[conspire] "The Tragedy of systemd" ... & livingcomputers.org Living Computers Museum+Labs

[Rick Moen]

I wrote:

> Well, not a dissection, but a few comments:

_And_ the inevitable follow-up, after some guy had decided to argue for
reasons unclear (possibly a spinal reflex):


Date: Thu, 14 Feb 2019 20:38:47 -0800
To: tech at golug.org
From: Rick Moen via Tech <tech at golug.org>
Subject: Re: [GoLugTech] Look what I found - "The Tragedy of systemd!"

Quoting David Krauser via Tech (tech at golug.org):

> I didn't read his talk as propaganda.

Well, to reiterate what I already made a point of stressing a couple of
times, neither did I.  (To repeat:)  What I said was that if one _did_
set out to create a propaganda fluff piece, it would share many notable
features.

> His omissions could simply be explained by the fact that this was a
> short talk....

Um, 47 minutes?  ;->

That's a standard talk length for linux.conf.au, if memory serves --
basically, a hour slot with some time for questions and a few minutes
between panels.  IMO, the nature and extent of omissions is pretty
notable, though I certainly was _not, repeat not_ calling them
nefarious, just notable.

He had absolutely no obligation to give any broader talk, of course.
For example, he had no obligation to mention that solutions to service
management have been commonplace in Linux and the other *ixes for
at least a decade or two.  At my recent employer, a large Internet
content firm, we'd been using supervisord for that purpose for long ages,
and every sysadmin who's been in that line of business is familiar with
either that or monit or daemontools or any of several others.

So, when we sysadmins see people like Mr. Rice give talks saying 'we
need to solve the service management problem', we have to wonder where
he's been hiding while we _did that_.  For decades.

Or, to put it another way, 'Hey, Mr. Rice, what exactly are supervisord
and monit and daemontools, then?  Chopped liver?'  (I'm omitting
relatively recent additions like runit and s6 and procer and freedt and
daemontools-encore from that list, but they're not chopped liver,
either.)


> The merits of systemd itself seem somewhat irrelevant to his points.

Um, actually, the use-cases where systemd shows strengths occupied about
the later half of the talk.

> What I got out of his talk:

Yes, well said.  But:

> I'd agree with those ideas.

Those points seem fine if you are unaware of what is already in common
use in exactly the areas where Mr. Rice claims we 'need to solve' this
and that.  But frankly, I expected better of an old-school BSD guy.
(I'm actually an old-school BSD guy, myself, having run 386BSD before
moving to Linux.  Not that I'm any expert, but I've been around.)