Hacker News new | past | comments | ask | show | jobs | submit login

I toy with Linux but I mostly use OpenBSD. So I'm thankfully not that affected by systemd.

I can completely understand what the OpenBSD init system does. It's a lot harder to fully understand systemd. Plus, as a benefit of systemd, you get headlines like "Don't panic, but Linux's Systemd can be pwned via an evil DNS query"[1].

Red Hat doesn't care if Poettering is a brilliant genius or just a useful idiot. Instead, Red Hat loves systemd for a very different reason: lockin. Most Linux distributions are now utterly dependent on systemd, and by extension dependent on Red Hat.

systemd gives Red Hat far too much control over Linux. They were already the 800 pound gorilla, now they're almost invincible overlords. But go ahead, keep drinking the Kool-Aid.

[1] https://www.theregister.co.uk/2017/06/29/systemd_pwned_by_dn...




> But go ahead, keep drinking the Kool-Aid

Personal swipes are not ok on HN, so would you please not post like this? Your comment would be much better without that last bit.

https://news.ycombinator.com/newsguidelines.html

https://news.ycombinator.com/newswelcome.html


It's quite clear that you didn't read my original message. I explicitly said that I don't like systemd (for some of the less inflammatory reasons you've mentioned), so I've not "drunk the Kool-Aid".

Red Hat conspiracy theories are quite interesting, but you might want to provide evidence for the claim that a unified init system somehow locks people into Red Hat. You do realise that you can have the exact same .service file work on RHEL just as well as it works on openSUSE or Debian? I will reiterate that I don't like systemd by any stretch of the imagination, but a unified init system makes life so much simpler for any user.

Rather than bashing systemd, the community should be working on alternatives. GNU Shepherd is a viable alternative, maybe we should work on that rather than sitting around complaining about what systemd is doing.


It's quite clear that you didn't read my original message.

I did read your original message. Whether or not you personally like systemd or whether or not SUSE likes systemd, it nevertheless is in openSUSE. I don't know enough about SUSE to know if you have any other distributions that don't have systemd.

Why is systemd in openSUSE? How was that decision made? If I were in the leadership of SUSE, I would hate being so dependent on key software that is essentially controlled by my largest competitor.

At this point systemd has become the entrenched incumbent. So "alternatives" are mostly wishful thinking. For people to switch away from systemd, they would need to be convinced that something like GNU Shepherd wasn't just equal, but was significantly better. That seems unlikely to happen anytime soon.

People aren't directly locked into Red Hat, but they sure are locked into a very key piece of software controlled by Red Hat.

This state of affairs has to hurt SUSE. When selecting a distribution, why wouldn't businesses buy from companies as far "upstream" as possible? Why buy software from SUSE if key pieces come from Red Hat? Why not just buy from Red Hat directly?


> For people to switch away from systemd, they would need to be convinced that something like GNU Shepherd wasn't just equal, but was significantly better. That seems unlikely to happen anytime soon.

You seem to be arguing two things: 1. systemd should not be used and 2. it is unlikely that you could convince current users to switch to an alternative.

These seem to me to be at odds with each other. If systemd is so bad, why do people keep using it? I'd guess that the alternatives are even worse for current systemd users. But that doesn't mean that the alternatives couldn't be improved to the point where they can replace systemd for most users. Unless of course all init systems are doomed to suck and you can only choose which one to complain about ;)


> Why is systemd in openSUSE? How was that decision made? If I were in the leadership of SUSE

That decision was made by the openSUSE community. openSUSE is not owned by SUSE in any sense, the community is run by the users and developers of the distribution. There is a board that is elected by the community (and no single company can have >50% of the board seats), but it's role is more dealing with conflicts than anything else.

openSUSE chose to use systemd because some people stepped up and did the necessary work to support systemd. And yes, people still complain about it, but the key point is that nobody has put work into replacing it. There is no reason that openSUSE couldn't support running everything without systemd -- nobody would stop you from doing that work -- but in our community the people who make such decisions are the people who do the work.

> I would hate being so dependent on key software that is essentially controlled by my largest competitor.

Ha-ha, it appears as though you don't understand how free software development works in this context. While Red Hat is a competitor to us, we work with them on their upstream projects just as they work with us on our upstream projects. I spend a large part of my day collaborating with my counterparts at Red Hat. Hell, I'm a co-maintainer with several folks from Red Hat and I contribute to their projects in my free time.

If a customer doesn't like us, they can go to Red Hat. If they don't like Red Hat, they can come to us. If they don't like either they can go to Canonical or wherever else. Hell, we even provide support for migrating to SUSE from Red Hat (and I believe they have the inverse). The benefit of building everything on free software is that you don't have vendor lockin, and systems like this really do "just work".

> For people to switch away from systemd, they would need to be convinced that something like GNU Shepherd wasn't just equal, but was significantly better. That seems unlikely to happen anytime soon.

So you agree that systemd solves problems that are not solved by other systems? Then I don't understand what you're arguing for -- should we intentionally ship software that doesn't solve user problems? Or wait for the community to decide on the best way to move forward before we ship a release (hint: those arguments will never end)?

If you want to get people to switch you need to have an alternative, it's a simple as that.

> This state of affairs has to hurt SUSE.

That's kind of like saying that because 'shadow' is developed by Debian it must hurt Red Hat. Or because Apache is developed by the Apache Foundation that must hurt Canonical. It's a nonsensical argument, that's now how free software works.

> When selecting a distribution, why wouldn't businesses buy from companies as far "upstream" as possible? Why buy software from SUSE if key pieces come from Red Hat? Why not just buy from Red Hat directly?

First of all, Red Hat uses many pieces of our software as well, this is a symbiotic relationship. Red Hat is not the only player (in fact we were around before them). Their new dnf package manager uses our libsolv RPM solver implementation. They are using openQA to perform testing of Fedora. kGraft and kSplice were merged upstream thanks to being able to compare the two approaches and come to a solid decision. There are many such examples.

But to answer your question, it's because we sell different systems with different opinions on how to do things. I'm not going to give you the marketing pitch (I'm an engineer), but we have plenty of really interesting technology that we ship in our products that Red Hat chose not to use (and vice-versa). SUSE and Red Hat both sell operating systems, but they are very clearly distinct and potential customers are given a choice with who they want to do business with.


Ha-ha, it appears as though you don't understand how free software development works in this context. While Red Hat is a competitor to us, we work with them on their upstream projects just as they work with us on our upstream projects.

Thanks for providing such a detailed write up. I hope that knowing a little about how "frenemies" work together is also of interest to others on HN.

Maybe Red Hat and SUSE will coexist happily well into the future. But you're both public companies, and you each owe certain duties to your shareholders.

In the software world the archetypical example of companies collaborating is Microsoft and (... any of dozens of companies go here ...). It seems that never ended well for anyone but Microsoft.

But perhaps the nature of open source / free software fundamentally changes this dynamic of collaborating with the dominant player in an industry.


> But go ahead, keep drinking the Kool-Aid.

This was unnecessary.


cyphar essentially said the same thing. So, point taken, this comment was way over the top. Sorry about that.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: