Everyone Has Different Views On The "Open-Source Community"

[brosis]

The people who dislike Lennart typically do not use Gnome. From their perspective, they see their systems suddenly being switched to Systemd, so that is the first thing they blame. People don't like change. Lennart is also a bit of a celebrity, and in Western society a lot of people feel satisfaction seeing a celebrity knocked down a bit. The Lennart haters are the same as the Shuttleworth haters; Systemd is the new Mir. There are also a few trolls around - people who don't care about Systemd, but think it's fun to bait Lennart and watch him suffer. The blog post he wrote was possibly the worst way of dealing with that, since it will only encourage them.

I dislike Gnome3 and I really appreciate Lennart's work. I dislike Gnome3 due to shift of paradigm - lost of customization, turning it from very customizable desktop into very locked-in in regard to changes, a "product". That doesn't mean I want to change Gnome3 or I make any hatespeech or threatening to the project - the principle of "Like - join, dislike - move along" is the best for free software.

[TheBlackCat]

The basic argument that has been given is that by making systemd a dependency of Gnome, distributions that wanted to ship Gnome had to adopt systemd (or develop a plugin compatible alternative that does the same stuff in the same way with respect to Gnome, which is hard). Since Gnome is popular and important, distributions chose the easier route of adopting systemd. Whether or not you agree with that perspective is up to you.

Considering that most major distros switched well before the Gnome dependency thing happened, and the only one that did so afterwards (Debian) cited many other reasons for the decision, I don't think it is really a matter of perspective. It is a matter of denying the facts. (Ubuntu switched too, but it did so because Debian did so)

[TheBlackCat]

Who owns systemd? Who owns Gnome? No, it couldn't possibly be

Who owns KDE?

[brosis]

You need to distinguish between de jure and de facto. And since your "support" is actually life support, Gnome pretty much belongs to Red Hat for all practical purposes.

Gnome's dependency on systemd wasn't something that was warranted by the internal design. It was - along with shitty new look&feel and overall functionality loss - the cost of being taken under Red Hat's wing.

What gives? If Red Hat starts contributing to GNOME it dies? If one dislikes what is done to GNOME by Red Hat is he denied in the right to fork?

[TheBlackCat]

Can somebody please explain me like I'm 5 what's with this systemd thing? From what I understand after the kernel is loaded init is started as the first process, and it goes on to initialize other processes and services aka daemons. Everything correct? The problem with init was that it couldn't start things in parallel so Canonical invented Upstart, which could. From here on what does systemd bring over Upstart? What do we need more than processes that start in parallel? Isn't that good enough?

It is a bit confusing because there are two things that are called "systemd".

The first is a general service manager. The goal is to provide more control and more reliability for starting and stopping services. The parallel initialization of services is a nice side-effect of that, but not the primary goal. Upstart did some of that, but its architecture made it difficult, if not impossible, to extend it to have all the control and reliability projects wanted. Also, note I said "service manager", not "init". An "init" handles system services, but systemd can also handle user services (which is why KDE is planning to use it).

The second thing is a collection of general Linux userspace tools. This is a group of applications that software targeting Linux (such as Gnome and KDE-based desktop environments) can rely on to provide convenient interfaces to lower-level system functions and kernel features. Of course it is possible for developers to write that themselves for their own applications, but systemd provides a convenient, high-level interfaces for these low-level functions. If these tools were available at all, they were available from a bunch of independent projects, some of which weren't well-maintained, and which tended to be used, configured, and started in different ways by different distros and thus were hard for user applications to effectively interact with (when I say "configured in different ways", I mean that different distros provided different ways for users to configure these tools, and trying to use the configuration approach from one distro can't be guaranteed to work on another distro).

[YoungManKlaus]

...

bookmarked because its such a nice explanation

[MoonMoon]

There's no such thing as "Gnome folks" anymore. Gnome belongs to Red Hat now. Just like systemd. So I don't blame Lennart, I blame Red Hat.

So who forced all the other projects that (will) have systemd dependencies, like KDE or Upower?
Does this turn into a Red Hat conspiracy thread now?

[RahulSundaram]

So which one of these two systemds gets to be so hated? The service part or the userspace tools part?

Neither and both. There are different groups of people with different perspectives so its kinda hard to summarize all of it but I will try neverthless.

* sysvinit is the best - group: There are some folks who really don't want anything to change from sysvinit or BSD style init scripts. They are potentially used to it (older sysadmins), don't really use it directly much (desktop or laptop users). This group is fairly small IMO since distributions have even before systemd, started using upstart, openrc and various minor variants of sysvinit. sysvinit also doesn't have a maintained upstream for it and the weaknesses are more broadly understood at this point.

* systemd is horrible - group: This is the set of people who really want anything but systemd. Typically used to praise upstart but that project has stopped and some of them advocate for openrc instead. Generally this appears to stem with prior experiences with other Lennart projects (avahi, pulseaudio etc which are quite innovative in Linux but were disruptive and took time to mature because they were really major changes) and isn't helped by Lennart's personality which is quite often very direct online (and from experience very very different from how he is in person) and they don't see all the coordination work he has done with major distributions to resolve their issues and convince them to adopt systemd because they often happen in places like conferences which they don't attend. They also perhaps don't have the need for the advanced features that systemd offers and distributions and technical experts tend to love. Some of them appear to hold the mistaken belief that the project is all about just one contributor (as opposed to 500+) or one vendor (as opposed about a dozen). Some of them really don't like Red Hat for other reasons ( Perhaps folks who love Gentoo or Debian) and rile up against systemd as a way of expressing their dislike for that vendor.


* systemd as only init - group: This is the set of people who like systemd as the init system and some of them even understand how logind or consoled is really useful but don't really want to consider systemd as this large project and would prefer to see it be just the init system although systemd was always envisioned to be more than that. There is some level of misunderstanding and confusion over systemd the project vs systemd the init system but this group is more linclined towards a philosphical debate rather than a technical debate and hence unlikely to be convinced by the features that systemd provides to them.

* systemd is awesome - group: This group really sees how components are intertwined and integrated well together. Perhaps they have seen all the problems with sysvinit and upstart and take advantage of systemd really well or perhaps they are people who don't have much of a history with older init systems and generally less change averse for whatever reasons.

So there you go. Hope that helps.

[erendorn]

So the systemd general Linux userspace tools collection is more like a form of Win32? I mean not the part of interface, that would be Gtk or Qt from what I understand, but more like a MFC over the basic Win32? I suppose that Linux already had its 'Win32' even before systemd. This is more of a wrapper that eases the workflow for programmers.

So which one of these two systemds gets to be so hated? The service part or the userspace tools part?

I guess it would be more akin to Windows core components and services (only part of it), winlogon, seesion manager, recovery console, DNS, location awareness, plug and play, task scheduler, windows time etc..

I think people hate that most of systemd project currently only works (because lack of other implementations) with other systemd project compoenents, in particular systemd proper (the init daemon).
And because the project provides a lot of valuable interfaces, it means that the whole systemd project and init tends to end up in most of distros.

[blackiwid]

to give a few point why I like it as enduser:

I had seen 0 disadvantages when using it yet.
I even start to like the colored errormessages of journald and the tools of it.
I liked from the beginning taht u have distro-independend init files. So I have for stuff like deluge init-"scripts" what I did nto have for ubuntu, the best I could do was maybe copy one from a forum, but there were no included in the distropackages.
I like that that many distros use it, so most of the time I can just do what I read in archlinuxes great wiki even I use fedora.
Great tools, like systemctl to manage that stuff, gentoos tools was similar nice to use, but debians/ubuntus are not so good or not installed by default. a easy systemctl enable apache as far as I know was not so easy to do in ubuntu and debian.

And another thing, the quality of the init files variated in this other distros, some tools had no restart task or no reload or they did give you no output so you did not know did it happen or did it got ignorered or what.