1/  The s6 documentation has been updated, and is clearer than it was.  (Perhaps it hasn't changed but I understand more about how it works?)

2/  s6-linux-init, s6-svscan and s6-supervise are good to go.

3/  The new s6-rc is still WIP but the current ("old") one is good to go.

4/  There needs to be a distro-specific effort to write s6 service file(s) for each daemon - e.g., to translate the SysVinit ones into s6-compatible ones.

5/  However, work is also underway on a "turnkey" s6 installation process - "s6-frontend" - which will provide a default set of s6 service files.  However that is still a while away.

(I wonder how easy it would be to transplant Artix s6 service files into Devuan?)

Maybe this will be the "Year of the Linux Desktop" (without systemd)...

I'm hoping @andyprough can give use his thoughts on s6 v runit speed.

My 2 cents:  Devuan + runit is very fast to boot up on an Xfce desktop installation.  Artix + s6 was faster on the same PC.  I would guess (relying on my memory of when I had Artix installed, 1.5 years ago) that Artix + s6 felt like an order of magnitude faster, but both are so quick that we're only really talking about a difference of  a few seconds.  Might be a different story, in favour of s6, for a server installation?

Just in general I wondered what the difference was. 

Also, which is a larger package would be interesting too.

My 2 cents:  Devuan + runit is very fast to boot up on an Xfce desktop installation.  Artix + s6 was faster on the same PC.  I would guess (relying on my memory of when I had Artix installed, 1.5 years ago) that Artix + s6 felt like an order of magnitude faster, but both are so quick that we're only really talking about a difference of  a few seconds.  Might be a different story, in favour of s6, for a server installation?

dvnUsr wrote:

Laurent Bercot now has a sponsor ...

Update: the project has found a sponsor! Expect more news in this space in the months to come.
...
My estimate is that it needs about one year of work...

(https://skarnet.com/projects/service-manager.html)

That's some great news right there.

Begs the question - let's say Bercot finishes the work in the next year, and s6 really does turn into the "ideal service manager". What then for Devuan? Start offering it as an alternative init? Making it the default init would seem the appropriate response if it does all that he claims it will.

And what will happen with Debian and any distros that thumb their nose at their users and continue to push a technology with increasingly inferior performance on them? At some point the wrath of the users will become palpable.

Full disclosure - I used s6 for a lengthy test run of a couple months, and yes what you heard is true - it's so fast it should be illegal.

Is it faster and more lightweight than Runit? That would be surprising if so!

Begs the question - let's say Bercot finishes the work in the next year, and s6 really does turn into the "ideal service manager". What then for Devuan? Start offering it as an alternative init? Making it the default init would seem the appropriate response if it does all that he claims it will.

We're way ahead of you!
https://devuan.org/os/init-freedom

Laurent gave a talk at the 2018 Devuan Conference!

Laurent Bercot now has a sponsor ...

Update: the project has found a sponsor! Expect more news in this space in the months to come.
...
My estimate is that it needs about one year of work...

(https://skarnet.com/projects/service-manager.html)

That's some great news right there.

Begs the question - let's say Bercot finishes the work in the next year, and s6 really does turn into the "ideal service manager". What then for Devuan? Start offering it as an alternative init? Making it the default init would seem the appropriate response if it does all that he claims it will.

And what will happen with Debian and any distros that thumb their nose at their users and continue to push a technology with increasingly inferior performance on them? At some point the wrath of the users will become palpable.

Full disclosure - I used s6 for a lengthy test run of a couple months, and yes what you heard is true - it's so fast it should be illegal.

There was a discussion about s6 on the /devel mailing lists a while ago:

https://lists.alpinelinux.org/~alpine/d … pit.net%3E

Looks like Bercot managed to convince Ariadne and others between August 2020 and March 2021.

There was a discussion about s6 on the /devel mailing lists a while ago:

https://lists.alpinelinux.org/~alpine/d … pit.net%3E

Very interesting reading, with a big colorful comparison table to show the exact points where systemd fails and where openRC is inadequate, and where s6-rc is currently lacking functionality.

Bercot states his plans on how he can complete the building of s6-rc into a service manager that fulfills all the needs of a distribution, and that it will be the first tool to ever accomplish this feat. Says he needs a year and some funding, and that since s6 is the darling of the docker space, he expects that funding the project will be a huge benefit to some large cloudy companies. It's basically a funding proposal, but with a ton of useful data.

Alpine has come out strongly in favor of getting this project funded, and says their long-term goal is to adopt Bercot's s6-based solution: https://ariadne.space/2021/03/25/lets-b … or-alpine/