Talk:Systemd/Archive 1
![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 |
First distribution to include systemd
In released anouncement of OpenSuse 11.4 milestone 2, it is written "D-Bus 1.4 adds support for systemd and fixes the threading issues which gave KDE 4.5 problems.", so it means systemd will be included in OpenSuse 11.4. So, maybe it will not be Fedora 14 which will include systemd. Pamputt (talk) 18:19, 7 October 2010 (UTC)
- Nothing certain, they're only considering it, see this blog post: http://lizards.opensuse.org/2010/10/08/systemd-and-osc2010/ -- intgr [talk] 19:58, 9 October 2010 (UTC)
Arch package
Now Arch packs systemd in its 'Extra' repository, along with a compatibility 'systemd-sysvcompat' package for use with the distro's official old fashioned init scripts. The native systemd unit files ('systemd-arch-units' package) are still in the community repo. I did't edit the front page because my english sucks and I don't want mess up with your language folks. Someone better english skilled please do it. — Preceding unsigned comment added by Marcosf0 (talk • contribs) 00:01, 21 February 2012 (UTC)
Sorry, the 'systemd-sysvcompat' package is not for what I thought. Installing it will remove sysvinit and initscripts. You will need systemd-arch-units (and have I pure systemd system). — Preceding unsigned comment added by Marcosf0 (talk • contribs) 23:31, 10 March 2012 (UTC)
wordplay "système d"
if you follow h-online it turns out the meaning of "the developers thought" does not mean the systemd developers thought so, but other developers. this can be seen at lennarts blog entry which is referred. there systeme d shows up the first time in the comment section, and even the wikipedia article is cited at nearly the same time. this same wikipedia article links then to h-online, which links to the blog entry citing the wikipedia article. it looks like a circular reference, the only thing which is pretty sure: the (german) system d developers did not have (the french) systeme d in mind. but the wikipedia introductory paragraph suggests exactly this - at least for a medium level english speaker (like me ...). that one can make a word play out of it is undisputed ... but there are many words where one might do this. --ThurnerRupert (talk) 18:29, 13 August 2012 (UTC)
- Good point. I interpreted the statement in h-online as "The developers [chose this name because they] thought that its name is suitably reminiscent of the French term "système D""
- But now that I also read the blog comment that you probably pointed to [1], I have to agree with you -- it does sound like Lennart was not aware of this term while naming the software... Quoting:
Eric: Well, I am sure that everything has some meaning in some language of this world. Also, reading the Wikipedia article I got the idea that the term wasn't negative at all?
— Lennart- -- intgr [talk] 08:10, 14 August 2012 (UTC)
Arch version?
Arch is version-less. It's weird to say "Arch-10.06.2012 and later", especially considering the install medium for Arch has been network based (no packages on the installer disk) since 7.15.2012. That means that installing from the 7.15.2012 disk will get you the exact same system as the 10.06.2012 disk, as would any net installer prior to 7.15.2012. 70.97.12.233 (talk) 21:29, 18 October 2012 (UTC)
- I couldn't agree with you more. The installer uses pacman to install the "base" group, which has included systemd since October 13th. →
UncleNinja
(talk)
19:03, 29 October 2012 (UTC)
Less code than init?
You have to be joking. Systemd has reams and reams of incomprehensible code with no comments about what any functions do.
The tiny source code of sysvinit is a transparent piece of glass by comparison.
24.85.131.247 (talk) 05:46, 22 May 2013 (UTC)
svg systemd utilities
Should it really say analize and not analyse? — Preceding unsigned comment added by 77.110.10.45 (talk) 21:12, 19 October 2013 (UTC)
- It should be analyze, as the utility name is systemd-analyze. I fixed the image, good catch. --Götz (talk) 05:07, 21 October 2013 (UTC)
Criticism
Like most software by Lennart Poettering systemd received a lot of criticism. Probably enough to be notable and therefor should be mentioned. No, I am not talking about bashing. For example there is criticism from the BSD folks, because his software focuses on Linux and from Linux guess because his development focuses on how Fedora does things. Another criticism is that his softer doesn't comply with the Unix philosophy, both the technical side as well as the idiomatic. A lot of these criticism can be referenced by a talk from 27C3.
http://www.youtube.com/watch?v=ZTdUmlGxVo0
Since the objections are always very similar I think it would fit into Lennart Poettering's article and systemd, pulseaudio, etc. should link there. Another possibility is just to mention that the approach of designing and implementing his software is very different from how the Unix world has done these things before (and therefor maybe aren't that portable, modular and replaceable). This would probably make it harder to use it for bashing. It would be informing instead of criticizing. --Athaba (talk) 08:27, 15 August 2011 (UTC)
- Sounds like you have an agenda going on there. Keep it off Wikipedia please. Almost every mainstream distro has adopted or plans to adopt all of the projects you mentioned. Obviously the people who actually have influence (i.e. the people who actually know what they're talking about and contribute instead of complaining) have voted in favour of these projects. There are some known issues and some pending improvements - much like any other software. That's not really a notable or objective "criticism" unless you want to be more specific. — Preceding unsigned comment added by 109.145.165.125 (talk) 09:23, 18 August 2012 (UTC)
- Please don't make insinuations. It's impolite, just like telling people to get off Wikipedia is. I didn't edit the article, just like I didn't criticize the projects on my on my own. I don't have an agenda on software projects. If people use software then there usually is a good reason to. I doubt that all the people with criticism are wrong and I also doubt there is a project, just like a person that's perfect. Also I have enough knowledge in this area (Unix) to know that at least some of the points mentioned are valid, even though it is a pro and contra decisions. That's the reason for taking a video were Lennart Poettering gave answer to all of his design decisions. Every bigger project and every big change brings criticism, how noteworthy it is and whether it makes sense to include it into Wikipedia is another topic of course and the reason I brought and put it on the discussion page rather than writing something on my own. Again, please don't make such assumptions on other people when you have evidence that they are not true, which I think you should have.--Athaba (talk) 10:56, 1 September 2012 (UTC)
Agree with the need for a criticisms section. This software is responsible for a much, much larger amount of controversy than the average piece of software. You could find this controversy with even a cursory glance at systemd related posts online that provide comment sections. Poettering's own Google+ page contains numerous comments criticizing systemd and his software in general, oftentimes accompanied with technical explanations based in fact. Also, the assertion that "Almost every mainstream distro has adopted or plans to adopt all of the projects you mentioned" is patently false. Ubuntu (which is the LARGEST distro) and Slackware are notable exceptions. Additionally, stating that the criticism of systemd is like other software does not make it so. As mentioned above, the controversy is very, very visible. 24.196.27.208 (talk) 04:51, 29 October 2012 (UTC)
- +1. I'm a happy user of systemd, but there's no question that it attracts lots of controversy, which should be covered on Wikipedia. Two points (perhaps obvious, but bear repeating anyway):
- 1. Don't call the section "criticism" as if you could only say negative things in that section. "Controversy" is better, as it can cover both sides of the argument.
- 2. Per WP:RS, only cover points made by reliable sources and always cite the source.
- -- intgr [talk] 20:19, 29 October 2012 (UTC)
- The "Criticism" title is used on plenty of other articles, but using "Controversy" here would indeed be a better idea in order to promote a more balanced article, as the rest of the article doesn't address its strengths too much. 24.196.27.208 (talk) 17:59, 30 October 2012 (UTC)
Having a separate "controversy" section is almost never useful. It would be better to just cover the history of the project, which would inevitable include controversies.Korin43 (talk) 19:19, 30 December 2013 (UTC)
- Shouldn't the start of this say "Like most core components of the Linux operating system systemd received a lot of criticism..."
Every change to the base packages, esp. those which impact the majority of users and how they have their desktop or small servers configured, will result in a good deal of push-back. Look at what happened with openSUSE dropped Sax2. I think a lot of the blame falls on those who made the decision to go ahead with systemd but failed to provide sufficient guidance to their non-Linux professional user base. As Linux becomes increasingly used on desktops (openSUSE has been the OS on my family's PCs for the last 8 years--I am not a Linux system admin or developer--I am a physician and am just someone who was sick and tired of Windows®), major distributions (e.g. openSUSE, Ubuntu, Fedora) aimed at this market have to step up and explain things with a bit more detail and clarity. I am still trying to figure out how to use Yast to configure it so that I have a common set of services whether I go into console (with no X) or GUI (KDE) mode.
It would be very helpful to document some of the features. While not a full "how to" (that is for WikiBooks!), discussing the various use cases and approaches would provide much needed information. — Preceding unsigned comment added by 75.73.1.89 (talk) 10:32, 11 February 2014 (UTC)
Encouraged editing by a website
There is a website debianfork.org which expresses hostile sentiments towards systemd. Under the "How can I help?" section it suggests: Also it can be helpful to monitor and update the Wikipedia page about SystemD. --77.12.0.101 (talk) 10:26, 19 October 2014 (UTC)
Looks like a systemd dev trying to stir up more controversy... Since when monitoring a page is vandalism? Please, give it a break. You're alienating many power users with that attitude of "who isn't with me is against me". — Preceding unsigned comment added by 136.173.162.129 (talk)
- It's good to have heads-up when some opinionated website encourages people to get involved in a Wikipedia article. No need to call it "vandalism" though, or insinuate that the OP has an agenda of "stirring up controversy". -- intgr [talk] 14:16, 20 October 2014 (UTC)
- Well, the "Reception" includes a lot of anecdotical critisism, even mentionning anonymous websites, while the massive enthusiastic adoption by distributions, including their sometimes highly-detailed feedback, are only linked as references. Whatever our opinion on the software might be, there is some massive NPOV going here, and all that popped-up recentely. I noticed, however, that over-focusing on criticism was a general trend in english-language wikipedia articles about softwares, so i'm not sure the readership would care that much, especially for such a niche software. — Preceding unsigned comment added by 130.34.163.29 (talk) 00:02, 30 October 2014 (UTC)
Unintended consequences
Because systemd is available on -- and the default on -- so many distributions, even in light of its (WARNING: AGENDA AHEAD) opacity and near total lack of documentation, it is becoming near impossible to *run* sysVinit, even if you are willing to continue packaging and building it yourself, because the package builders for various distros are not only ceasing to maintain the requisite initscripts for all the *user level* packages in their repos, *they are ceasing to include them in the packages*. So you're pretty much stuck with your system's most critical code being something poorly documented and understood. Surely in that rant there's at least one or two facts that merit inclusion in the main article? :-)
--Baylink (talk) 18:46, 30 November 2013 (UTC)
- The "merit of inclusion" is decided by the verifiability policy. If you have good, reliable sources saying that it's poorly documented then sure.
- I happen to disagree, sysvinit's documentation is much worse and it shows in how badly most initscripts are written. But my opinion has little relevance to Wikipedia. -- intgr [talk] 00:43, 1 December 2013 (UTC)
- in other words, for 99.9% of the computer users in the world, nothing important has occurred ? cause 99.9% of users don't even know what compile or runtime shell scripts are ???
From gnomeasia2014 (PDF of slideshow from discussion): "Turning Linux from a bag of bits into a competivive General Purpose Operating System." This is essentially turning Linux from a modular 'bag of bits' (Unix-like) to Windows where even the browser and GUI are dependent on the other parts of the OS. No longer packages but a 'big blob'? Or perhaps like Android. This seems like a classical power grab or disruptive action. 73.3.211.0 (talk) 09:29, 30 November 2014 (UTC)
Can anyone define "Major Linux Distribution"?
OpenELEC is listed as a "Majour Linux Distribution." I'm failing to grasp how it's considered to be a majour one. gh5046 (talk) 16:14, 21 October 2014 (UTC)
- Amongst media-center focused distributions on the raspberry pi, OpenELEC is probably the most popular one, and the rpi has shipped 3.8 million units as of october 2014 according to its wikipedia page, so I guess that counts for something. 80.212.191.68 (talk) 03:27, 28 November 2014 (UTC)
- Never heard of it. Gronky (talk) 19:54, 1 December 2014 (UTC)
Layout and readability
Putting frontends and backends in unrelated sections doesn't make sense. They're all software components.
Also, the "Components and libraries" section is not readable. It's fourteen sections, half of which are empty, a further quarter of which give little more than the date of integration into systemd. And the section contains a lot of external links - this is an encyclopedia, not a web portal. This section needs serious compacting. Gronky (talk) 11:51, 5 December 2014 (UTC)
- Yeah, but graphical front-ends are not official components of systemd. They're more like independent projects, that's why I've moved them later into the article; in other words, they pretty much represent "related work", going well together with forks. Regarding the systemd § Components and libraries section, I do agree that it should be compacted. Will do something. — Dsimic (talk | contribs) 12:09, 5 December 2014 (UTC)
- Should be better now, please have a look. — Dsimic (talk | contribs) 12:25, 5 December 2014 (UTC)
- The article calls systemd-ui "official". Is it an official part of systemd or not? Gronky (talk) 12:30, 5 December 2014 (UTC)
- That's a who-knowns-how-old leftover, systemd-ui shouldn't be called "official" as it's pretty much unmaintaned. Please have a look at one of the latest commits to systemd-ui's source, here's an excerpt:
- This tool has been largely obsoleted by the commandline tools like systemctl and systemd-analyze. Bugs are likely to be present.
- That's a six-months-old commit, and there have been no newer ones. — Dsimic (talk | contribs) 12:56, 5 December 2014 (UTC)
- That's a who-knowns-how-old leftover, systemd-ui shouldn't be called "official" as it's pretty much unmaintaned. Please have a look at one of the latest commits to systemd-ui's source, here's an excerpt:
- For what it's worth, a certain GUI-heavy UNIX with a process 1 that has some similarities to systemd :-) has no graphical front end to process 1, so I'm not entirely surprised by "this tool has been largely obsoleted by the commandline tools like systemctl and systemd-analyze", although a certain GUI-heavy very-much-non-UNIX does have a GUI tool to talk to its closest equivalent. Guy Harris (talk) 19:46, 5 December 2014 (UTC)
- Hehe, well, probably there hasn't been much interest in having such a low-level utility on that certain GUI-heavy UNIX. :) At the same time, that certain GUI-heavy very-much-non-UNIX operating system recently adventured into the world of unfriendly typed-in commands – go figure. :) — Dsimic (talk | contribs) 19:55, 5 December 2014 (UTC)
- Yes, to, for example, turn sshd or Samba^WSMBX or ftpd or... on, you'd go to the Sharing pane of System Preferences and enable the particular service in question. I suspect some Linux distributions could do the same ("distributions" rather than "desktops" because the desktop might not be able to count on the system using systemd and maybe might not be able to count on particular services having particular names). Guy Harris (talk) 20:09, 5 December 2014 (UTC)
- That shows how well OS X is integrated; another example might be the proxy configuration in OS X, if I recall correctly – it surely changes port redirection configuration on a system-wide firewall level. On the other hand, the power of Linux has always been in its diversity and pretty much internal disparity, so it could be questionable which direction would be better for Linux. It might be the best to offer both approaches for Linux, so desktop users can have easily accessible functionality, while all the gears continue to be clearly exposed to those who want to be in more direct control. — Dsimic (talk | contribs) 20:21, 5 December 2014 (UTC)
Is sysvinit just /etc/init or does it also include some scripts?
sysvinit currently links to init#SysV-style, which describes a style of init implemented by at least two independent families of implementations - the /etc/init from (various releases of) System V and various commercial-UNIX derivatives of them, and the /etc/init from the Linux distributions that implement System V-style init.
Are there multiple different "System V-style /etc/init program" projects for Linux, or is there just one? Do those projects have anything more than the source to the /etc/init program and possibly a template /etc/inittab, such as template scripts?
The article says "systemd replaces sysvinit and the traditional init daemon"; if sysvinit is just /etc/init, then it is the "traditional init daemon", and that should probably be "systemd replaces sysvinit and the scripts it runs"; if sysvinit is a project supplying scripts as well as /etc/init, then sysvinit should perhaps be a page of its own, discussing the entire project, and this article should probably say "systemd replaces sysvinit" and possibly note that sysvinit is both "the traditional init daemon" and the scripts it runs. Guy Harris (talk) 01:35, 11 December 2014 (UTC)
- That's a very good remark, thank you for pointing it out! In a few words, /sbin/init is a separate project and pretty much the same in all "traditional" Linux distributions, and it has no idea whether it runs SysV-style or BSD-style startup scripts – it does whatever /etc/inittab specifies. The actual startup shell scripts are developed and supplied by a particular distribution. Furthermore, the init daemon and actual startup scripts are usually packaged separately; for example, CentOS 5 separates them into SysVinit and initscripts packages. Got it improved in the article, it was really confusing – please, check it out. — Dsimic (talk | contribs) 06:31, 11 December 2014 (UTC)
- "it does whatever /etc/inittab specifies" ...which makes it more System V-style than BSD-style; BSD's init is a descendant of the original V7 init, which had no notion of runlevels. I think System III was the first UNIX with run levels (it definitely had them, as per the fine manual, although, as I remember, its scripts were less complicated than the ones in later System V's (SVR1's scripts, according to its fine manual, don't appear to have been that complicated; what people think of as "SysV-style init" may have started with SVR3). Guy Harris (talk) 09:44, 11 December 2014 (UTC)
- With Linux, "BSD style" refers to the lack of per-service scripts in /etc/rd.d/init.d and associated start/stop symlinks in /etc/rc.d/rcN.d. Instead, "BSD style" comes with only a few startup scripts placed in /etc/rd.d, and services are enabled or disabled by editing them manually. Of course, that isn't a true BSD style, at least because runlevels are still there. — Dsimic (talk | contribs) 10:07, 11 December 2014 (UTC)
If this article is right and everyone hates it, why are the distros adopting it?
I've no opinion on systemd. I came here to do some first reading about it, but this article seems to contradict reality.
Reality is that pretty much all the major distros are adopting systemd, which suggests that pretty much all the senior makers of technical decisions think that systemd is great.
So I come to Wikipedia to read about why they think systemd's great, but I just find a list of people who have criticised it. That means something is missing. Either:
- People generally hate systemd, but it's getting adopted anyway. If this is the case then the article content is fine but something needs to be added to explain this counter-intuitive situation. Or,
- Lots of people actually love systemd. If this is the case then the article should include those views.
I'd love to roll up my sleeves but I know very little about systemd, so I'm not the person to review this article.
One possible source for good things about systemd is the comments to this article on slashdot recently: Ask Slashdot: Can You Say Something Nice About Systemd?
Can anyone make a start on fixing this article in whichever way that needs doing? Gronky (talk) 19:52, 1 December 2014 (UTC)
- Hello! Well, the article shouldn't be "fixed" in any way, as Wikipedia's mission is to present both pros and cons in an unbiased way. Speaking of systemd, it's no wonder that Linux distributions like it, simply because it takes a lot of work off their hands. At the same time, it is true that systemd turns Linux into a "blob-like" structure; a good example, among many others, is system'd support for NTP, which is almost ridiculously pointless. — Dsimic (talk | contribs) 13:41, 4 December 2014 (UTC)
- Great! If you know something about systemd, maybe you can make the necessary changes ("fix it" in my terminology, "be unbiased" in yours). The "Reception" section can be summarised as:
- Linus critised it
- Ted Ts'o critised it
- Mark Shutleworth critisied it
- Patrick Volkerding critised it
- Paul Venezia critised it
- Some folk boycotted it
- Then the next section (Adoption) says that every single major distro has adopted it. Can you either explain why distros are jumping on a piece of software that (accordin to the article) everyone hates, or can you add something to indicate why some people do like systemd (how does it "take work off their hands"?). I'm not pro or anti, I'm just a casual reader who'd like the article to make sense. Gronky (talk) 16:14, 4 December 2014 (UTC)
- Great! If you know something about systemd, maybe you can make the necessary changes ("fix it" in my terminology, "be unbiased" in yours). The "Reception" section can be summarised as:
- Well, I know a thing or two, ;) and I'll see to make the article tied together better. Regarding how systemd helps Linux distributions do less work, they for example no longer need to maintain their own sets of startup scripts and associated shell libraries, what's a quite time-consuming thing to do. — Dsimic (talk | contribs) 16:22, 4 December 2014 (UTC)
- These few edits should help, please have a look. — Dsimic (talk | contribs) 17:21, 4 December 2014 (UTC)
- What the article needs is a section about what is different in a GNU/Linux system when you replace the current init system with systemd. I've rearranged the sections and made a start on this but knowing so little, it takes me a lot of research work to write just a sentence or two. If you could also add a sentence or two of objective description of systemd, then I think the article would be on its way to being useful. (We should avoid "it's simpler" - the critics say the opposite is true; but we can provide numbers for boot time, and we can say it manages ten services and provides d-bus interfaces (why is that useful?) and other objective things - and leave he subjective (verbose) stuff for other sections.)Gronky (talk) 20:22, 4 December 2014 (UTC)
- These few edits should help, please have a look. — Dsimic (talk | contribs) 17:21, 4 December 2014 (UTC)
- Should that be in a named section, or should it be in the introduction? Perhaps the introduction should, at least, mention the launch-on-demand, with declarative specifications, architecture of systemd-the-daemon, although the launchd article first mentions its launchd-on-demand, with declarative specifications, architecture in the "Components" section, not in the introduction. Launchd, however, is just the daemon (and its control command) - Apple haven't bundled any other system daemons into the launchd project - so perhaps systemd's page needs to be structured differently. Guy Harris (talk) 01:23, 5 December 2014 (UTC)
- I'm guessing the features of systemd are enough to deserve a named section. I don't know much about systemd, but with all the controversy I'm guessing it's features (good or bad) are substantial. (If some highlights could be mentioned in the intro, I guess that would be good.) If you can make content additions but are holding back because of questions about layout or style, you could make your additions and leave it to someone else to massage the text into a structure (I might do it). Gronky (talk) 11:56, 5 December 2014 (UTC)
- Until the article grows substantially, I'd say that expanding the systemd § Overview section further should be just fine, while adding brief summaries to the lead section. — Dsimic (talk | contribs) 12:12, 5 December 2014 (UTC)
- Hello! It is worth noting that systemd gained little traction with non-RedHat distributions until udev adopted a dependency upon systemd. Several udev-using Linux distributions adopted systemd soon following the introduction of this dependency. It could be that they only adopted systemd to continue using the latest udev releases, but that would be difficult to prove. TTK (talk) 04:01, 23 December 2014 (UTC)
- Welcome here! :) Hm, finding a reliable source for that would be quite tough, though it would be a good addition to the article. — Dsimic (talk | contribs) 04:05, 23 December 2014 (UTC)
Developer is not Red Hat Inc
From FAQ on http://0pointer.de/blog/projects/systemd.html :
Is this a Red Hat project?
- No, this is my personal side project. Also, let me emphasize this: the opinions reflected here are my own. They are not the views of my employer, or Ronald McDonald, or anyone else.
—Preceding unsigned comment added by 209.132.186.34 (talk) 09:18, 26 November 2010 (UTC)
- All developers are paid for exclusive systemd development by the one company Red Hat Inc. All name their employer in their email-adress, social network profiles etc. Suppressing this information from the article might give the impression of a slightly biased POV. -- 92.226.33.150 (talk) 01:46, 6 December 2014 (UTC)
- That same page indicates a diversity of employers, actually. Quoting:
- Well, the current code-base is mostly my work, Lennart Poettering (Red Hat). However the design in all its details is result of close cooperation between Kay Sievers (Novell) and me. Other people involved are Harald Hoyer (Red Hat), Dhaval Giani (Formerly IBM), and a few others from various companies such as Intel, SUSE and Nokia.'
- If there are other (perhaps more recent) sources which claim something else, that would be interesting. -- Beland (talk) 17:29, 30 December 2014 (UTC)
- That same page indicates a diversity of employers, actually. Quoting:
Chart
File:Systemd components.svg by User:ScotXW is somewhat informative (thanks for pulling it together) but leaves me a bit confused. It's unclear what "targets" and "core" are. For example, are they programs that are running, or are they like libraries? The biggest question is, what is the overall relationship between utilities, daemons, targets, core, libraries, and the kernel? Some lines showing call paths or other relationships would be helpful. -- Beland (talk) 17:35, 30 December 2014 (UTC)
- Targets aren't really "components" so much as "states". It's important to note that this is really just a set of rings around various terms within systemd: it's not an architecture diagram. Chris Cunningham (user:thumperward) (talk) 15:55, 31 December 2014 (UTC)
Overview still in need
I've removed the third paragraph and its associated 3-point list from the "Overview" section. Here's what that section looked like before my edit: [2]
I read it a few times and then realised that I learned nothing about systemd. It gave me nothing at all of an overview. It could be summarised as: "Poettering says systemd is actively developed and is a service manager, a platform, and glue". A bit like having a paragraph to say systemd is software, made of ones and zeros to be run by a processor. If we add filler, just for the sake of filling sections, we'll end up with an unreadable and unmaintainable article.
But I do hope people can add info to that section to say concretely what systemd does for a GNU/Linux system. Gronky (talk) 16:54, 9 December 2014 (UTC)
- No, it can't be summarized this way at all. He is not describing it this way. He is describing it as a "system and service manager." That means it manages both the system and its services. There is a big difference between that and just a service manager. He is also describing it as a platform, which means that it is a basis for developing other software. He is also describing it as glue -- not in some vague generic sense -- but glue between the kernel and applications. He is summarizing systemd as having three main functions, which are clearly defined as three numbered points. It's very clear, and straight from the horse's mouth. Please don't dumb it down and remove the important details. I'm not sure why you feel it is necessary to remove important sourced information like this, or modify it in a way that is less accurate... :-( Huihermit (talk) 23:36, 9 December 2014 (UTC)
- Is that "systemd the program that runs as PID 1" or "systemd the suite of programs, including but not limited to systemd-the-program-that-runs-as-PID-1"? That's a significant distinction - I get the impression that some of the ZOMG SYSTEMD DOES TOO MUCH!!!!!111ONE!!!!! complaints are based on the belief of the complainers that some of the functions performed by other programs in the suite are actually bundled into systemd-the-daemon, hence "PID 1 is a critical process, why should it be so big and bloated doing XXX" complaints. Guy Harris (talk) 23:48, 9 December 2014 (UTC)
- Yeah, there's little to no need to delete well-sourced and precisely stated details. The Systemd § Overview section describes pretty well what systemd is about, and it should be clear to anyone already familiar with the traditional init systems; otherwise, why anyone should bother to understand it without already knowing some "history"? Also, the description continues into more details in Systemd § Components and libraries section, and those two sections complement each other rather well. Beside reverting the deletion, I've tried to clean up and clarify the overview a bit further. Hope it makes the thing even more understandable. — Dsimic (talk | contribs) 01:20, 10 December 2014 (UTC)
- Huihermit, that text meant something to you because you already know what systemd is. To me it meant nothing. Can you add something to the article about what difference is made by systemd "manag[ing] both the system and its services"? That could be informative.
- The meaning you draw from "platform" says nothing. Every system component gets built on. And most system components do the "glue" role of interfacing applications with the kernel.
- That paragraph, in its current form, provides no information. It's a waste of space. Cruft.
- You think it means something because in your head you are expanding the words and attaching them to facts you already know. The text doesn't have a meaning to you, it's just reminding you of stuff you know. Can you use the stuff you know to produce a text which would inform someone who doesn't know what systemd is?
- ("well-sourced info"? - I'm not so sure. Poettering is far from an independent commentator. IIRC WP prefers third-party sources, so this is only sourced not well-sourced.) Gronky (talk) 13:06, 10 December 2014 (UTC)
- Gronky, may I ask how familiar are you with traditional init systems and things like daemontools? — Dsimic (talk | contribs) 19:46, 10 December 2014 (UTC)
- You're asking the wrong question. FWIW, daemontools isn't installed on my system and I think I've never used it (unless I know its components by other names), and I have lower-intermediate level knowledge of sysv.
- But my knowledge is irrelevant. This article should be understandable (and informative) to someone who's never touched a computer with sysv init. Asking about daemontools makes me wonder if you're way off course here.
- Can we get back to discussing making the Overview section understandable and informative? Gronky (talk) 22:01, 10 December 2014 (UTC)
- Gronky, may I ask how familiar are you with traditional init systems and things like daemontools? — Dsimic (talk | contribs) 19:46, 10 December 2014 (UTC)
- Everybody's knowledge is relevant; we simply can't expect a Joe "Wintel" Average to jump into this article and be immediately educated on what Linux is, what those init systems do and why do they exist at all, why do we need some systemd as yet another one, etc. Some prior knowledge is required for pretty much all technical articles in Wikipedia. Moving forward, could you please explain what's actually unclear to you? Starting from the article's lead section (or from Systemd § Overview if you prefer), which is the first sentence you don't understand? — Dsimic (talk | contribs) 22:45, 10 December 2014 (UTC)
- The problems with this article are glaringly obvious, but when I point them out I get various replies about "Gronky can't understand it" or "Gronky is handicapped" (see below). As I keep saying, this article is uninformative. Even if we leave Windows or Mac users aside, let's say a normal GNU/Linux user reads the article, will they know what systemd is at the end? Will they even read to the end?
- A line-by-line review is suitable when there are fine-grained problems that one needs to go through with a comb in order to find. This article's problems are still way too big for that.
- If you're all so knowledgeable, can any one of you write an overview of what systemd is? Gronky (talk) 23:29, 10 December 2014 (UTC)
- Well, it is already written, here's an excerpt from the Systemd § Overview section:
- Before systemd, the majority of Linux distributions had their startup sequences controlled by sysvinit. As an integrated software suite, systemd replaces sysvinit and the traditional init daemon, and also integrates many other services that are common on Linux systems, such as login, device management (replacing udev), scheduled execution (replacing cron), and a unified system for logging these services. Services are all started in parallel, which is faster than the traditional sysvinit sequential startup sequence.
- Systemd is not just the name of the init daemon but can also refer to the entire software bundle around systemd. This includes the daemons systemd, journald, logind and networkd, and many other low-level components such as libraries and utilities. In January 2013, systemd author Lennart Poettering described systemd not as one program, but rather a large software suite that includes 69 individual binaries.
- Sorry, but I really don't understand what isn't clear there? Systemd replaces the traditional init and sysvinit startup scripts, and integrates a lot of other stuff. It's pretty much clearly expressed already. At the same time, I do agree that the overall layout of the article became much better in the last few days – it was quite a mess before. — Dsimic (talk | contribs) 23:43, 10 December 2014 (UTC)
- (I think my reply further down doubles up as a reply to this, so I guess we can just continue in the further down thread.) Gronky (talk) 23:52, 10 December 2014 (UTC)
- Well, it is already written, here's an excerpt from the Systemd § Overview section:
- Gronky, if you don't know what systemd is, and if you don't understand the article, you shouldn't be deleting information from it. Just because you don't understand something doesn't mean that others are so handicapped. Huihermit (talk) 22:29, 10 December 2014 (UTC)
- I understood what I deleted and I understand the article - which also means I can see that it's badly layed out and missing a lot of basic info. I'd like to turn it into an easily understandable article, i.e. well written, logically layed out, and complete. It's slow work, but the article is already a lot better than how it looked before I got here.
- Given how controversial this topic is, I expect a lot of people come to Wikipedia looking for an overview. I'd like them to get a useful overview, rather than sending them away with a bad impression of Wikipedia.
- You guys insult me because I say I'm no systemd expert. If you mean to imply that you know systemd better than I do, then please prove it by adding something informative to the Overview section. Gronky (talk) 23:07, 10 December 2014 (UTC)
- Please don't get me wrong, I do not want to insult anyone. I might have better-than-average knowledge in some areas, but that by far doesn't mean I'm a superhuman that knows everything under the sun.
- Let's compare your intentions to the article with those of the people who oppose systemd's approach – it's relatively easy to say something isn't as good as it could or should be, but it's not so easy to actually make things better. :) As the opponents of systemd should sit down and write some code to offer a better alternative, could you please say what actually you'd like to improve in the Systemd § Overview section? As I've asked above, which is the first sentence you'd like to be clarified further? — Dsimic (talk | contribs) 23:17, 10 December 2014 (UTC)
- I try to make things better, I've made edits but they're often reverted. The problem is not that I need something clarified, the problem is that the Overview section doesn't give an overview of what systemd is. The only substance is the first two paragraphs, and they're only there because I rolled up my sleaves, did some research and wrote them. But those two paragraphs aren't enough, and here I am talking with three guys who know systemd better than me, and no one's improving the article.
- I'm not sure how else to explain the problem. I can try, but first I have to wonder how we're seeing these things so differently: do you (plural) think the current Overview section gives a real summary of what systemd is?
- If yes, then I'll try to explain why I think it's lacking. If no, then let's discuss improving that section. Gronky (talk) 23:48, 10 December 2014 (UTC)
- I should add that I take none of these comments personally :-) I mention the personal aspect of some comments only to highlight that it's the wrong thing to be discussing. Gronky (talk) 00:06, 11 December 2014 (UTC)
- To me, Systemd § Overview section describes systemd rather well – thank you for your contributions to it! The section does it briefly without going too deeply into various details, and while the first two paragraphs start it very well, the rest of the section provides a more generally oriented continuation of the overview (brings some kind of a "big picture", so to speak).
- Regarding the involvement of personal aspects, we should always try to focus on improving the article(s), but they do inevitably slip in from time to time. :) — Dsimic (talk | contribs) 00:22, 11 December 2014 (UTC)
- Gronky, do you find it any better after the latest few edits? — Dsimic (talk | contribs) 12:02, 11 December 2014 (UTC)
- Dsimic, yes, they are in the right direction but are a lot more minor than what I think this article needs. The article still presents systemd as a simple, logical and uncontroversial technical upgrade. The uproar we saw in the media and the spate of resignations in Debian suggests that systemd is way more than a logical upgrade, so I'm guessing this article is still just a superficial and wordy stub waiting to be expanded. Gronky (talk) 08:40, 31 December 2014 (UTC)
- Well, Thumperward joined the efforts today, so we should see some improvements to the article layout. — Dsimic (talk | contribs) 15:36, 31 December 2014 (UTC)
Layout is one thing (mostly sorted now: there's still too much bullet-point trivia and "controversy" waffle, but every article suffers from that), content is quite another. I'm in full agreement with Gronky that we simply don't do a good enough job of describing what systemd is at the moment. I can think of at least four excellent sources from which to draw this material (the systemd documentation official, Lennart's blog posts on the same, Russ Allbery's documentation of his experiences with it, and Neil Brown's multiple-part LWN series), but it's a lot of work. Chris Cunningham (user:thumperward) (talk) 15:50, 31 December 2014 (UTC)
- It's true that the article could cover and explain systemd much better, especially its various components, but it surely would require at least a couple of days worth of dedicated work. However, I still don't feel that the current "big picture" explanation is that bad. I'm not sure whether I could dedicate that amount of time; however, systemd does deserve good coverage in Wikipedia. — Dsimic (talk | contribs) 16:12, 31 December 2014 (UTC)
- It's going to take a lot of time to do well, but it needs done. The "big picture" is not necessarily bad, but "it's a replacement for init" is a decent enough "big picture" summary by itself. We also need to explain how it does what it does. At present, that's almost entirely confined to the one paragraph that talks about cgroups, and the one paragraph about the config file using a declarative language. Chris Cunningham (user:thumperward) (talk) 16:24, 31 December 2014 (UTC)
- Totally agreed that a much better coverage is needed. I'll review in detail the changes you've introduced once you're done with the restructuring, and will also see to dedicate some time to further improvements. — Dsimic (talk | contribs) 16:37, 31 December 2014 (UTC)
- Oh, you can count me as done with the restructuring for the next 48 hours or so. :) Chris Cunningham (user:thumperward) (talk) 17:19, 31 December 2014 (UTC)
- Hehe, Ok. :) Have fun, and best wishes for the new year! — Dsimic (talk | contribs) 17:30, 31 December 2014 (UTC)
- Good job on restructuring the article, Chris! Looks and reads much better, and I'll see to expand the article further. — Dsimic (talk | contribs) 08:15, 3 January 2015 (UTC)
Gentoo / Gnome 3.8 now requires systemd
So please forgive my laziness, but I suppose this should be added somewhere now. If you want to upgrade to Gnome 3.8 on Gentoo (or any distro from what I can tell) then you must switch to systemd. They have an upgrade guide here. Basically, this means uninstalling udev and OpenRC -- seems quite significant to me and is another reason for me to either dump gnome or just delay the systemd upgrade. Anyway, sorry I don't have the time to update the article myself. Daniel Santos (talk) 07:32, 17 December 2013 (UTC)
- There is an unofficial way of installing GNOME 3.8 on Gentoo without systemd by using the openrc-force use-flag. But it is not supported by the developers. — Preceding unsigned comment added by 85.228.199.174 (talk) 05:53, 29 December 2013 (UTC)
- I think the colors in the "Adoption" section should be reversed. Gentoo not defaulting to one-opaque-process-to-rule-them-all should be the good green, the rest should be the bad red. 72.177.34.205 (talk) 20:00, 18 May 2014 (UTC)
- Exactly what I came to say: here the colors are misleading. Please reverse them. Oh and thanks for the mentioning of openrc-force use-flag. 84.60.221.9 (talk) 06:39, 7 October 2014 (UTC)
- I think the colors are misleading, but more because red/green have a connotation of "good" and "bad" and due to the controversy of this subject I would suggest avoiding them. Perhaps either no color or neutral colors would be more appropriate. Rich0 (talk) 20:11, 5 January 2015 (UTC)
Systemd-the-init-daemon and systemd-the-software-collection?
The "How systemd differs from current init systems" section says
- As well as initiating the operating system at startup, systemd integrates many services which are common on GNU/Linux systems, such as login, device management (replacing udev), and scheduled execution (replacing cron) and provides a unified system for logging these services. Services are all started in parallel, which is faster than the traditional sequential startup sequence.
How much of that refers to systemd-the-init-daemon and how much of that refers to other daemons in the systemd suite? I'm guessing, perhaps incorrectly, that the other daemons in the systemd suite work only if launched by systemd-the-init-daemon, and can't be launched by traditional startup scripts, but, presumably the programs that the other daemons in the systemd suite replace could be launched by systemd, as I think it's capable of running traditional startup scripts for backwards compatibility (along the lines of launchd, until Yosemite, running SystemStarter and its scripts). Guy Harris (talk) 21:33, 4 December 2014 (UTC)
- Hey Guy. I've changed that section title now to be clearer. The article is about the whole systemd suite, so unless otherwise specified "systemd" should refer to the suite rather than just the init system. Gronky (talk) 22:44, 4 December 2014 (UTC)
- Then Type = "init daemon" is wrong here, as the software collection does far more things than your typical init daemon. --94.217.249.128 (talk) 23:05, 17 January 2015 (UTC)
- That's a good point. Fixed the infobox, in which "system software" should fit better as a genre. — Dsimic (talk | contribs) 03:01, 18 January 2015 (UTC)
Adoption / Non-Adoption
Regarding the big controversy, it might be interesting to list distributions that have not chosen to adapt systemd (or even chosen to not adapt systemd). Docdrum (talk) 12:47, 8 October 2014 (UTC)
- Agreed. This might be a good place to start collecting names: http://forums.gentoo.org/viewtopic-p-7648392.html Some (but not all, so be careful) of these distributions switched from udev to eudev in order to avoid a systemd dependency. TTK (talk) 03:56, 23 December 2014 (UTC)
- +1. PCLinuxOS has chosen to stay away from systemd: http://www.zdnet.com/article/hands-on-with-pclinuxos-a-terrific-release 188.113.89.84 (talk) — Preceding undated comment added 21:11, 2 January 2015 (UTC)
- Just found this: http://forums.debian.net/viewtopic.php?f=3&t=118319 an "operating systems without systemd" thread. TTK (talk) 03:38, 27 February 2015 (UTC)
- Also this, which lists not only distributions which do not use systemd, but whose developers have declared the intention to not use systemd in the future: http://www.gabordemooij.com/escape_from_systemd TTK (talk) 03:27, 28 February 2015 (UTC)
Goals of systemd
Hello. I think information about the goals and design philosophy of systemd should be added. I'd really prefer if this is "from the horse's mouth" so to say, rather than third parties speculating about ulterior motives the software and its backers might have. I've already added a brief mention of systemd's goals to the lead, but more information would be much appreciated! --PaulDejean (talk) 00:52, 15 July 2015 (UTC)
systemd-boot
I think systemd-boot (gummiboot) should be mentioned. --Kıverti qwerty (talk) 12:51, 21 August 2015 (UTC)
- As the article about gummiboot says it became part of the systemd project, yes, it probably should. Make it so. Guy Harris (talk) 19:16, 21 August 2015 (UTC)
out of place reference to login shell
- In August 2015, systemd now provides a login shell, callable via machinectl shell.[86]
- It's a misleading simplification. It's an su(1) replacement, not a login shell.
- Its link doesn't reference any controversy. Phoronix does but it's not as good a summary
- Why is it in "History and controversy" section? It is a new feature added over time and thus "History", but nothing else in this section covers systemd's expansion (aka mission creep, tentacles in everything, evil takeover plans, Lennart==satan, etc. :-) ) except in the context of controversy.
There are many other features that systemd has accrued over time. Some are mentioned in "Ancillary components". Repeating these and other to present a History of systemd expansion section would be moderately interesting but a lot of work. Perhaps instead a section following "Ancillary components" for "Other feature expansion" could mention such things like machinectl shell.
Overall, this is a fine article, kudos! -- Skierpage (talk) 21:37, 4 September 2015 (UTC)
Controversy sections
The article's discussion of the controversy in general seems pretty one-sided, actually. I should think such discussion ought to be confined to the Controversy section, but complaints about systemd permeate this entire article. And throughout, almost *all* the arguments are negative. We should try to be impartial here, but the article hardly presents any of the arguments and counter-arguments in favor of systemd. --Android the Andrew (talk) 04:45, 11 October 2015 (UTC)
The article has been cleaned up some now, but could still use some balance. The "Controversy" section includes critiques of systemd, but few (none?) defenses.
Mention of Linus and Kay
I edited the sentence a bit. The criticism of Linus was not about systemd; in a video discussion, he said that he does not care overly much about systemd itself, in one way or the other. So the problem with Kay is not about systemd at all but about patches and modifications sent to the linux kernel itself, NOT systemd. It can perhaps be mentioned in the main article BUT it is not the main topic for systemd here since it is peripheral to the topic at hand. 2A02:8388:1600:A880:BE5F:F4FF:FECD:7CB2 (talk) 19:35, 13 November 2015 (UTC)
Citations for the increasing complexity claim?
It seems to say systemd depends on Gnome? Citation? — Preceding unsigned comment added by Ehrlichja (talk • contribs) 22:24, 2 December 2015 (UTC)
- No, the coupling to which it refers is that GNOME depends on systemd. Perhaps it should state that more explicitly, indicating that the coupling is a one-dimensional dependency. Guy Harris (talk) 22:49, 2 December 2015 (UTC)
List of Linux distros without Systemd
I came here looking for a list of distros with sysvinit/openrc intact and NO systemd. Hope someone works a article which is meant for us, millions of *nix users who want to see a sane system without systemd. — Preceding unsigned comment added by 116.68.127.11 (talk) 18:37, 5 November 2015 (UTC)
- There are couple of web forum threads and a blog entry dedicated to this (and related), links to which I've collected here for easy reference: http://ciar.org/ttk/public/systemd.html TTK (talk) 20:59, 11 February 2016 (UTC)
sysv-rc-conf for systemd
![](http://up.wiki.x.io/wikipedia/commons/thumb/3/32/Sysv-rc-conf.png/220px-Sysv-rc-conf.png)
Is there something similar to sysv-rc-conf for systemd? Since its configuration by far exceeds the complexity of that of sysv, I would deem it necessary. As soon as somebody writes such a program, this should be documented in the Wikipedia. User:ScotXWt@lk 13:39, 13 February 2016 (UTC)
Call to clean up history and controversy section
I would offer to clean up and fix the historical and controversy section, but Wikipedia lately has become so rigid that any changes I've made (even minor ones) are rolled back. Not worth my time to help Wikipedia in this manner any more. RyanNerd (talk) 03:59, 24 February 2016 (UTC)
Thinking about recanting my refusal to clean up historical and controversy section
I made some edits in other pages in wikipedia waiting for them to be rolled back and to my surprise they (so far) are still standing. The "Controversy and History" section although related -- they need to be split into their own sections with the controversy section internally referencing the history section thus making it clear to the reader that the controversy section by nature contains bias and non-neutral language. — Preceding unsigned comment added by RyanNerd (talk • contribs) 17:45, 22 March 2016 (UTC)
systemd 230 KillUserProcess Backlash
I'm thinking we should add a bit on the recent KillUserProcess controversy (here and here) with the release of systemd 230. They changed it to yes by default, so when a user logs out, all other processes they were running are killed as well. It might make sense from a security perspective, however, it breaks processes which were intentionally left running, e.g. screen and tmux. So far, Debian, Fedora, Arch, and Gentoo have gone and reverted that setting. It seems most of the linux community does not appreciate systemd arbitrarily flipping a setting that breaks long standing processes and workflows for the sake of a bit more security (and most notably to fix Gnome issues.) At least, not without advance notification to the community at large. AfroThundr3007730 (talk) 15:09, 31 May 2016 (UTC)
Factual error
consoled was quietly removed from systemd by its author almost a year ago. Jonathan de Boyne Pollard (talk) 21:40, 1 June 2016 (UTC)
- Then fix the factual error by updating the article. Guy Harris (talk) 00:24, 2 June 2016 (UTC)
"Opinions" section does not belong
Since when is Wikipedia a sounding board for anyone's arbitrary reception of something? This adds nothing of value to the article. Merely noting that there is some controversy surrounding systemd is completely sufficient - there is no need to highlight any one individual's opinion of why systemd is good *or* bad. And the few opinion pieces this editor has cherrypicked certainly set an unbalanced tone for the article. No, I'm not proposing adding negative reception, I'm proposing that an entire section dedicated to systemd's reception is pointless and inflammatory no matter how you spin it. 98.86.105.195 (talk) 22:15, 10 October 2016 (UTC)
- I notice that 98.86.105.195 is the same IP from which the edit came (since reverted) which implied that systemd was used in all linux distributions. TTK (talk) 19:28, 23 December 2016 (UTC)
Wrong citation?
At the beginning of the article there is the sentence The name systemd adheres to the Unix convention of naming daemons by appending the letter. It is followed by a citation, which links to an article which says The developers thought that its name is suitably reminiscent of the French term "système D", an expression that relates to "thinking on your feet" and describes high-speed technical problem-solving abilities such as those displayed by TV action hero MacGyver.. Also I am not sure (but also not wanting to claim otherwise) that daemon is really the right term for an init system, so it's not really adhering to it. But might be wrong there.--Athaba (talk) 17:19, 23 March 2017 (UTC)
- "But might be wrong there." Apple seems not to agree with you. Guy Harris (talk) 18:37, 23 March 2017 (UTC)
- And I found a citation for "the name ends with "d" because it's a daemon and daemon programs on UN*Xes have names that end with "d"". Guy Harris (talk) 18:50, 23 March 2017 (UTC)
- It's funny: When that citation was added (2012-05-17, see diff), it was to support the statement "It is also a wordplay on the term System D, which refers to a person's ability to adapt quickly and improvise to solve problems." That sentence was then removed some time later (2012-08-12T21:21:24, see diff), but the reference was kept.—Tea2min (talk) 08:36, 24 March 2017 (UTC)
- OK, I put that sentence back - the comment on the edit was "pls provide a reference for the wordplay", but the reference was already there, although you had to know to look for "système D" rather than "system D" to find it. Guy Harris (talk) 09:49, 24 March 2017 (UTC)
Is it worth mentioning this in the article, which was covered in the press at the time it was announced?
Pwnie for Lamest Vendor Response
Awarded to the vendor who mis-handled a security vulnerability most spectacularly.
SystemD bugs 5998, 6225, 6214, 5144, 6237
Credit: Lennart Poettering
Where you are dereferencing null pointers, or writing out of bounds, or not supporting fully qualified domain names, or giving root privileges to any user whose name begins with a number, there's no chance that the CVE number will referenced in either the change log or the commit message. But CVEs aren't really our currency any more, and only the lamest of vendors gets a Pwnie! [1]
90.155.73.34 (talk) 20:46, 14 August 2017 (UTC)
- While I myself happily admit that I am a critique as well, wikipedia has to be as objective as possible and focus on facts. The situation of individual maintainers, and neither lead developers, is not necessarily central to the software at hand. Perhaps it may or may not go towards an article of any particular person on wikipedia, but I do not think that it should be part of the main systemd article. How many issues do you have in total there? 3000? 5000? 10.000? That will also grow as time passes by ... so now, I do not think that one should focus on individual issues or not per se, from the point of view of wikipedia. Wikipedia should be objective in ALL of its main articles - that is one of its main strengths too by the way.
The main article
Isn't systemd a big umbrella project that includes a init system? The Linux Geek (talk) 16:05, 17 February 2017 (UTC)
- Yes. The init process in the init system is also called "systemd". Guy Harris (talk) 17:40, 17 February 2017 (UTC)
Why don't we move the page to "Systemd (Init)" or something similar? The Linux Geek (talk) 22:09, 30 April 2017 (UTC)
- Because it discusses the entire project. Perhaps you meant "why don't we create a "systemd (init)" page for the init daemon, and link to it from the "systemd" page for the project"? Guy Harris (talk) 22:42, 30 April 2017 (UTC)
Yeah, perhaps I worded that wrong. The Linux Geek (talk) 23:54, 30 April 2017 (UTC)
- The problem is that systemd as a whole is by now a lot more than "just" an init system. The initialization is one component. Wikipedia should note this down, in particular because other comparisons would be unfair. For example, it is unfair to compare systemd to sysvinit due to the former having a lot more integrated functions than the latter, and also having a larger code base. The current article is fine to discuss systemd, but the wording is not completely correct how it starts, aka "is an init system". Yes, it is an init system but it also is a parallelization system for controlling daemons, cgroups and processes. It even has a fallback shell too. 2A02:8388:1603:CB00:3AD5:47FF:FE18:CC7F (talk) 15:10, 11 September 2017 (UTC)
History of the project
It may be better to add a subsection called history of systemd. Right now the article is a tiny bit chaotic and jumps about. I think a history article would be useful - there, one can also link in to other parts that are only marginally related to systemd itself, such as the situation of linux distributions adopting or not adopting systemd. 2A02:8388:1603:CB00:3AD5:47FF:FE18:CC7F (talk) 15:12, 11 September 2017 (UTC)
"increasingly complex"
The whole second paragraph of this public article reads like one big kvetch. Does a casual learner really care about the internal conflicts among kernel developers being one of the first ideas they ever read about this system? Fact: Almost all modern Linux systems use an implementation of systemd. Opinion: systemd is overly complex.
I feel like if this stays in the article at all it should be moved to its own "Controversy" or similar section. It does not help people learn about systemd or what it does.
Chaim1221 (talk) 18:54, 21 March 2018 (UTC)
- +1 Fixed by removing the kvetch. KMeyer (talk) 02:21, 20 May 2018 (UTC)
- FSDO "almost all" = about 75%, and that 25% includes Android, the most widely used Linux distribution today. The wording of your "fact" seems a bit biased. I'm not trying to pick a fight, just saying that if you want to avoid opinions in this article, you should set a better example. TTK (talk) 00:15, 9 June 2018 (UTC)
Does this page need a complete list of all systemd bugs, fixed and unfixed?
"In September 2016, a bug was discovered that allowed any unprivileged user to perform a denial-of-service attack against systemd."
Yes? And? The bug was fixed, no?
Is it now notable that some software had a bug?
HughesJohn (talk) 09:29, 17 February 2018 (UTC)
- +1 Why not throw in every other bug systemd has ever had to make the list complete? It's pretty silly. KMeyer (talk) 02:20, 20 May 2018 (UTC)
- Reasonable questions. Bug completionism would indeed be silly. The answer is that notability depends on the bug. The most notable are generally security bugs: they allow malicious actors to do bad things to victims' computers. That particular 2016 bug, for instance, allows a malicious local actor to perform denial of service attacks.
- Whether a particular security bug is notable or not is the next question. The answer largely depends on ease of exploitation and extent of damage. At one extreme is the Heartbleed security bug, which is notable enough to have an entire Wikipedia article devoted to it, while at the other extreme are bugs that are practically harmless. I don't know enough about that systemd bug to place it on this spectrum.
- I'll change the text quoted above from "a bug was discovered" to "a security bug was discovered", which suggests to readers why the bug might be notable. I'll also add a security bug discovered in 2017, CVE-2017-9445[3], which was notable enough to headline an article by prominent technology news site The Register[4].
- Lastly, I recommend asking a different question: Have security bugs in systemd been significant enough to warrant a section called "Vulnerabilities"? Usually two facts do not a section make, but perhaps there have been other, serious security bugs in systemd that I don't know about. My vote is no for now. Adelphious (talk) 03:15, 6 July 2018 (UTC)
Deletion of Negative & Controversial Aspects of Systemd's History (Attn: KMeyer)
Hi KMeyer. Please can you explain why you want those parts of Systemd's history removed? People have taken the time to research, cite, write and link up that information. The information is also true and an important part of Systemd's history that I personally remember. I am perfectly willing to contemplate any suggestions to remove those paragraphs. I am not anti-Systemd. I am pro-truth. Thank you. Akima AE (talk) 01:13, 3 July 2018 (UTC)
- Hi Akima AE. Simply scroll up on this talk page to see some of the earlier discussion on removing the sections. Your 2 July edit adds tabloid-level gossip. Unfortunately, just because something has had some effort put into it and is factual does not mean it meets the bar of notability or relevance.
- Devuan might belong on a Debian page, but political Linux distro forks are created all the time and we don't list all of them (outside of maybe "List"-type articles) because they are generally not notable. (Distrowatch's popularity index (not a perfect representation, but useful as a rough metric) puts Devuan well below notable.)
- The Shuttleworth quote is pure tabloid. "Oh, look, some millionaire had to eat his words." If you think it's worth including in Shuttleworth's article, great, but I don't think it adds anything to the Systemd page.
- Yes, the adoption of systemd was contentious and fiercely argued. No doubt about that. I believe the article covers that better without the gossip.
- I do not feel strongly about systemd. And I too am "pro-truth" (who isn't?). All the best, KMeyer (talk) 17:26, 8 July 2018 (UTC)
- Hi KMeyer. Thanks for responding. I didn't write either of those paragraphs you have once again deleted. I only restored them.
- I disagree with your stance on this matter. The wording of the deleted history components could be better, sure, but you haven't reworded it to simply improve the quality of the article, you have removed the content entirely. Right now the history section gives the false impression that Ubuntu simply adopted systemd after Debian. This doesn't paint an accurate picture of what happened. The paragraph that follows the sentence about Ubuntu adopting systemd also doesn't accurately portray history. It could easily be interpreted by a reader that the reason those 4 Debian people left the project was because of technical difficulties integrated systemd. Obviously this is not the case. The now-deleted content about Devuan made it more clear that the stress those 4 Debian people experienced was due to a divide within the Debian community about whether Debian should adopt systemd or not.
- Sadly I don't have the time to resist what you've done, so the systemd history section will likely remain in this state and readers will be misled. There is very little controversy surrounding the gradual work done to transition GNU/Linux desktops from X11 to Wayland, however there has been a great deal of controversy and unwillingness among some users and developers to transition popular GNU/Linux desktops towards usage of systemd. As such I would expect this -- now major -- component of a GNU/Linux system to have a wikipedia page that accurately portrays that controversy. I think that if you truly care about truth, then you wont leave the article in this state. It is better to have some slightly lower quality content that -- left in place -- paints a more accurate picture of history, than to have higher quality content which creates a false picture of history. Akima AE (talk) 17:11, 11 July 2018 (UTC)
notsystemd not defunct
The section on notsystemd states "Development of the software ceased in September 2017 and the project may be defunct." It was neglected for a bit, but I was hoping to release v239.1 today... because of some lingering bugs, I don't think it's going to happen tonight. Anyway, not defunct. I don't want to edit the article myself because I'm the primary person behind notsystemd, and WP:COI. That said, I'm surprised (and flattered!) that someone thought notsystemd was notable enough to include in this page (actually, looking through the page history, it was added to the page by a fellow Parabola developer, User:Megver83). ~ LukeShu (talk) 00:58, 22 July 2018 (UTC)
- notsystemd v239.1 is now out: https://lists.parabola.nu/pipermail/dev/2018-July/006882.html ~ LukeShu (talk) 06:14, 24 July 2018 (UTC)
- Fixed in http://en.wiki.x.io/w/index.php?title=Systemd&diff=851637405&oldid=851340053 . Can you point to the development branch of notsystemd? https://git.parabola.nu/~lukeshu/systemd.git/log/?h=notsystemd%2Fmaster still appears defunct — no activity from 2017. KMeyer (talk) 21:37, 28 July 2018 (UTC)
- Being released as a patchset, it has a rebase-based workflow; looking to git for the history won't be terribly helpful. And even then, I don't really mean for the git repo to be authoritative (if I did, it would be under /packages/, not /~lukeshu/), the authoritative source of the source code is https://repo.parabola.nu/other/notsystemd/ . The notsystemd/master branch has a lot of skunkworks-type stuff that has never actually been included in a release, so it was never a good place to look for development activity. A better place to look to gauge current development would be "notsystemd/wip/${largest_number}". ~ LukeShu (talk) 05:10, 29 July 2018 (UTC)
- Ok. That is kind of an uncommon development pattern but makes sense. If possible, I would encourage you to remove stale "master" branches and instead publish your actual WIP/master branch in a single branch. Yes, it will require force pushing on that branch, but that makes sense for a rebase-based patchset project. Of course, it is your effort, do as you wish :-). All the best, KMeyer (talk) 15:02, 7 August 2018 (UTC)
- It's not too uncommon for other patchset-base projects. I've renamed many of the branches to have better names (the old "notsystemd/master" is now "notsystemd/reorg0/master"), and introduced a new "notsystemd/master" branch that tracks releases the same way that "systemd/parabola" tracks what Parabola ships as plain systemd. I do publish my WIP/master branch as a single branch, it's named "notsystemd/wip/v${version_it_is_based_on}", or I may be working on a feature branch off of that. In general, I won't force-push to a branch that doesn't have "wip" or "rebase" in the name, as to manage expectations of "are commit ID's in this branch meaningful?" ~ LukeShu (talk) 02:34, 17 August 2018 (UTC)
Configuration of systemd
![](http://up.wiki.x.io/wikipedia/commons/thumb/3/32/Sysv-rc-conf.png/220px-Sysv-rc-conf.png)
sysv-rc-conf
is a TUI-based tool to configure sysvThe section "Configuration of systemd" should be reworked. Screenshots of **ALL** available graphical configuration tools belong here!
- systemd-manager for GNOME based on GTK
- systemdgenie: for KDE based on Qt
- etc.
User:ScotXWt@lk 14:53, 6 January 2019 (UTC)
"at the cost of abandoning long-standing compatibility with POSIX and other Unix-like systems."
I removed this clause because it had nothing to do with the preceding one and I was unsure how to rework it to make it a useful part of the article. To some extent it is accurate, in that systemd itself is not portable software and cannot be run on any POSIX system — only Linux[2]. But the removed clause made a stronger claim than that. It seemed to imply Linux was no longer POSIX compatible, which is not true, or that systemd had long-standing portability, which is also not true. If someone feels it is valuable and can rework it to make sense, I do not object. KMeyer (talk) 14:56, 7 August 2018 (UTC)
- The "POSIX" part of the claim you removed is nonsense; there is nothing in POSIX about the details of the init system. There are Single UNIX Specification-compliant systems that use SV init and there are Single UNIX Specification-compliant systems that use launchd, for example.
- I suppose you could say "many Unix-like systems other than Linux use an SV-style init, so abandoning SV init in favor of systemd makes Linux distributions less like, say, Solaris", but the *BSDs are also "Unix-like systems", and use BSD init plus various init script schemes, so Linux-with-systemd and Linux-with-SV-init (and other systems with SV init) differ from *BSD.
- So there's less to the "ZOMG systemd is different from what UNIX has!" than people who make that complaint might think, i.e. 1) POSIX doesn't say what a Unix(R) has to have and 2) "UNIX" doesn't have an init system, different UNIXes have different init systems, unless you define "UNIX" as Research UNIX (in which case the *BSDs are the least different from "UNIX" in that regard - less so than UNIX System V!). Guy Harris (talk) 22:47, 12 February 2019 (UTC)
References
- ^ https://pwnies.com/winners/#lamestvendor
- ^ Poettering, Lennart (30 April 2010). "Rethinking PID 1". Pid Eins. Retrieved 7 August 2018.
It should be noted that systemd uses many Linux-specific features, and does not limit itself to POSIX.
Unclear wording
The following wording is not clear to me: "the first process (PID 1) has a special role on Unix systems, as it replaces the parent of a process when the original parent terminates." That is, I cannot tell from the sentence how the first process parent process works. It seems to imply that when a process parent terminates, the process ID of the parent process becomes "1". I think. Someone who understands this better than I do should take a stab at rewording this sentence fragment to clarify how this works. Other than that, the grammar and punctuation around this sentence fragment is excellent. — Preceding unsigned comment added by Ignus3 (talk • contribs) 18:50, 14 February 2019 (UTC)
- What that sentence means is "when a process terminates, any processes that had that process as the parent process are reparented so that process 1 is now their parent process"; I think you're saying the same thing, i.e. "the process ID of the parent process becomes "1"" is referring to the "parent process" attribute of a process changing from the old value, the PID of the process that terminated, to a new value of 1. Guy Harris (talk) 19:03, 14 February 2019 (UTC)
Lower Case Title vs MOS
I see a bunch of people going around in recent months change systemd to lower case. See here for what the Wikipedia Manual of Style has to say on it. Excerpts
With the exception that immediately follows, trademarks promoted without any capitals are capitalized like any other:
- use: I found a Thirtysomething DVD and a pair of Adidas shoes while browsing Craigslist.
- avoid: I found a thirtysomething DVD and a pair of adidas shoes while browsing craigslist.
In the article about a trademark, it is conventional to give the normal English spelling in the lead section, followed by a note, such as "(stylized as ...)", with the stylized version (which may include simple stylization, like capitalization changes, decorative characters, or superscripting, but not colorization, attempts to emulate font choices, or other elaborate effects),[c] then resume using an alternative that follows the usual rules of spelling and punctuation, for the remainder of the article. In other articles that mention the subject, use only the normal English spelling, not the stylization.
--Voidvector (talk) 01:24, 15 July 2018 (UTC)
- Is systemd, in fact, a trademark for the systemd project or for the daemon in that project or both?
- If not, is the idea that the MOS item in question also applies to non-trademarked names that refer to a particular item - systemd, for example, refers the daemon and to the project? Guy Harris (talk) 01:34, 15 July 2018 (UTC)
- There is the concept of unregistered trademark, which are trademarks not legally/commercially registered. In addition, MOS:TM mentions LaTeX, which I am fairly sure is in the same legal status category. --Voidvector (talk) 02:10, 15 July 2018 (UTC)
- Yes, I know, that's why I said just "trademark", not "registered trademark".
- But what causes a word to become an unregistered trademark? Merely using it as, for example, the name of a particular thing? Guy Harris (talk) 02:53, 15 July 2018 (UTC)
- IANAL, but since Systemd people care so much to go around Wikipedia enforcing their branding/stylization, I take that as an indicator that they are treating it as a trademark that's worth protecting. --Voidvector (talk) 04:59, 15 July 2018 (UTC)
- I.e., they've been changing Wikipedia pages to say "Systemd" rather than "systemd"? Guy Harris (talk) 05:37, 29 July 2018 (UTC)
- There were a number of edits in the past year on other Linux=related pages of people changing "Systemd" to "systemd" even though the proper grammatical usage should "Systemd" (i.e. start of a sentence). --Voidvector (talk) 22:12, 29 July 2018 (UTC)
- Please read more carefully: "However, if the title of the article is the stylized version of the name (e.g. iPod), it should be given in the boldfaced title recapitulation at the beginning of the lead (i.e., without a 'stylized as' note), and used throughout the text". The "systemd people" are not protecting a trademark. They are well-intentioned users fixing an error. 2600:1015:B048:4698:D42F:3CBA:C203:754C (talk) 00:52, 22 January 2019 (UTC)
- Hello, I'm new to Wikipedia and was just wondering, why not edit it to say something like "The systemd software suite is designed to provide a fundamental building block for a Linux operating system." This allows us to maintain the style preference of the software designer, while also ensuring that MOS:TM isn't violated. Would that violate some rule I'm not familiar with? ThatUnnamedRedShirt (talk) 00:18, 20 February 2019 (UTC)
Systemd GUI Interfaces
There have been multiple UIs shown on in the past systemd-ui and currently systemd-manager that are no longer maintained. We should remove unsupported UIs. No section within the article discusses systemd-manager either. Specifically `https://github.com/mmstick/systemd-manager` has been deleted and is now abandoned. [1] 108.45.63.81 (talk) 01:09, 13 September 2019 (UTC)
References
- ^ https://aur.archlinux.org/packages/systemd-manager/ "Seems upstream ceased development and remove the source repo. abandoning."
"ConsoleKit" listed at Redirects for discussion
![](http://up.wiki.x.io/wikipedia/en/thumb/2/28/Information.svg/30px-Information.svg.png)
An editor has asked for a discussion to address the redirect ConsoleKit. Please participate in the redirect discussion if you wish to do so. All the best: Rich Farmbrough (the apparently calm and reasonable) 21:29, 29 February 2020 (UTC).
21:29, 29 February 2020 (UTC)
Hard-coded name servers
I'm surprised there is no information about the hard-coded name servers. There was quite a bit of criticism and a vivid discussion. --78.35.82.207 (talk) 05:27, 29 January 2021 (UTC)
Security Vulnerabilties
Systemd has accumulated a rather large number of CVEs over the years, I wonder what the cost to society this software has caused due to the possible increase in cybercrimes as a result of its widespread usage? — Preceding unsigned comment added by 86.130.94.30 (talk) 19:51, 3 October 2021 (UTC)
- I don't think cybercrime rate is increased due to systemd having bugs. I think that is a bit hyperbolic. My question is, is there an exorbitant amount of CVEs? Or is it the notoriety and widespread usage of software like systemd that lead to these CVEs even being discovered in the first place? Leaning towards the second personally. Mpaglamas (talk) 15:21, 12 November 2021 (UTC)
Listing of the greek version
Hello, I am writing the greek version of the article on systemd, and I cannot seem to find how to add the rest of the languages on the sidebar on it. I added the greek version on this article on Wikidata, but on Greek Wikipedia the "Languages" list is not updated with the other languages. Adding any language on it manually throws an error (as expected), because the article has already been added on Wikidata. Any help would be appreciated. Mpaglamas (talk) 15:14, 12 November 2021 (UTC)
- @Mpaglamas: I've purged the page for you and now the links appear. Cheers, Hoo man (talk) 15:18, 12 November 2021 (UTC)
- Thanks a lot! Reading the link you provided now. Mpaglamas (talk) 15:22, 12 November 2021 (UTC)
Citation Needed
I think the [citation needed] tag on the summary of the page is a bit unwarranted. What kind of source can be provided for the following phrase?
- Since 2015, the majority of Linux distributions have adopted systemd, having replaced other systems such as the UNIX System V (also know as SysVinit). Systemd may be one of the most, if not the most polarizing piece of software in the Linux community.
The first statement (for SysVinit) is confirmed by multiple sources on the summary and on the history, so I did not add it here too. (Perhaps I should have?)
The second statement is an opinion granted, but I really don't think anyone on the *nix community would disagree about System being the most controversial project. Meaning either someone likes or dislikes the project, it being immensely controversial is objectively true. I can't think of any project that has stirred the waters more, thus the phrasing on the text.
But I'll let other user share their thoughts before editing the tag. Mpaglamas (talk) 11:38, 17 November 2021 (UTC)
Forks section includes software that is not forked from systemd
E.g. ConsoleKit2 and s6 are not forked from the systemd codebase. Also elogind and eudev are more or less just patch sets on top of systemd rather than full fledged forks. They sync with systemd regularly.
Thoughts? What is this section intended to convey? Can we use different terminology to be more accurate?