May 2, 2017 12:00 PM, "Laurent Bercot" <ska-supervision_at_skarnet.org>
wrote:
> Old? Let's see...
> The MTA used by this very mailing-list is netqmail-1.06, i.e.
> qmail (latest version released in 1996) with user-contributed
> patches, the latest of which is from 2005.
> So yeah, that's 12 years old software. It could be going to middle
> school right now.
> And it works. I think the usual motto is "if it ain't broken, don't
> fix it". :)
To be honest, it's surprising to say the least to hear of a program
that fully meets its purpose, is complete and has no bugs. The "ain't
broken, don't fix it" motto is usually invalid because of this, as
people demand new features, related tools and libraries get updated,
code becomes larger and taking care of multiple tasks, programmers
create bugs, and so on. You know better than I do.
> The latest runit version is from 2014 - that's only 3 years old.
> It's not even kindergarten age, come on. Did you know that some Linux
> distributions still ship GNU make 3.81, a version of GNU make that is
> 11 years old?
No, I didn't, usually I just do `./configure && make && make install`
and it just works. Despite of this, GNU make 4 exists, and it has been
last updated on June 2016. So perhaps software is improvable after all.
> systemd, on the other hand, hasn't left the maternity hospital - but
> how could it? it's such a fragile baby, it needs constant attention,
> it can barely leave the incubator. If you ask me, it should never
> have left the womb in the first place. :-Ă
That fragile baby must be the child of a demigod, as it is one of the
most used inits out there already, and most complaints come from either
its (somehow working and usable) centralized design, having to type
Ctrl-Alt-Del 7 times in less than 2 seconds for rebooting, or it being
a core part of Red Hat's evil plan to conquer the world (less likely).
Anyway, don't take me wrong. I *do* prefer the model of Runit over the
model of Systemd. But those who don't are quite abundant. And think
about this: there is no "correct lifespan" for a program to last..
People expect things to be done for themselves and demand new ways to
do things that get closer to how they think. You can't know what's
happening everywhere or what will happen, however you do know from
barely every project and from your past experience that updates keep
constantly rolling. Besides, if you could magically do perfect small
applications, that means you'd have to keep working with and on
millions of small tools, right? Wouldn't that just add up complexity!?
tl;dr: Not everyone is DJB, so not everyone thinks like DJB. And some
of them make me write mails to people who think like DJB. Others make
and use Systemd instead. Sad.
>> Has Runit been so well tested and hardened,
>> is it such a simple codebase? Or are there not enough interested,
>> capable people maintaining the project?
>
> The answer to both questions is partly yes.
> There is one lingering bug in runit that I know of, but it only
> triggers in extreme corner cases. Apart from that, it just works,
> does what it'supposed to do, and is a very simple codebase, so
> there's no need to constantly tinker with it.
> It is also true that it would benefit from closer maintenance.
> Gerrit Pape, runit's author, is still around and still reads this
> list, but is not as active as he was a few years ago (typically
> during runit development). Fortunately, there are not many feature
> requests.
OK, I see. So there's a single bug left. It could be fixed. A single
known bug can either mean a really overlooked program or something more
stable than Systemd. And I incline myself towards the latter,
considering it has over 13 years of development on its back, being most
likely today more popular than it's ever been (i.e. it is actually the
default on a distro as well as packaged on many popular ones).
> As others have mentioned, if you're looking for a closely maintained
> supervision suite that is similar to runit, may I recommend
> http://skarnet.org/software/s6 ? It's not going away any time soon,
> and neither is its author - I hope ;) Lots of work have been put into
> s6 in the last few years, more than in any other supervision suite,
> with the possible exception of nosh; the fact that Void Linux has not
> yet switched to s6 is proof that runit is still working well enough
> for them and they don't feel the need to change
Great then! I still don't feel either the need to switch. I'll
definitely try it out on a QEMU VM or a Docker container, but so far
Runit is seriously impressive.
Greetings to the creator of s6! I hope too that he doesn't go anytime
soon!
Francisco / _at_espectalll
Received on Tue May 02 2017 - 17:16:24 UTC