Runit as an official init system for GhostBSD

[Prince]

Runit, as a simple startupscript, is not out of keeping with the traditions of Unix. Runit is, also, the fastest. It is robust and avoids the instability and complexities of similar programs. Gentlemen, run it, please!

[ASX]

viewtopic.php?f=70&t=589

Beside you already got an answer, please consider that GhostBSD is open source, you are free to fork it and deliver your own implementation, including every crappy init system you may deem necessary to broke it.

When you will have done that, and demostrate that it work as intended, we will consider its inclusion here.

[Prince]

Crappy? This response surprised me. I know I shot myself in the ankle with the request for OpenRC. I ought to have never mentioned OpenRC, because, now, my true request will be forever overlooked. I had just came from a few sites that raved about it. In the past, I believed I have suggested some of the best software, out there, for GhostBSD. These were software that I believe would bring new users a rewarding experience. Runit is the init startup system of choice by a NetBSD maintainer, like another user mentioned, in Void Linux. BSDs are about the convergence of ideas. Sometimes, a project will move so forward that the developers of FreeBSD will request a special feature from another BSD project, like pf from OpenBSD, and other features from other BSD projects. Even the developers of the FreeBSD kernel don't overlook possibly rewarding suggestions, as crappy. I'm a strong proponent of suckless.org's philosophy. They have a list of software that rocks and sucks. They are very critical of poorly written software. Runit does not suck.

[ASX]

You misundertood: it is not runit that is crappy, it is the idea to change GhostBSD to adopt runit that is crappy.

[kraileth]

I ought to have never mentioned OpenRC, because, now, my true request will be forever overlooked.

It won't, you may be happy to hear that. I intend to try out several init systems and compare them. Don't take that as a promise, though, as I have no idea when the right time will be for it. But what I can promise is that I plan to include Runit in these tests. I actually liked it in the past but I admit that I didn't dig deeply into it.

You've put it right in the title: Runit as an official init system for GhostBSD may or may not happen. Runit as the official one probably not. We're trying to follow FreeBSD closely and what they chose will likely remain the default for us, too. But with packaged base there might be the option to offer another init system.

BSDs are about the convergence of ideas. Sometimes, a project will move so forward that the developers of FreeBSD will request a special feature from another BSD project, like pf from OpenBSD, and other features from other BSD projects.

Agreed. And that's why ASX - despite him openly taking a stance against changing the init system - has proposed that you invest some time and effort and build something to showcase Runit. If you have a working solution that proves to be beneficial to the project, chances of it's inclusion are not that bad!

You misundertood: it is not runit that is crappy, it is the idea to change GhostBSD to adopt runit that is crappy.

Why's that? Are you afraid of the additional work to test things or to leave the path of FreeBSD in such a central matter? Or do you know any details about Runit that would ruin GhostBSD that I'm not aware of?

[ASX]

Why's that? Are you afraid of the additional work to test things or to leave the path of FreeBSD in such a central matter?

Yes, I am. Not much about testing, but more because it is another departure from FreeBSD man pages.

More or less that already happened with Grub2, where every man pages that reference to add this or that to loader.conf become invalid ... similarly all references to services would become invalid and would need to be adapted.

The difference is that GRUB2 is what allow us to deploy hybrid images (unlike FreeBSD), so it was worth the pain, I cannot say the same for an alternate init system.

[kraileth]

Why's that? Are you afraid of the additional work to test things or to leave the path of FreeBSD in such a central matter?

Yes, I am. Not much about testing, but more because it is another departure from FreeBSD man pages.

More or less that already happened with Grub2, where every man pages that reference to add this or that to loader.conf become invalid ... similarly all references to services would become invalid and would need to be adapted.

The difference is that GRUB2 is what allow us to deploy hybrid images (unlike FreeBSD), so it was worth the pain, I cannot say the same for an alternate init system.

I see. It's funny, though, that GRUB2 is something that I don't have a lot of love for. But it does make sense to have it for the ISOs at least. Ok, documentation is a strong argument. Offering things that work is nice but doing things right would mean to have quality documentation, too. I've just checked: There's a man page but it fits on a single screen. At least the instructions to use it on FreeBSD sound quite simple and it's available as a package already so it would be easy to test.

[Prince]

Thank You! I hope this will help along the journey: http://blog.darknedgy.net/technology/2015/09/05/0/