as they are interested and willing to maintain/develop a distro it will happen. If there are other users is largely irrelevant to many projects, it certainly helps to have more users (for example for bug reports), but it is by no means mandatory for a distribution to exist.
Of course, there are distros that will die without users, mostly those that need users for commercial viability (RHEL, SLED/S, Ubuntu, Slackware, ...), but there are also many distros that are not dependent on their users.

Developers are also human beings. Personal life may interfere and take away free time needed for keeping distro running. You suddenly form family and get a kid, do you know how little free time you have left in your hands, all of a sudden? Simplest of all reasons.
Without viable pool of users, there shall be no new devs replacing the drop-offs, distro withers and dies off.

[Guest]

When they run into systemd-related issues (especially if they did not before) they form a certain opinion. It does not even have to be a big issue. For example slight change in fstab would trigger systemd hanging up the boot process. "Average Joe" does not get what's wrong or how to fix it."Average Mary" even less. All he/she knows is - it used to work before this "newfangled crap". Even when there is written instruction on console STARING BACK AT THEM about how to skip and proceed. Which actually does not solve the issue but at least allows user to get out of hung-up boot. It's just single common issue.
I've also hit systemd not properly handling dhcp, not being able to shut down the system, journald eating the max cpu resources and so forth.. why the fuck should I tolerate it for a 10sec faster boot into desktop? It's like getting a race car which is really awesome, when it works.. but put it on the line with average Opel and by the end of 100km trip, only Opel reaches the fucking finish.

Problems with dhcp and booting up were so common with sysv init. There's a long list with systemd advantages, but you should already know this. There's probably no bigger crap than sysv init.

You do know about first impression forming definite opinions in inter-personal relationships? It carries over to software as well. You had issue with something, which replaced stuff that "used to work" - you will instinctively feel dislike for it. Older you get, more intense the feeling becomes.

The worst first impressions with sysv init comes to mind.

Without viable pool of users, there shall be no new devs replacing the drop-offs, distro withers and dies off.

It seems there are many more users who prefer systemd over sysv init. You have to live with this.

[Guest]

There really is no choice when it comes to Linux init. Not for average user. For power user certainly. IF he/she has sufficient free time in hands..

Is there a choice when comes to BSD init? Can I use systemd?

[aht0]

Is there a choice when comes to BSD init? Can I use systemd?

Wrong person to ask from. Why don't you ask Poettering? Question implies claim that systemd incompatibility with BSD is somehow fault of the BSD flavors themselves. Where in fact Poettering has clearly stated that he is developing systemd for Linux and only for Linux. And btw, for Linux with GNU glibc to be more particular. You could of course fork systemd and port it to the BSD. Never mind the multiple hundreds of thousands of lines of code and ton of linuxisms to work around/remove. I am wishing you all the luck in your effort.

If you really want to hammer more on this, look at Slackware's default init. It's the same BSD's are using. Research-Unix-derived. It works, much better than SysV init actually.

[starshipeleven]

Where in fact Poettering has clearly stated that he needs a bunch of specific features found only on Linux and glibc.

Fixed. He never said he hates other libc, he said he needs features they don't provide.

Given what systemd is, it has to be as integrated as possible with kernel and use security and compartimentalization features that are there already.

So the choice of leaving out BSDs isn't exactly out of spite as you seem to imply. He had to choose between doing a half-assed but BSD-compatible job, or a great and Linux/Glibc only.

You could of course fork systemd and port it to the BSD. Never mind the multiple hundreds of thousands of lines of code and ton of linuxisms to work around/remove. I am wishing you all the luck in your effort.

The issues aren't "linuxisms" it's actual features that are either missing or different.

You'd be much better off rewriting it from scratch.

If you really want to hammer more on this, look at Slackware's default init. It's the same BSD's are using. Research-Unix-derived. It works, much better than SysV init actually.

That is stone-age stuff even if compared to OpenRC. Even embedded devices like LEDE/OpenWRT have a better initsystem that is reactive to config changes or other dynamic situations/conditions.

[bkor]

Wrong person to ask from. Why don't you ask Poettering?

What would Lennart know about BSD? Why do you even suggest he'd even know anything about BSD?!? You want choice in init systems but cannot even bother to answer someones question if there's init system choice in BSD?

[aht0]

What would Lennart know about BSD? Why do you even suggest he'd even know anything about BSD?!? You want choice in init systems but cannot even bother to answer someones question if there's init system choice in BSD?

Simple. Who is the lead dev of the systemd? Who decided it has to be "linux-only"? It's not about "he knowing about BSD", it's about zero effort on portability from his part, or let me rephrase it: he simply does not give a shit.

Before you still start bitching over it, go to github, check SLOC of FreeBSD and then do same check for systemd. Do you think someone wants to tackle porting this pile of code? Try working around around all the very linux kernel specific stuff Poettering/Red Hat&Co put in it in the first place.. why bother? Amount of code is equal to substantial portion of FreeBSD OS itself, if measured by lines of source code.

[DeepDayze]

Because 99% of users do not care about this subject ?

You never know.., but maybe for the Linux purists out there they may well care to have options. Systemd still is a work in progress and it has its bugs and glitches that can throw new users to a loop.