Re: interesting claims

From: Laurent Bercot <>
Date: Tue, 30 Apr 2019 08:22:50 +0000

>"suckless init is incorrect, because it has no supervision capabilities,
>and thus, killing all processes but init can brick the machine."
>a rather bold claim IMO !
>where was the "correct" init behaviour specified ?
>where can i learn how a "correct" init has to operate ?

For instance:

  Or, as Guillermo mentioned, several posts in the ML archive.

  init is a subject that little study has been put into (though it
is also the subject of a whole lot of talk, which says something
about whether people would rather talk or study). But I think you'll
find that things are different around here.

>or is it true since s6-svscan already provides such respawn
>capabilities ? ;-)

Do not mistake causes for consequences. Things are not correct
because s6 does them; s6 does things because they are correct.

>there is actually NO need for a "correct" working init implementation
>to provide respawn capabilities at all IMO.

Then you are free to use one of the many incorrect inits out there,
including sinit, Rich Felker's init, dumb-init, and others. You are
definitely not alone with your opinion. However, you sound interested
in process supervision, which is part of the more general idea that a
machine should be made as reliable as possible *at all times* and
*under any circumstances*; if you subscribe to that idea, then you
will understand why init must supervise at least 1 process.

>so this looks like a rather artificial and constructed argument for
>the necessity of respawn functionality in an init implementation IMO.

  Maybe you've never bricked a device because init didn't respawn
anything. I have. The "rather artificial and constructed argument"
happened to me in real life, and it was a significant inconvenience.

Received on Tue Apr 30 2019 - 08:22:50 UTC

This archive was generated by hypermail 2.3.0 : Sun May 09 2021 - 19:44:19 UTC