Talk:Linux

I dispute the existence of "Linux" as a family of operating systems

A. Linux is a kernel, which is a component of an operating system. Many operating systems use Linux, most of which are GNU systems (For further context see my proposal for GNU as a family of operating systems), however the Linux kernel is not the unifying characteristic that should be used to determine what OS family the software distribution falls into. Chronologically the GNU project came first and operating systems that derive from it naturally ought to be categorized in terms of their historical lineage as such, derivatives of the GNU Project, therefore the GNU family of operating systems. Other operating systems which use Linux but don't belong to the GNU family of operating systems, are simply Unix-like operating systems, not meaningfully a part of any particular OS family that I know of. Linux is a kernel Unix-like operating systems make use of.

B. I took a look at the archived page of the first citation after the pronunciation citation, 17. Linux#cite note-23 archived at archive.org and this citation on the Wikipedia appears to be a misquotation, on the Wikipedia page on the citation it says "The shared commonality of the kernel is what defines a system's membership in the Linux family; the differing OSS applications that can interact with the common kernel are what differentiate Linux distributions." But this is not what the archived source says! According to the that archive source on what is presumably page 33 it says "kernel The central, core program of the operating system. The shared commonality of the kernel is what defines Linux; the differing OSS applications that can interact with the common kernel are what differentiate Linux distributions" provided I didn't misspell anything. This is absolute nonsense, a blatant lie and a misquotation! First there is absolutely nothing about any "family" of operating systems in the source cited which makes this mis-citation, and second, this segment is used to define what "Linux distributions" are not a "Linux family". I argue there is a difference in the use of "Linux distribution" as a colloquial or casual term refer to operating systems making use of Linux, versus a technical categorization of operating system families.

C. Furthermore the archived source says that Linux is supposedly an operating system originated by Linus Torvalds, which is blatantly false, as the operating systems which use GNU software obviously share an older common origin dating back further into the founding of the GNU Project by Richard Stallman than any Linux kernel added in later. Essentially that sourced article is a misrepresentation, and a poor record of (or even intentional attempt at rewriting) history, thus I say an unreliable source. Linus Torvalds did not create an operating system, his work is not even what brought about any operating systems, his contribution was a kernel, a component that would be used to form complete system distributions out of Richard Stallman's GNU operating system.

D. To be clear, my case is that GNU is a family of operating systems. Since Richard Stallman did not complete his operating system where third party entities formed their own complete system distribution out of Richard Stallman's GNU Project, those independent distributions of software are thus distributions of the GNU family of operating systems because they historically derive from the software by the GNU Project + with the Linux added. Note that my argument here isn't on the GNU/Linux naming controversy, this is about the categorization of the GNU family of operating systems, and that there is no Linux family. My post here does not dispute the colloquial use of "Linux distribution" as a casual way to refer to any operating systems, I am disputing the existence of a supposed 'Linux family of operating systems'. I argue the historical lineage is what should be used to classify them under any families and umbrella terms. The historical lineage of the operating systems claimed to be a part of the Linux family of operating systems do not have their historical roots in the Linux kernel, therefore Linux has no claim for a family of operating systems, it is a component used in many different Unix-like operating systems. Softwareperson1000 (talk) 13:20, 2 May 2026 (UTC)

So the GNU family includes both Linux-kernel-based systems with userlands based on GNU software (and including other non-GNU software), Hurd-based systems based on GNU software (possibly including other non-GNU software), and possibly other systems.

The subset of the GNU family consisting of systems using the Linux kernel have features based on Linux kernel mechanisms that systems based on kernels lacking those features do not have; the same applies for the subset of systems using the Hurd, mutatis mutandis.

There are also system using the Linux kernel and a non-GNU-based userland, such as the one on the majority of smartphones out there. Those systems may have many of the same Linux kernel features that the ones with GNU-based userlands do, but will not have GNU-userland features not present in their userlands.

The Linux-kernel-based subset of the GNU family is a significant subset; you probably won't have, for example, cgroups in non-Linux-kernel-based members of that family. Guy Harris (talk) 03:11, 3 May 2026 (UTC)

That's essentially it. Let me further explain why and elaborate:

1. My line is that the historical lineage is what is mainly used to determine what OS family the system belongs in, minor technical variations aren't enough to sway that, so long as the overall design is similar. Differences do become relevant when the overall design differs too much (I'll elaborate on this later in the Android section of this post). You will notice for example if you look at the various *BSD systems that they have differing features from each other, yet they belong in the BSD family. OpenBSD is different from FreeBSD which is different from NetBSD etc.

2. The larger Unix-like family of operating systems (which GNU also belongs in), includes many operating systems which use the Linux kernel, they are all Unix-like because they follow a Unix-like design philosophy and behave as Unix. Most of the system distributions which make use of Linux just happen to be a part of the GNU family, but they don't have to be and there is no inherent connection. Let's look at one Alpine Linux as an example operating system which doesn't use GNU software but does use Linux, so what to make of that one might ask, well it's an Unix-like system and that's it. Unless there is a better categorization for Alpine I don't know of, and I argue Linux isn't one, then there is no more to it than that, it's just a Unix-like system among many. Linus Torvalds developed a kernel used on many Unix-like systems.

3. I do have doubts that Android qualifies as an Unix-like operating system though, to me it seems like Android shouldn't be called Unix-like, for more reading I also posted here. Android would thus be it's own independent operating system altogether, so it isn't Unix-like. This would mean Linux is also used on non-Unix-like operating systems as well, even if Android does contain Unix somewhere inside though, but that's not too important here as this is a different discussion whether Android is Unix-like. I point out here that Android at least does not support the case for "Linux as an operating system family" because Android is too different from the more Unix-like systems which I argue are the GNU family of operating systems, to be put in the same family.

In fact the technical composition and features do become relevant once they are great enough like the overall Unix-like design itself. Minor technical differences, are not important enough to warrant not putting the systems as a part of the same operating system family, in that even binary compatibility (ABI compatibility) for example is not expected from systems that are in the same OS family anyway. There is no established precedent that would suggest ABI compatibility is required for systems to qualify as a part the same family. Many Unix-like systems categorized to be in the same family are often binary incompatible and have many small differences in terms of features too, but here is the important fact, they all retain the same overall Unix-like design and fundamentally they all follow the Unix philosophy and also roughly work the same way. Which means the internal variations within the family are just the natural variations of the Unix design put in practice. It's only when a system strays off from the Unix design like how I see Android has done, that I would say such a system is no longer Unix-like and therefore it should not be included in another different OS family that is for Unix-like systems which is what the hypothetical "Linux family" would have to generally be.

If Android is determined not to be Unix but did hypothetically still somehow get included in the supposed Linux family, the Linux family would be so diverse and indeterminate as to be an incoherent mess, which isn't even entirely Unix-like anymore. I don't think sharing one OS component is enough to warrant the inclusion of a totally different OS design as a part of the same OS family. Basically put short, the Linux family would be too big (in terms of scope of different OS designs) if it included Android as a non-Unix system, which I say Android probably is that. Softwareperson1000 (talk) 11:35, 3 May 2026 (UTC)

This article is not the place to put forward novel arguments (WP:OR) that contradict what sources state (WP:NPOV) with regard to Android and any GNU family of operating systems. - Aoidh (talk) 17:54, 3 May 2026 (UTC)

A. "novel arguments" You are just calling it a name, I will point out that categorizing and classifying operating systems based on historical lineage is standard practice, and is not my original research. Me pointing out an error in many articles failing to follow such an established convention on Wikipedia, is not original research. Also an "Unix-like" operating system is a well established term and a concept in the world of computing, and is most definitely not my original research.

B. For me to point out that I am not convinced Android as being categorized under that "Unix-like" concept, is in fact supported by the lack of sources that show Android is Unix-like, rather than my "original research" suggesting it isn't Unix counter to some well established line. In that, who ever said Android is Unix-like? Even on the Wikipedia page on Android the term "Unix-like" appears twice (not counting the links near the bottom of the page), once as the 'OS family', and second time is literally an important figure disputing Android being Unix-like (according to the Wikipedia article itself). Actually, I would appreciate for you to show me the sources you speak of that show Android is in fact an Unix-like operating system. But I don't think I need to prove Android is not an Unix-like system, when given the context, the burden of proof to show it is Unix should be on those who wish to prove that. I can't actually on a quick glance see on the Wikipedia page on Android itself a good source saying it is Unix-like, so I ask you, what sources?

C. Also it is important to retain focus whilst discussing this, my argument is still on the Linux as an operating system family, the Android issue is only adjacent to my actual argument about the Linux as an OS family.

D. Whether GNU as an operating system family is "original research" is unclear to me, however, the way it seems to me is that Wikipedia made-up the Linux family of operating systems already as is, so to that end why accept Linux as a family but not the GNU? Again need I remind you, of what I said earlier, the instance of Linux as a "family" being cited on this article was literally a misquotation, the word "family" did not once appear in that archived source the citation pointed to, if that's the case in support of "Linux" as a family then how is that not original research? Softwareperson1000 (talk) 19:27, 3 May 2026 (UTC)

To be clear, I have no issue with removing the "family" verbiage, which was added when the quote was altered in 2017. However, suggesting the addition of GNU family of operating systems would not comply with WP:NPOV given the lack of sourcing supporting that. As for the Android issue, this isn't the talk page to discuss issues with that article; I would suggest raising that issue at Talk:Android (operating system). - Aoidh (talk) 02:54, 4 May 2026 (UTC)

1. Fair enough on the Android thing to be raised as an issue on that other Talk page, however I did initially bring it up here because it is still relevant to my main argument on the "Linux family", because if Android is indeed included inside such a Linux family that causes trouble for that OS family concept itself, because that would mean this supposed OS family contains both Unix-like an non-Unix-like operating systems at the same time, which is nonsense.

2. Given you gave an OK to removing "family", I need to ask you, just how far does this go? In that many Wikipedia articles on the various operating system distributions, which make use of Linux, say that they are a part of the supposed Linux family. An example: Arch Linux on the article reads "OS family Linux (Unix-like)", there are no citations to back up the "family", and that word only appears that one time right there. I would either replace that with "GNU (Unix-like)" or simply "Unix-like". If GNU is found not to be a good option (continues in 4. of my post) then I am fine with "Unix-like" as the sole family.

3. So long as the 'OS family' should exist on these Wikipedia articles, for each operating system, there is a taxonomic need for some sort of such classification and therefore a name to call it. Now that I have presented my case for why "Linux" is wrong for such a use, something has to replace it. My main proposal for that replacement is the GNU family of operating systems because it is clear and evident that out there in the world there exists operating systems based on the use of GNU software and are historically derived from it as I see it.

4. The issue of GNU family as original research; I am open to that possibility that my proposal for the GNU family is in fact original research as per Wikipedia's definition, but I'd still like to argue in support of it. My understanding is that Wikipedia's rules are not intended to cause the editor to ignore basic common sense and ignore basic facts and reality and write false nonsensical statements and incorrect information based purely on a lack of good sources. The question is of which one takes priority, writing nonsensical statements into article so that one can adhere to the spirit of a reference site (as understood by the editor), versus writing an accurate record of human knowledge despite the lack of good sources, there is an inherent conflict in Wikipedia's mission in that sense. When there are no good sources at all for an article to exist in the first place, then that article should not exist as per the policy, however the presence of the articles on these OS systems is already justified, the only question here we have now is just the classification "OS family", specifically (which may already be a concept Wikipedia made up anyway (as used on Wikipedia in this form)). If it is determined that GNU is not a good option to replace "Linux", then I am willing to compromise and reach an agreement in saying it should then be just "Unix-like". Using just "Unix-like" then serves as a fallback for my argument here in opposition to the "Linux family", we can therefore dismiss the GNU issue here for now (I will still argue in favor of it elsewhere on a separate Talk page). So do we have an agreement then that each such Wikipedia page should reflect "Unix-like" as the sole 'OS family'? I am happy with that compromise and I'll drop the "GNU family" as an issue here on this Talk page for now, until I can find good sources for it in the future perhaps. Softwareperson1000 (talk) 10:31, 4 May 2026 (UTC)

GNU would not be an appropriate family for Linux distributions for several reasons. First, it is an unsourced label. Second, though all Linux distributions are based on the Linux kernel in some way, not all Linux distributions or Linux-kernel based operating systems use GNU software. This wouldn't be an improvement and would create issues where something like Alpine Linux wouldn't be considered the same family as Arch Linux, and Gentoo Linux may or may not be part of a GNU family depending on how you configure it. I cannot find any third-party reliable sources that state that GNU is a family of operating systems, and while the removed source did not contain references to Linux being a family of operating systems, I pulled out my old copy of the CompTIA Linux+ Study Guide (ISBN: 978-0-470-50384-3) and Page 2 states that Linux isn’t a single OS, but rather a family of OSs... and I've also found references to such online (e.g., [1][2]). Given that there are, in fact, sources supporting the family verbiage, I'd be interested in hearing User:Guy Harris's opinion on whether that should be restored. - Aoidh (talk) 17:57, 4 May 2026 (UTC)

A. I can find a few odds and ends of some miscellaneous instances of "GNU/Linux" being used as a "family". Also it doesn't have to be just "GNU" specifically, any similar variation (including GNU/Linux) essentially serves the same role for the context here to replace "Linux".

https://www.ubishops.ca/wp-content/uploads/bucstr-2014-001.pdf

https://www.iitj.ac.in/PageImages/Gallery/03-2025/IITJ_AR_2022_2023_English.pdf

https://blog.dataumbrella.org/juan-luis-opensource-journey

https://plusformacion.us/what-is-the-full-form-of-gnu/

https://digerty.com/guides/free-libre-and-open-source-what-does-that-really-mean-no-really-really/

Mageia as a notable example of an OS distribution calls itself an "operating system of the GNU/Linux family"

https://sourceforge.net/software/product/Mageia/

https://slashdot.org/software/p/Mageia/

B. Even if GNU wasn't the way to go, I still don't see why simply "Unix-like" isn't good? It's already "Linux (Unix-like)" on the many Wikipedia articles, simply dropping "Linux" would just leave "Unix-like" as the 'OS family', that's fine by me.

C. Regarding Gentoo, Gentoo is a project, that distributes many systems, you will find if you look into it, that not all Gentoo variants with it's Portage package even use Linux. So in fact, you are wrong in asserting that Gentoo must be in the same supposed "Linux" category as some of those others, Gentoo also distributes for other kernel bases. Also, let me bring you some news you might not yet be aware of: Gentoo has officially announced a GNU Hurd variant https://www.gentoo.org/news/2026/04/01/gentoo-hurd.html so Gentoo is not Linux. Softwareperson1000 (talk) 18:11, 4 May 2026 (UTC)

A few more:

https://www.ubuntubuzz.com/2020/11/kubuntu-2004-lts-review-familiar-operating-system.html and https://www.ubuntubuzz.com/2021/04/download-fedora-34-full-editions.html written by same person

https://hackaday.com/2019/11/05/will-the-real-unix-please-stand-up/

https://www.physics.ntua.gr/konstant/ComputationalPhysics/Book/ComputationalPhysicsKNA.html

As to whether you think these are any good as sources I don't know, but there certainly are some third party uses of "GNU/Linux" as a "family" Softwareperson1000 (talk) 18:15, 4 May 2026 (UTC)

For Linux as a family: the references seem good. I gave a look at https://link.springer.com/chapter/10.1007/978-1-4842-1392-6_3, but didn't check to see if any of the distributions he cited were "GNU/Linux" distributions as opposed to "some-other-userland/Linux" distributions.

For GNU as a family: Gentoo isn't the only case of an OS that's "GNU but not Linux", Debian/kFreeBSD was another case (according to https://www.debian.org/ports/kfreebsd-gnu/, development stopped in 2023). There are also, as noted, GNU/Hurd systems.

For Unix-like as a family: systems with a Linux kernel generally fall into that category, but they also form an interesting subcategory of that category, given that there are APIs that don't exist in other Unix-like systems, and daemons and command-line utilities that use those APIs. That's similar to macOS, which is not just Unix-like, it's a Unix(R), but there are APIs that aren't in other Unixes. (As far as I'm concerned, if you don't do at least one thing differently from other Unixes, you don't qualify as a Unix. :-)) It's listed in macOS as belonging to families Mac, Darwin, BSD, Unix-like, and Unix (the latter two could be considered redundant, as any system passing the Single UNIX Specification could be considered "Unix-like" as well as being a Unix(R). I'd say most Linux distributions could fit into "Unix-like", "Linux", and "GNU" if the latter two are considered OS families. Some might be "Linux" but not "GNU" if they don't have a GNU userland.

For Android as a Unix-like Linux distribution: Bionic has most, but not all, of POSIX implemented (https://android.googlesource.com/platform/bionic/+/master/docs/status.md) - some of the exceptions appear to have been omitted for security reasons. Termux (https://termux.dev/en/) is an add-on that provides a terminal emulator and may also add some tools not shipped with Android.

For Gentoo: there are multiple systems using a GNU userland and Portage. The one that uses a Linux kernel could be considered a "Linux" or a "Linux distribution". The ones that use other kernels might be considered "GNU" systems but not "Linux" systems. The same applies to Debian. Guy Harris (talk) 19:02, 4 May 2026 (UTC)

A relevant discussion discussing the Linux (Unix-like) label on distros: Talk:Linux/Archive 52#Replace the 'Unix-like' field with 'Linux' or remove the field. - Aoidh (talk) 22:15, 4 May 2026 (UTC)

1. That's an interesting idea, of including more than one family as Guy Harris here suggested above, pointing to the example of MacOS, which has more than one entry there for 'OS family'. I hadn't thought of that. To me the issue was that having Linux is wrong, but it seems like a good idea to have more than one entry instead, that solves our problems. Since if GNU nor "Unix-like" alone are found to be any good, I will agree to a compromise such that more than one entry is included. So it would look like this with each distro falls into each separately as appropriate per system distribution, where ones which use both Linux and GNU are therefore in both families, GNU ones without Linux are GNU alone, and the Linux ones which aren't GNU are just Linux then. That makes sense and fixes the problem I perceive with Linux exclusively as the OS family. I still hold that as a priority for me "GNU" (or variations) should be the 'OS family', however indeed I don't see an issue with there existing an overlap with more than one family existing separately and simultaneously as with Mac OS which seems to set up a good precedent which should be followed, which I hadn't considered. So do we then have an agreement with the proposal of Guy Harris to have more than one family?

2. I must still point out, one single entry that covers each GNU, Linux, and Unix-like, still makes sense, to say "GNU/Linux (Unix-like)" which means for systems which use Linux but no GNU they are then just "Linux (Unix-like)" and systems with GNU but no Linux are "GNU (Unix-like).

3. From https://web.archive.org/web/20180615191615/https://link.springer.com/chapter/10.1007/978-1-4842-1392-6_3 "The first Linux distributions were created from scratch, having only the Linux kernel and a bunch of dispersed tools (like the GNU Project ones) with which to work."

False, the GNU Project was never a project to develop a series of arbitrary system tools. Make note, that even the Hurd kernel was also already in development (1990) by the time Linus first released Linux (1991). While the Hurd kernel didn't get finished and thus didn't get any official release as early, the Hurd kernel is still older than Linux as a project, GNU can hardly be called a bunch of dispersed tools as it is was obviously developing a complete operating system, and was well on the way towards getting there. What happened, and this is a historical fact, is the early distro developers took the GNU Project which sought to make an operating system, threw in some third party software and a kernel, and out comes a Unix-like operating system which has it's historic roots tied to the GNU Project. For me to say that isn't in violation of Wikipedia's neutral point of view policy, it is the accurate series of events in history as they played out. The historical roots tie the operating systems to the GNU Project irrevocably, and GNU should not be dismissed. I don't see why one should prefer such a source which obviously seeks to rewrite history, where that is not an accurate account of events. It was not a random series of coincidences for the developers of those early system distributions to just happen to run into a bunch of dispersed tools by coincidence and not recognize it was the GNU operating system, of which they formed the final complete system distribution out of.

4. I'll drop the Android issue Softwareperson1000 (talk) 10:42, 5 May 2026 (UTC)

See WP:VNT; saying that reliable sources are wrong based solely on your commentary and that this commentary should determine content is contrary to how the content of Wikipedia articles is determined. GNU/Linux is a minority term that is not the name of this article or the family of operating systems, as shown by reliable sources. - Aoidh (talk) 11:36, 5 May 2026 (UTC)

1. Your sources aren't reliable sources though, if they are obviously wrong in failing to recognize the key historic role the GNU Project played, which is a well recorded historical fact, or do you intent do dispute this? If the source depicts an interpretation of history which is proven false (such as asserting the Linux kernel was the founding point of the system distributions, as that one link Guy Harris pointed to did) when better record of history exist, such sources should be disqualified. There is also a question of politically motivated attempts at rewriting history in such bad sources, in removing mentions to the GNU Project as a controversial political entity many do not wish to have an affiliation to. Not wanting to affiliate with the politics of such an organization artificially warps the sources against the true prominence of the GNU project and it's relevance. So I question your sources on these grounds.

2. On another thing, in terms of pure numbers of how much popularity the "Linux" name has, I question the competency and familiarity random secondary source news paper website types would have on the technical nature of operating systems design, which is what the vast majority of support for the "Linux" name would inevitably end up being in terms of pure numbers, since the bigger popularity is the main argument in support of "Linux" after all. Being a secondary source, alone, is not a qualification to make one a reliable source. We are dealing with a technical matter which requires the opinion of people who understand the technical subject at hand, not the opinion of a news paper the only qualification of which is that they are a secondary source (which again in terms of pure numbers and popularity inevitably makes the most substantial portion of support). Softwareperson1000 (talk) 12:21, 5 May 2026 (UTC)

Reliable sources cannot be discarded simply because they do not support your perspective on a topic. - Aoidh (talk) 13:04, 5 May 2026 (UTC)

1. Fortunately, they are not reliable sources, so we can dismiss them.

2. Well recorded historical facts are not my perspective.

3. Do you dispute the historical facts about the role the GNU Project played, in bringing about the various operating system distributions, specifically, as a series of events, where the system distribution developers took the GNU Project software started by Richard Stallman, and then formed out of it complete system distributions, using some other third party software in combination with the GNU Project's software?

4. If you do acknowledge that as an accurate and a well recorded series of events, then how do you reach the conclusion that any source which fails to recollect this accurate series of well recorded historical events and even contradicts that historical record, is in any way a reliable source? Contradicting well recorded historical facts should disqualify any such source as being reliable in relation to that issue then, or do you disagree with this? Being a 'secondary source' does not give a nonsense bad source a free pass according to the WP:NPOV, that's not the purpose of that policy, and you have misunderstood it if that's how you use it. Softwareperson1000 (talk) 16:26, 5 May 2026 (UTC)

I do dispute the claim that "the system distribution developers took the GNU Project software started by Richard Stallman, and then formed out of it complete system distributions, using some other third party software in combination with the GNU Project's software." That's like saying that the architect who designed my house and the carpenters, etc. who built it share the credit with the wood from Georgia Pacific, the concrete from Quikrete, and the nails from Meite because they "formed out of them a complete dwelling." When you build something, the parts you buy and the tools you use don't get credit.

The GNU contribution to Linux was a bunch of small programs and a compiler -- parts and tools. GNU didn't create the small programs or the compiler. They copied and adapted the work of Bell Labs. What GNU really contributed -- the huge contribution that everyone has built on -- is the free-as-in-freedom license. That is the key creation that enabled everything else. The only reason most of the actual software made by GNU matters is because once someone makes a version of ls, mkdir, or pwd that work perfectly fine, nobody bothers reinventing the wheel.

The existence of a sawmill that performed a necessary step between a tree growing and the house I live in is also a historical fact. So is the steel mill and the cement factory. I can give you reliable sources for where nails and lumber come from. Everything is built on something else. That doesn't imply that the something else gets credit. --Guy Macon (talk) 22:09, 5 May 2026 (UTC)

1. You have now made an argument that is on the topic of who is given credit, but that is not the topic of discussion here. This topic is specifically on the technical designation of the 'OS family' entry on the Wikipedia page as well as the use of the "family" terminology used for the operating systems which use the Linux kernel and beyond. What name the system is generally referred to by, or, who is given credit is a separate matter entirely. Since your argument now seeks to dispute giving credit to GNU instead of addressing the actual family terminology and topic of this discussion in question, the GNU/Linux naming controversy and with it what the systems should be in general called and who is to be given credit being a separate issue, and with arguments on this thread not to be made one way or the other in support or in opposition of broader rewriting of articles such as to reflect one side or the other when referring to the system on that controversy (ie giving credit), I therefore, dismiss your argument on these grounds, and I don't want to hear more about the issue of giving credit.

2. However, I will still respond to your analog, the sawmill, cement factory and steel mill, and such things, are not a part of your house, are they? Such things were used throughout the process of building your house, but they weren't previous phases of, or previous iterations of, your house, and they certainly still aren't in your house nor a part of it. GNU software is still there (in your house) in the GNU operating system distributions today as a part of it, as well as being the historical and contemporary technical foundation and a framework for building the operating system, both. GNU therefore wasn't used to build an operating system and then discarded, it is the actual foundation of it on top of which the rest of the operating system currently exists. Also you can only push analogs so far. Here is another important counter argument within the other part of your analogs about the lumber used for building your house, which actually addresses the topic of OS families at hand this time, maybe the materials used in construction of your house, do, in fact, categorize your house. Maybe it's a house made of bricks, maybe some type of wood, perhaps concrete, or maybe your house is carved out of the stone on a cliff-side, categorization of your house in some metaphorical equivalent of 'OS family' in your analog, fits in perfectly. The "family" terminology escapes the analog perhaps, but as it turns out the house can be most certainly be categorized based on it's features and what it is built out of.

3. Lastly I'll mention that the part about no one wanting to replace the ls command and the like, because they can't be bothered to or whatever, is just entirely irrelevant and a total non-argument. Softwareperson1000 (talk) 12:00, 6 May 2026 (UTC)

That's like saying that the architect who designed my house and the carpenters, etc. who built it share the credit with the wood from Georgia Pacific, the concrete from Quikrete, and the nails from Meite because they "formed out of them a complete dwelling." When you build something, the parts you buy and the tools you use don't get credit. That would mean that neither the GNU project's software, the Linux kernel, or the other software would be credited. And, in fact, I usually see names such as "Red Hat" and "Fedora" and "Debian" and "Ubuntu" or..., not, for example, "Red Hat Enterprise Linux" or "Fedora Linux" or "Debian GNU/Linux" (Canonical don't even appear to speak of "Ubuntu Linux").

GNU didn't create the small programs or the compiler. They copied and adapted the work of Bell Labs. To be fair, what they copied and adapted were the external behavior of the Bell Labs programs and compiler; they reimplemented those from scratch. (It's not as if Stallman took the PCC source and made GCC from it.) They also significantly extended those behaviors. I.e., it's a reimplementation of Unix, not a new OS created from scratch. (And the extensions arguably are what make it a Unix - if you don't do at least one thing in your particular Unix that's not something other Unixes do, it's not really a Unix. :-))

Riiight. You started this thread with "my case is that GNU is a family of operating systems" but a response arguing that Linux is a family of operating systems and GNU is a collection of tools is "is just entirely irrelevant and a total non-argument". Not just something you disagree with, Not just something that is wrong. Entirely irrelevant and a total non-argument. Fine. I won't bother you with any further irrelevancies or non arguments. I will simply state that multiple reliable sources say that [A] Windows is a family of operating systems[3] (Windows 3.11, Windows 2000, Witndows 7...) [B] BSD is a family of operating systems[4] (FreeBSD, NetBST, OpenBSD. Ghost BSD...), and [C] Linux is a family of operating systems[5][6][7][8][9] (Linux Mint, Slackware Linux, Tiny Core Linux...). To be fair, a google search did find two non-reliable sources that say "GNU/Linux family of operating systems" - the FSF[10] and Wikipedia[11] and one that says "Linux/GNU family of operating systems" (a public comment to a proposed law).[12] I could not find a single source that says "GNU family of operating systems". --Guy Macon (talk) 14:34, 6 May 2026 (UTC)

Some early userland-for-Linux history:

• Linus spoke of having ported bash 1.08 and GCC 1.40 to the Linux kernel in his announcement of said kernel. He did not say anything in that post about a C library.
• Some of his early postings speak of a "really minimal" library with "partially free" sources (so perhaps some parts came from the GNU C library, but the non-free parts obviously didn't), with "Earl Chews estdio" being used.
• At some point, apparently, developers forked the GNU libc to make the "Linux libc"; eventually, they used standard GNU libc. Other C libraries, such as musl, have also been used.
• What appears to have been the first Linux distribution for people other than complete Linuxheads, MCC Interim Linux, shipped with GNU utilities as well as some non-GNU utilities. This document for MCC Interim Linux doesn't say what C library is used, although it does mention the GNU C++ library.

So, yes, software from the GNU project was used early in Linux development. The original C library wasn't from GNU, and I don't think there was a GNU init at that point (according to https://www.oldlinux.org/Linus/, the mid-October 1991 release didn't even have init, it just came up single-user as root). Networking tools may have come from BSD. Guy Harris (talk) 02:12, 6 May 2026 (UTC)

I will post here as a reply to yours Guy Harris, but with the intention of generally continuing the discussion without addressing any one specific editor individually, in fact I invite anyone to reply,, posting here, because what I will say ties directly into what has been said before in the above thread continuity.

1. As it is well recorded history, that the GNU Project was integral to the early distros and also what came after (note that Debian Manifesto reads "in the spirit of Linux and GNU", and Linus Torvalds earlier mentioned GNU in his announcement of Linux, therefore showing GNU wasn't seen as a mere set of tools they just accidentally stumbled on as some false and uninformed accounts would claim). [continues in the separate part 3. of my post] If GNU is supposedly only instrumental as a series of tools to develop a "Linux" family of operating systems, then how do you (you as in the supporters of that case) explain, how in your system the broader kernel offerings GNU systems make use of, fit in? See GNU variants here for context. Let's look at Debian GNU variants, there are more kernel offerings than one, of which FreeBSD and Linux had an official release, with Hurd still experimental but usable, and there are also several other more or less experimental and/or historical ports that didn't gain much success, where what each such option has in common is that each is obviously still Debian, and every time Debian uses the GNU userland, which together with Debian's package manager and it's overall design (as an Unix-like implementation) are what define Debian as an operating system. Debian is a part of the GNU family of operating systems because of it's historical lineage dating back to the founding of the GNU Project, with GNU also serving as a common unifying identity within the Debian ecosystem that ties all the different kernel branches together, the kernel being one interchangeable component. Debian thus, can not be a member of any "Linux" family, as proven by an official release of Debian using the FreeBSD kernel (note that being end-of-life is irrelevant, actively receiving updates is not required to be an operating system, many OS's never get updates). Question, how do you explain the presence of Debian GNU/kFreeBSD and still support a "Linux" family of operating systems at the same time, when there is such a glaring contradiction and/or an exception made where somehow different versions of Debian aren't in the same family, or either you would automatically insert Debian as a whole into the "Linux" family without giving further thought to the other kernels Debian also has?

Notice that Debian is currently, correctly designated as being "Unix-like" in the 'OS family' of it's Wikipedia article, which means this adds further support for my case in providing established precedent in favor of such categorization "Unix-like" for similar system distributions (with "Unix-like" being a compromise I said earlier I would agree to, to remove "Linux" as the family). The same should be done I argue, for other similar GNU derived systems which use Linux, even if they do provide alternative kernels to Linux, on the basis that the fundamental framework of those other similar operating system distributions is essentially the same, as is shown by some distros like Guix System and Gentoo with their introduction of Hurd kernel variants (Gentoo especially having broad Unix-like support with their package manager), and the Arch Hurd project, they all belong in the same family and Wikipedia's 'OS family' should reflect that. So I argue therefore that OpenSUSE or Mageia or Slackware as an examples of operating system distributions, are not that dissimilar from Debian, so therefore they should all belong in the same OS family on Wikipedia, which currently they are not assigned as such in 'OS family'. So in fact, there already does exist an inconsistency on Wikipedia as a malpractice of poor standardization and categorization, before I even started this argument, where applying my argument will fix this. This current state of things with it's current inconsistencies also further supports my case in showing the "Linux family" has failed, and does not in fact work because it causes such inconsistencies and has such a poor track record to show for it.

Furthermore, that inconsistent application of 'OS family' present in Wikipedia now, supports my other side-argument within this larger debate, that this is a nonsense made-up concept on Wikipedia to begin with, "OS family", which means my opponents can not pull the Wikipedia's policy of Neutral Point Of View against me on this issue in an attempt to evade my argument, as has been done several times on this thread so far, when this whole premise of "OS family" as it is de-facto applied on wikipedia is already a made-up concept which the original editor who presumably thought of it, would have already been the original violator of that WP:NPOV policy. Therefore such attempts to police the WP:NPOV against me now, would indicate seletive enforcement of that policy against those who object to the idea, but the original support of that concept of "Linux" as an 'OS family' is not subject so such policy enforcement. However, with that said, I do not object to there being 'OS family', what I object to, is selective policy enforcement and having inaccurate nonsense information on the 'OS family', which I seek to correct by making this case here.

2. With all this here, I say, my system works, and fixes exsiting problems. I have what is a consistent system where Debian as an example is always a member of the GNU operating system family, regardless of the specific kernel offering in question (or "Unix-like" instead of "GNU" as a compromise), Fedora and all the others belong in the same family, Alpine is just Unix-like, whereas your system (supporters of "Linux family) would have holes, gaps, be riddled with inconsistencies, and would require the likes of massive double standards when considering the broader scope of Unix-like operating systems and beyond, where it is not well established precedent to pay this much attention to the kernel specifically (you are treating Linux as a special snowflake, and with that making arguments which only make sense in the immediate proximity (if even that much) of Linux, while these operating systems in question are a part of a broader scope of the Unix-like sphere).

I argue you (supporters of "Linux family") do not have a holistic case in support of your claim, yours being myopic and Linux centered, but I have a full case which makes sense in a broader context and fixes existing problems. I argue the historical facts, as well as the established precedent in the Darwin and especially the BSD family of operating systems conventionally being seen as such, are against your case of "Linux" family, thus on the grounds as explained in this post, and including on the basis of the existing questionable state of the 'OS family' on Wikipedia, I declare, that such sources which contradict the well recorded historical facts regarding the GNU Project's relation to the system distributions using it's software and/or the well established precedent in categorizing operating systems families based on the historical lineage as with the BSD family, as unreliable sources which can not be used to support the case for a "Linux family" of operating systems.

3. GNU was not a means to an end, to make such an assertion that GNU was merely a series of miscellaneous tools that were used to build an operating system, is entirely arbitrary, and more importantly, it is based on personal motivations to say so, as in, to say from a certain point of view what the perceived motivation behind using the GNU software was, therefore that violates the WP:NPOV. Statements made from a biased point of view regarding the motivation, therefore are not based on actual historical facts or technical facts regarding the nature of the operating systems, such arguments can therefore be dismissed as being against the policy. GNU was it's own Project to form a complete operating system, where GNU Project already had it's own kernel in the works well before Linux, 1990 versus 1991. GNU having it's own kernel proves (one proof among others) it is not a set of tools, but a development effort to form a complete operating system. As such, it is then a fact to point out, that any third party who then forms and distributes a complete operating system derived from the use of such software from that Project, is therefore obviously doing just that, forming a complete operating system distribution based around the use of software from that Project, in other words, they are completing as a derivative work an operating system which the GNU Project sought to make using their software. There is the historical lineage, and thus the established precedent as with the BSD family, the GNU Project is the original Berkeley Unix equivalent in the comparison. The distro developers did swap out the kernel from GNU Project's one for another, had they not swapped that one component, there wouldn't be a case for any "Linux" family. Therefore the only support the "Linux" family idea has is that the supporters of it have to artificially over-inflate the importance of the fact that the distro developers use a different kernel as that component of the operating system. The kernel is a component of an Unix-like operating system which is interchangeable, as the Debian GNU/x variants show. The kernel is important to the function of an operating system alright, but it's not that important that it trumps everything else and to become the single deciding factor. Softwareperson1000 (talk) 14:43, 6 May 2026 (UTC)

correction: instead of neutral point of view having been pulled against me several times, WP:NPOV twice and WP:OR and WP:VNT have been suggested against me. Softwareperson1000 (talk) 15:26, 7 May 2026 (UTC)

The kernel is a component of an Unix-like operating system which is interchangeable, as the Debian GNU/x variants show. Changing the kernel results in some visible changes to the system. For example, the cgroups feature is a function of the Linux kernel, and won't be available for GNU/something-other-than-Linux unless it's reimplemented on the latter kernel, and Mach messaging is a feature of the GNU Hurd, and won't be available unless the GNU part is atop a Mach-based kernel such as XNU or the Mach messaging mechanism is reimplemented in the other kernel. There are other kernel-dependent mechanisms, such as the mechanisms for passing file descriptors from one process to another over Unix-domain sockets, packet-capture-and-injection mechanisms, kernel-dependent socket mechanisms, and so on.

I.e., it's not as if there are no changes in capabilities if you switch from one kernel to another. A lot of userland stuff won't notice, but there is userland code that's either dependent on one particular kernel's mechanism or that has platform-dependent code for handling different kernels (I'm a core developer and maintainer for one of those bits of userland code).

So there are things common to Linux-kernel-based systems, regardless of the userland, and not present in non-Linux-kernel-based systems, regardless of the userland. Guy Harris (talk) 05:56, 18 May 2026 (UTC)

Changing any part of the system as "interchangeable" does not have to stand for being a 1:1 drop in replacement as I use it here, in that of course changing any part of a system that deviates at all from some universal standard is necessarily going to introduce changes and breakage for other software that depends on those specific features, but this is true to any software and so not unique to kernels and so kernels don't deserve special credit for this. Any software specific features, at any part of the operating system, when another software depends on those features, is going to break things when that software being relied on is swapped for another one, this is inevitable, but is also not what I meant by "interchangeable". This is well explained using another example, the Unix shell is "interchangeable", there are lots of them, bash, ksh, fish, etc. however all such shells are not necessarily 1:1 feature parity with one another, yet they are an interchangeable component of an Unix-like system. The shell is an especially interchangeable component of an Unix-like system, where the kernel may be more difficult to "swap out", but it can be done for the kernel too as the GNU variants evidently prove. Linux specific features are in concept no different to how Bash extends the features as compared to a pure POSIX shell, both are arbitrary extensions or deviations, obviously when software relies on them things break, they are still interchangeable. Softwareperson1000 (talk) 17:37, 18 May 2026 (UTC)

Nevertheless, there are, as I've noted, things common to members of the set of XXX/Linux systems, just as there are things common to members of the set of GNU variants, and the set of things common to the set of XXX/Linux systems that are not in all GNU variants. That was the pointer I was trying to make.

That's why I do not dispute the existence of "Linux" as a family of systems. Guy Harris (talk) 21:39, 18 May 2026 (UTC)

OK. Is your position then that you also do see there exists a GNU family which overlaps with the Linux in some places? As with the Mac OS example you brought up earlier which has multiple entries in OS family. Softwareperson1000 (talk) 10:44, 19 May 2026 (UTC)

I agree with "the other Guy". My position is that the Linux family of operating systems does exist, and consists of all OSs that use the Linux kernel. The GNU family of operating systems also exists, and consists of all OSs that use the GNU Mach kernel.

BSD is more difficult. BSD does not have a stand-alone kernel but is developed as a whole operating system. That's why we have Debian GNU/kFreeBSD[13] and not Debian GNU/kBSD. --Guy Macon (talk) 16:17, 19 May 2026 (UTC)

Since you acknowledged GNU as a family of operating systems when it's own kernel is used, would you then agree that distros like Guix System and Debian which offer Hurd as as an option for the kernel, then need to be categorized under 'OS family' as including "GNU" there? Arch Hurd is currently formatted as "Unix-like (GNU/Hurd)" which is a bit odd, repeating the kernel is redundant as Hurd is already the kernel of GNU, so it's stating GNU twice where it's GNU userland then GNU's kernel but then in cases where the kernel is actually different the current state of things in Wikipedia is that then the kernel absorbs the entire system and becomes just "Linux" in case that kernel is used, so why this insane system on Wikipedia where the own kernel needs to be differentiated in such a context as if that makes any sense?

I will point out, that there is nothing difficult about the system I propose, simply find the oldest common point of origin as the historical root for any given system distribution, and have that as the OS family, with some edge cases where two systems are roughly evenly combined which makes it one / the other as convention. With a rule like that everything neatly falls into distinct categories, sure there may be some oddball systems because Unix-like systems just are like that, but my system is infinitely superior to a nonsense system riddled with double standards and exceptions as with "Linux", where no one calls XNU an OS family. Why have I not heard anyone ever refer to systems based on Darwin as being based around the use of the XNU kernel, or the XNU family of operating systems, Darwin is a complete operating system, so there is no special reason to pay attention to the kernel specifically, BSDs are complete systems, the only difference between them and GNU is that GNU was left incomplete at the time the forks started appearing, meanwhile those other systems were released as completed by the time any other systems based on them rolled around, but GNU was definitely a project to create a complete operating system, it just happened that it was other people who finished the task before the GNU Project themselves.

On the BSDs, FreeBSD, OpenBSD, and NetBSD are their own independent operating systems, they are a part of the BSD family of operating systems. There could be a Debian GNU/BSD in case such a variant used the kernel from the original Berkeley distribution before it split into the different projects. Just like how the original BSD existed and got split into the other BSD systems which form the family, even if GNU was incomplete it was a development effort to build an operating system, then it too was split into different independent systems. However their level of independence is actually weaker than in the BSD world, where GNU/Linux systems still use the same software from GNU and haven't diverged into their own projects like the BSD ones have, which actually means GNU/Linux has an even stronger case for having to be it's own designated group of operating systems, as either "GNU/Linux" or just "GNU" where the BSD systems have less common with each other than the GNU/Linux ones do. The only thing tethering the BSD systems together other than some code spreading from one project to another, is that historical root. So if the BSDs as different systems which don't share a common development project as the base on top of which they build separate implementations of that system as with Linux and GNU but are still a part of the same family, then how come GNU as not only the historical root and point of origin in the timeline, but also the still current development base, alongside Linux with GNU/Linux ones, isn't a valid base for an "OS family" for all such systems derived from it then? Softwareperson1000 (talk) 09:50, 31 May 2026 (UTC)

The question is who decides what an ancestor is. If someone keeps the kernel but changes literally everything else, is it still a descendant? How about if they change just the kernel and keep everything else the same?

I made up a special rule just for BSD. If it comes from Berkley, who hold the UNIX license, I call it the BSD family simply because saying NetBSD and OpenBSD are different OS families each with one member is silly. Maybe they are all part of the UNIX family of operating systems? --Guy Macon (talk) 13:03, 31 May 2026 (UTC)

If a currently "GNU/Linux" distribution removes all GNU software, when it removes the last remaining part of the GNU heritage I would say it is not a GNU anymore. This may be the case with Ubuntu in the future, (possibly depending on whether it actually happens) where Ubuntu might remove all the GNU software from their system which they have started doing. In that case Ubuntu will then be "Unix-like" just like Alpine is. So append the oldest living ancestor so to speak to my criteria then, which I stated earlier about finding that ancestor. Ultimately such a categorization of "OS family" is always entirely arbitrary, but my case is that "GNU" or "GNU/Linux" or even "Unix-like" simply has a stronger claim to be the family rather than "Linux". In other words "Linux" has the smallest claim to the systems is my argument at it's core.

Unix > GNU > Debian GNU/Linux. Unix and Unix-like should be used interchangeably for an OS family because the only difference is trademark status

Unix > BSD > OpenBSD as another example

GNU/Linux a taxonomic subcategory of GNU, which means every GNU/Linux system is a GNU system, the default for a GNU is in terms of it's kernel the Hurd, so it doesn't need to be specified so unless otherwise noted GNU alone is GNU with Hurd. Each GNU system is also a part of the Unix-like family of operating systems which also include the BSDs, Darwin, all the proprietary ones, and others.

For a system that would be a full GNU with it's kernel too, that removed all userland GNU software but retained the Hurd, that would still be a GNU system, because it obviously still retains GNU in there, only once everything is removed the oldest living ancestor then dies I would say. Softwareperson1000 (talk) 13:37, 31 May 2026 (UTC)

See Talk:Linux#The proper name is Apache/Busybox/GNOME/GNU/Java/KDE/LILO/LLVM/Mozilla/Perl/Slackware/TeX/Minix/Xorg/Xfree386/Linux. Your logic is flawed. You start with the false assumption that Linux should be called GNU/Linux, and from that assumption you draw conclusions about the family it should be classified in.

Again I tell you, when you create a free and open source project, you get to name it. It doesn't matter what you based it on or what other software you include. Common sense, Wikipedia consensus and international trademark law agree that you and you alone get to name it. The ONLY "GNU/Linux distributions" are those that voluntarily choose to add "GNU" to the name. Their choice. Not yours, not Stallman's. --Guy Macon (talk) 18:30, 31 May 2026 (UTC)

"Again I tell you, when you create a free and open source project, you get to name it."

I see we have an agreement then, Richard Stallman created the GNU Project and he gets to name it. GNU it is Softwareperson1000 (talk) 12:36, 1 June 2026 (UTC)

Yes, GNU is called GNU. That doesn't mean distributions have to include "GNU" in their names. Skynxnex (talk) 12:58, 1 June 2026 (UTC)

but what is it a distribution of? I am not convinced most people participating in related discussions on this sort of topic even understand what the word means. A distribution, is not a type of operating system, software distribution in case of GNU software means, that this thing right here, is a distribution of GNU Softwareperson1000 (talk) 13:03, 1 June 2026 (UTC)

• "Unix is a family of multitasking, multi-user computer operating systems that derive from the original AT&T Unix, the development of which started in 1969 at the Bell Labs research center by Ken Thompson, Dennis Ritchie, and others."
• "Research Unix [Also known as Version 2 UNIX] refers to the early versions of the Unix operating system ... developed in the Bell Labs Computing Sciences Research Center. The term Research Unix first appeared in the Bell System Technical Journal (1978)" --Research Unix
• "The origin of C is closely tied to the development of the Unix operating system, originally implemented in assembly language on a PDP-7 by Dennis Ritchie and Ken Thompson, ... The C compiler and some utilities made with it were included in Version 2 Unix [1978], which is also known as Research Unix." --C (programming language)

AT&T owned the trademark for Unix until it sold it to Novell in 1992. Novel transferred it to The Open Group in 1993. Nobody other than the Open group gets to decide what Unix is called, so please stop trying to rename it to Multics/Unix. That's for the Open Group to decide. They own the name.

• "Development of the GNU software was initiated by Richard Stallman while he worked at the MIT Artificial Intelligence Laboratory. It was called the GNU Project, and was publicly announced on September 27, 1983" --GNU
• "The GNU Compiler Collection ... was first released in 1987 by Richard Stallman, GCC 1.0 was named the GNU C Compiler since it only handled the C programming language." --GNU Compiler Collection

Richard Stallman (or perhaps he has transferred ownership to the GNU project?) owns the trademark for GNU. Nobody other than Stallman gets to decide what GNU is called, so please stop trying to rename it Multics/Unix/GNU. That's for Stallman to decide. He owns the name. He also owns GCC, so please stop trying to rename it to K&R C/GCC That's for Stallman to decide. He owns the name.

• "Linux is a family of free-and-open-source Unix-like operating systems based on the Linux kernel, which was first released on 17 September 1991 by Linus Torvalds." --Linux

Linus Torvalds owns the trademark for Linux. Nobody other than Torvalds gets to decide what GNU is called, so please stop trying to rename it Multics/Unix/GNU/Linux. That's for Torvalds to decide. He owns the name.

• "Slackware is a Linux distribution created by Patrick Volkerding in '1993." --Slackware

Patrick Volkerding owns the trademark for Slackware. Nobody other than Volkerding gets to decide what Slackware is called. (He decided to call it "Slackware Linux" but everyone else calls it "Slackware"), so please stop trying to rename it Multics/Unix/GNU/Linux/Slackware. That's for Volkerding to decide. He owns the name.

• "Android is an operating system owned by Google which is based on a modified version of the Linux kernel and other free and open-source software, [none of it from GNU] designed primarily for touchscreen-based mobile devices such as smartphones and tablet computers. Android has historically been developed by a consortium of developers known as the Open Handset Alliance, but its most widely used version is primarily developed by Google. First released in 2008" --Android (operating system)

Google owns the trademark for Android. Nobody other than Google gets to decide what Android is called, so please stop trying to rename it Multics/Unix/GNU/Linux/Android. That's for Google to decide. They own the name. And they have specificly rejected "Andoid Linux" as a name.

What's that I hear? You (Softwareperson1000) don't want to rename Slackware to Multics/Unix/GNU/Linux/Slackware? It's only Linus Torvalds that you want to seize ownership of his trademark from? That's an ... interesting ... decision. --Guy Macon (talk) 14:36, 1 June 2026 (UTC)

What on earth did you just post? What has Multics got to do with this? Also who said this discussion thread is supposed to be about renaming things in the first place? You brought up the convention that the one who starts a thing gets to name it just recently, and to that I humorously remarked with sarcasm that that RMS started the GNU Project and as I've already established earlier that the complete system distributions which are the descendants of GNU, such systems ought to therefore be rightfully called so, accepting this line premise of using that convention you just brought up, however that is actually my position, the sarcasm was the part that we agreed. I only responded to what you just said within the scope of you first mentioning that convention of naming things, prior to that I already told you this thread is not about giving credit Talk:Linux#c-Softwareperson1000-20260506120000-Guy Macon-20260505220900 and it is also not tied to the naming of things, I have explicitly made it clear the GNU/Linux naming controversy in terms of what the system is generally referred to as ie what it's called or what it's name is, is not the issue here. I have humored this topic enough, stop talking about the name. This discussion is about the specific classification in terms of a designated "OS family" entry or the use of "family of operating systems" and related terminology, which is not the same as what the systems should in general be referred to as. I do have opinions and viewpoints also on those other issues, and we can discuss them separately elsewhere, but please keep it out of this discussion you are not contributing anything useful to it and I will dismiss such statements by you. Softwareperson1000 (talk) 17:01, 1 June 2026 (UTC)

Please don't write sentences such as "GNU/Linux a taxonomic subcategory of GNU, which means every GNU/Linux system is a GNU system" If you don't want responses talking about "GNU/Linux". GNU/Linux does not exist. It isn't an operating system, a distribution, a family of anything, or anything else. GNU/Linux will only exist if Richard Stallman and Linus Torvalds (the owners of GNU and Linux) ever agree that it exists.

Also, please don't write things like "I've already established earlier that the complete system distributions which are the descendants of GNU, such systems ought to therefore be rightfully called so" if you don't want replies saying that you have established no such thing and have failed to explain why we shouldn't tack Multics on to the name of anything that is a descendant of Multics. You keep applying your descendant rule in one place and one place only and you can't explain why you don't apply the same rule anywhere else.

In my experience, the best way to stop talking about something is stop talking about it. Demanding that other people not talk about it so you can get in the last word is far less effective. I'm just saying. --Guy Macon (talk) 17:36, 1 June 2026 (UTC)

I am not concerned with your life advice given in the last paragraph, this is a specific discussion thread taking place in the Talk page of a technical software project that focuses on a specific established topic, if you are unwilling to keep the conversation and your comments on-topic then you need to go somewhere else.

If you want to talk about Multics, then talk about Multics, that's fine, I have no issue with you bringing it up in relation to families of operating systems as per the lineage and descendants, so you saying "You keep applying your descendant rule in one place and one place only and you can't explain why you don't apply the same rule anywhere else." is entirely unfounded, I am willing to discuss Multics, what I am not willing to discuss is renaming operating systems, on this thread Softwareperson1000 (talk) 17:52, 1 June 2026 (UTC)

New question to supporters of Linux as a family. If Linux as a kernel qualifies as the base around which the "family" is defined against, where is the Mach microkernel family of operating systems? Mach was developed at Carnegie Mellon University as a standalone OS kernel project following the microkernel architecture just like Linux was, not as a part of any specific holistic operating system project. How Mach differs from Linux is that it only implements the core of the microkernel design and not the rest of the kernel, where it's kind of like the core of the kernel of the operating system, where the OS kernel is itself the core of the OS, and Mach is microkernel based where Linux is monolithic. If Mach as the deepest most core component of an OS kernel qualifies as the foundation on which later complete operating systems and their implementations of complete kernels based around Mach, qualify as being in the same sort of historic category that Linux is under which grants Linux the claim to the family of operating systems, then we should also have the Mach family of operating systems which encompasses Darwin and it's derivatives as well as GNU Hurd based operating systems. Is there a reason not to classify Mach similarly to Linux as the base and the foundation on which the rest of the operating systems were built, and thus has the strongest claim to the operating system family, if Linux has that? Softwareperson1000 (talk) 10:21, 19 June 2026 (UTC)

Mach was developed at Carnegie Mellon University as a standalone OS kernel project following the microkernel architecture just like Linux was, not as a part of any specific holistic operating system project. Presumably "just like Linux was" means "just like Linux was developed as a standalone monolithic kernel, not as part of any specific holistic operating system project".

we should also have the Mach family of operating systems which encompasses Darwin and it's derivatives as well as GNU Hurd based operating systems In the sense that it's a family of OSes that support Mach messaging, a case could be made that all OSes that support Mach messaging (as an API exposed to developers) are a family. Guy Harris (talk) 10:46, 19 June 2026 (UTC)

Yes I meant by the part on just like Linux was that they were both standalone projects, I perhaps worded that poorly but that's what I was trying to say.

Darwin's XNU and GNU's Hurd, both actually take the Mach and build the rest of the kernel around it. It's not a separate different project that is merely compatible with Mach, that they both made, they actually use the Mach. Not necessarily Mach in the exact original form, where both would have forked it into their own projects such as GNU Mach being distinct from the original Mach as developed Carnegie Mellon University but it's still essentially the same thing. Both XNU and Hurd actually use Mach at the core. Softwareperson1000 (talk) 10:56, 19 June 2026 (UTC)

Yes, they do, albeit somewhat differently:

   $ fs gnu/hurd/gnumach | xargs wc -l | egrep 'total$'
     499731 total
   $ fs macos/xnu | xargs wc -l | egrep 'total$'
    2382167 total

(fs is a script wrapped around find that finds source files). The kernel-mode code is about a factor of 4-5 larger for XNU (and that doesn't include all the kernel-mode loadable modules that implement various device drivers, file servers, etc.. Much of that is done in userland in the Hurd:

   $ fs gnu/hurd/hurd | xargs wc -l | egrep 'total$'
     346153 total

But both of them support Mach messaging as an IPC mechanism that developers can use, so they're both members of the Mach family of OSes. Guy Harris (talk) 11:20, 19 June 2026 (UTC)

So if you say they are indeed a part of such a family, should the Mach article reflect that and speak of the "family" of such operating systems, as well as articles on Darwin, GNU, and Mac OS etc. be rewritten to refer to them as a part of the Mach family? And the "OS family" reflect that also, where currently eg MacOS does not mention Mach as any "family". Softwareperson1000 (talk) 11:27, 19 June 2026 (UTC)

Yes, that might make sense. An OS can, of course, belong to more than one family, e.g. a GNU/Linux distribution would belong to the family of Unix-like systems, as well as belonging to the Linux family and the family of GNU variants, and macOS would belong to the family of Unix-like systems (and the subfamily of Unix systems, as it's passed the Single UNIX Specification test suite), as well as the Mach family. Guy Harris (talk) 11:35, 19 June 2026 (UTC)

I'd say "the family of GNU variants" really is just the GNU family of operating systems where each of the variants are essentially the members of that family, GNU. I have no issue with there being a family entry defined around the use of any particular kernel present, alongside other entries like how you describe. My problem is that I don't see the kernel alone as being sufficient to describe the family around, and in particular Linux but Mach too I would say. So do you and I have an agreement then on this topic that there should be multiple entries, and that for those systems which are GNU and Linux they should include "GNU" or variations as one of the entries? Softwareperson1000 (talk) 11:49, 19 June 2026 (UTC)

References