Page 1 of 1

finit - init system

Posted: Tue Jan 19, 2016 11:37 am
by machinebacon

Re: finit - init system

Posted: Tue Jan 19, 2016 2:24 pm
by wuxmedia
ruby bootloader? Why not, can use /bin/yes as PID 1 :)
Fascinating init walkthrough though.

Re: finit - init system

Posted: Tue Jan 19, 2016 3:11 pm
by rhowaldt
nice write-up indeed, like how he doesn't bash back and forth but just proves some points both against sysvinit and systemd.
what i don't get is why he does such a basic, low-level thing as making an init-system, then proceeds to run all manner of bloat on it. perhaps this is for testing purposes, to be sure his init works with all the "modern" stuff? dunno, found it weird: "i program an init from scratch to then run SLiM"

Re: finit - init system

Posted: Tue Jan 19, 2016 3:22 pm
by wuxmedia
yeah - he certainly wasn't railing against one or the other, kinda showing how 'easy' it is to knock up an init.
ie run something as pid 1. mount some things, make some files/dirs. couple a sockets and you're off

Although he did take tools from systemd

Re: finit - init system

Posted: Tue Jan 19, 2016 4:07 pm
by machinebacon
I don't think he wanted to run SLiM - he wanted to make a point that even a graphical login is possible, before people start bitching that "yeah this is just like busybox init, and then you get stuck, asshole". He proves that X can be started even via session manager ;) He clearly stated it right at the beginning, /bin/yes would be the proof of concept only, and in order to make people STFU he completed the task. Of course there is no point in running SLiM, and what he lends from systemd (udev) can also be replaced with mdev -- the problem is that most distros don't have it in the repositories, so I guess installing it would be an extra step that he didn't want document because it is not the scope of the project.
Slack's and Arch's (former) initscripts were not much more complicated than this one, so the argument of 'adding bloat on top of a script' is always a bit relative :) The absolutely stripped version is kernel plus busybox: you can try to run that shit and then you'll quickly come back to something more 'bloated' :D

Re: finit - init system

Posted: Tue Jan 19, 2016 4:08 pm
by simgin
Brilliant share Julius, bookmarked it. Very good write up!
Now even my meager brain activity is getting a better picture of it, sweet.

cheerio

simon

Re: finit - init system

Posted: Tue Jan 19, 2016 7:11 pm
by Snap
Thanks for the links. Really interesting.

Re: finit - init system

Posted: Wed Jan 20, 2016 11:13 am
by rhowaldt
thanks for that explanation Jules, this was my inkling as well, simply because it didn't make sense, philosophy-wise. so now it does :)
of course the bloat is all relative - remember who wrote the wiki article on that? :D

Re: finit - init system

Posted: Wed Jan 20, 2016 12:15 pm
by dkeg
I like the format and conveyance of the article. Might be another snowy weekend experiment. Now that my studying is done, I have some spare time.