Talk:Unix-like/Archive 1

Archive 1

Archive 2

Archive 3

misrepresentation of GNU and BSD

I've fixed a lot of mistakes about GNU, but I think some remain about BSD. Are the BSDs really "non-commercial "Unix-like" operating systems ... developed to serve as inexpensive or free substitutes for Unix."? ...is that really what their developers wanted them to serve?

I've fixed the timeline of the "Development of Unix-like systems" section. It presents information in what seems is meant to be a mostly chronological order, so I've added GNU to the first paragraph where systems developed between 1978 and 1985 are discussed. I've also fixed the very mistaken classification of GNU as being developed for the purpose of being inexpensive.

For mentions of "Linux, GNU, and Minix", I've replaced them with "GNU/Linux and Minix". The old wording would confuse many people into thinking that Linux and GNU are simply two operating systems. That can be one interpretation of the terms, but the issue is not as simple as that. The actual relationship between them is explained in the article, and I've noted that GNU+Linux systems are often called "Linux". Using the term "Linux", for GNU systems which use the Linux kernel, would be completely misleading in this article since the project to develop an OS that would be Unix-like started 8 years before Linux.

Anyway, my main question is: can someone confirm/fix the BSD references? Thanks. Gronky 02:35, 10 November 2005 (UTC)

For the BSD thing, it's a little complicated -- back in the day BSD was what all the cool kids used, instead of stodgy System V. These days, it's all about the low-cost thing. As for the GNU/Linux distinction, may I suggest we change it to "GNU Hurd, Linux, and Minix"? As has been pointed out, the GNU/Linux thing is a minority opinion (viewed by many, including myself, as a Stallman power grab), and Hurd and Linux are very different OSes that happen to share similar userlands. Like I said, Linux is not part of the GNU project, never has been, and a Linux distro can be implemented using very little GNU software. Haikupoet 06:01, 10 November 2005 (UTC)

Over 8 years (before Linux appeared), the FSF staff and the GNU volunteers wrote the substantial part of an operating system. The GNU project was the only project that was working towards making an operating system available and is therefore the only reason that an operating system came into existence. This makes "GNU" or "GNU/Linux" a valid name. Many people call the operating system they produced "Linux". This makes "Linux" a valid name. But if this article calls GNU+Linux operating systems "Linux", then it has to say "Richard Stallman started Linux in 1983". Which would be very silly.

No, it does not have to say any such thing; your premise that naming is viral is flawed, as demonstrated by the absurd conclusion it reaches. Furthermore, Wikipedia policy requires us to avoid saying thing that are deliberately confusing and contrary to popular usage of terminology, regardless of whether you can torture a semantic argument into supporting them. Tverbeek 12:44, 10 November 2005 (UTC)

You've given nothing to support your first sentence, so I can't respond. And the Wikipedia policy you cite about not using confusing names is exactly what I cite when I say that you can't use the name "Linux" for an operating system that was started 8 years before Torvalds even thought of writing his kernel. Can you say what you are suggesting is done, and offer something to back it up? Gronky 15:10, 10 November 2005 (UTC)

You most certainly can use the name "Linux" for an OS using software written before Torvalds got started on his kernel; countless people do it every day. Names are not genetic (the flawed premise of your argument); they are decided by the people who use them. The grapefruit is in no way related to the grape, but that's what people call it, so that's what its correct name is. Insisting that we call it a "big yellow orange" or a "Citrus paradisi" - contrary to common usage - is what would lead to confusion. The only possible "confusion" caused by the use of "Linux" instead of "GNU" or "GNU/Linux" is who should get the credit for it. Which is unfortunate (I happen to agree that the FSF and RMS should get more) but not relevant to Wikipedia, which prohibits the use of articles to promote a cause. If you don't like that policy, go fork your own encyclopedia. Tverbeek 16:32, 10 November 2005 (UTC)

There are no GNU-less Linux-based operating system distros (maybe there are some tech toys, academic projects, or bootable utilities - but there are no fully funciontal OSs), but there are Linux-less GNU-based operating system distros (see the article). At some point in the future, there may be replacements for GNU packages but those replacements won't have been the reason that an OS came into existence. Someone can write a new libc, but this is not a significant contribution since it would simply be redundant. There is already a working libc. Linux was a significant contribution, since there was no working, free software, Unix-style kernel.

On BSD being all about the low cost thing: I'm surprised. This contradicts the little I know about the BSDs, but still, I'm not very involved in BSDs. Gronky 10:50, 10 November 2005 (UTC)

Calling them Linux and GNU Hurd would be totally wrong. The GNU Hurd is a unix-kernel replacement, Linux is a unix kernel, and Minix is an operating system. The GNU operating system with the Linux kernel - GNU/Linux - is often referred to as just Linux, so you could say Linux and Minix are operating systems. If you do this, GNU Hurd doesn't fit in anymore, because it is not an operating system. As for the "Stallman power grab": as he said it himself, "When I do this, some people think it's because I want my ego to be fed. Of course, it's not like I'm asking you to call it Stallmanix." Have a nice day, I wish you good luck trying to run Linux without GNU! Geronimooo 15:42, 10 November 2005 (UTC)

edits by Tverbeek

Tverbeek's edit makes Linux look as a derivative from GNU, which it is not. Only one part (though important) is replaced - the Hurd by Linux Geronimooo 17:36, 10 November 2005 (UTC)

Also, it lists GNU and Linux as different Unix-like operating systems on the list of current examples, and removed GNU/kFreeBSD and GNU/NetBSD! Geronimooo 17:38, 10 November 2005 (UTC)

Linux is not GNU

It is true that the bulk of most Linux systems is made up of GNU software, but the core of the system is not and has never been part of project GNU, and indeed apart from GCC it is possible to build an entire Linux system using non-GNU software. (You'd be giving up a lot of functionality doing that, mind you, but there are those who would argue that you're also removing massive amounts of bloat from the equation, a flamewar for another time.) Therefore it is incorrect to include Linux as part of GNU. Haikupoet 02:03, 9 November 2005 (UTC)

It's not incorrect, but as Tverbeek said it indeed does not reflect popular usage, which Wikipedia adheres to. ¦ Reisio 03:22, 9 November 2005 (UTC)

What is incorrect is to say that Linux or GNU/Linux is descended from something where the "kernel is the GNU Hurd plus a micro-kernel". Ditto for the GNU/*BSDs. -- Karnesky 19:50, 20 November 2005 (UTC)

Why not List of Linux distributions?

List of Linux distributions is apparently appropriate for List of open source software packages (where I used it to replace a growing list of distributions), and on GNU. Different distributions are:

1. Run by different people
2. Can use different variations of the kernel
3. Can have different userland apps

if all of the *BSDs are listed, I see no reason not to link to various Linux distributions. -- Karnesky 23:28, 20 November 2005 (UTC)

Linux or GNU/Linux is an operating system. Linux distributions are just that - distributions of essentially the same operating system. To argue that BSDs are the same as Linux distros doesn't quite make sense, as Linux distributions are typically just the Linux kernel (or one of a handful of popular modified versions) and a handful of supporting software that is approximately the same regardless of distro. BSDs, on the other hand, do not all use the same kernel. I realize that when you get right down to it they're all the same - Unix clones living off mostly the same pool of Unix software - but there are very few BSDs compared to a staggering amount of Linux distros, partially for the reasons I have already explained.

In addition to it not making all that much sense to list distros as separate operating systems, the Linux distribution article is linked from Linux, GNU and here. ¦ Reisio 04:27, 21 November 2005 (UTC)

O.K. Your placement of it to the See also section is fine. -- Karnesky 05:05, 21 November 2005 (UTC)

such as Linux, GNU or Minix

Really, right now this makes no sense. "Functional Unix or Unix-like -- Broadly, any Unix-like system that behaves in a manner roughly consistent with the Unix specification; more specifically, can be used to refer to systems such as Linux, GNU, or Minix that behave similarly to a Unix system but have no genetic or trademark connection to the AT&T code base."

This is factually incorrect, because it assumes that "Linux" (GNU with the Linux kernel) and GNU (with the Hurd in the place of the kernel) are different operating systems. I won't change it to just GNU anymore, because tverbeek doesn't like it, but something needs to be done about this. Geronimooo 20:42, 9 November 2005 (UTC)

You are right that if one word were required to describe them both, it should be "GNU", but they _are_ different operating systems. ¦ Reisio 22:20, 9 November 2005 (UTC)

The assertion that systems built on entirely different kernels should be considered the same operating system because they share libraries and userland tools is not only an opinion (rather than a fact), but a minority one. In fact, I've never heard it asserted in any circumstance except when someone is trying to subsume Linux systems into GNU. The list of "Linux, GNU, and Minix" makes perfect sense to the vast majority of readers. (Even to those who would rather have Linux subsumed into GNU understand it perfectly well; they merely don't like it.) "Something needs to be done about this." Why? Tverbeek 02:14, 10 November 2005 (UTC)

So Debian GNU/Linux, Debian GNU/Hurd, Debian GNU/kFreeBSD and Debian GNU/NetBSD are different operating systems?

Yes. They're very similar but the fact that they use different kernels warrants mentioning them seperately I think. If you don't wish to mention them seperately then you should just call them GNU. ie. Not mention the work `Linux' in the page, which would be weird. Alternatively you _could_ call the systems by thier respective kernels, however I think the FreeBSD People wouldn't be happy with GNU/kFreeBSD being called FreeBSD!

Now the articles of the different operating systems, which are variants of the GNU operating system, are merged in GNU variants. -- mms 17:29, 8 January 2007 (UTC)

GNU vs GNU/Hurd

User:Reisio removed GNU/Hurd from the list of GNU-based systems, saying that "'GNU/Hurd' is just 'GNU.'" Yes, they lead to the same article. But I think it helps to differentiate them in the list. If "GNU" refers to the GNU userland (or operating systems which use it, in addition to a F/OSS kernel), then GNU Hurd should be listed (as being on the same level as GNU/Linux and GNU/*BSD). If "GNU" refers to "GNU/Hurd," then GNU/*BSD and GNU/Linux should not be listed under GNU. In short, please pick a way to use "GNU" & be consistent. -- Karnesky 22:34, 20 November 2005 (UTC)

GNU Hurd is just the Hurd. GNU is an operating system which uses the Hurd. This system is also called GNU/Hurd sometimes. GNU/*BSD and GNU/Linux are GNU variants. -- mms 17:24, 8 January 2007 (UTC)

I agree that GNU/HURD is the GNU Operating System. But GNU is an ambiguous term, used for the operating system, the "GNU project" (or the organization responsible for it), the GNU licenses, or other projects. Because of this ambiguity, GNU is not commonly used alone to describe the OS & an encyclopedia is no place to change it. Either refer to it as "GNU/Hurd" or, at least, explicitly as "the GNU Operating System." --Karnesky 21:38, 8 January 2007 (UTC)

agreed. -- mms 00:37, 9 January 2007 (UTC)

Hurd's not GNU

Well yes, Karnesky, Hurd's not GNU. Hurd is correctly named GNU Hurd and it is a kernel. GNU is the name of the operating system developed by the GNU Project. Even though the Hurd did not yet exist there were GNU variants released. An operating system consists of more than a kernel. It is true that most people call GNU/Linux Linux --- but still this is not correct. I just read about that some GNU/*BSD variant is »Linux based« which is ridiculous. This article is about an advanced topic and we should apply the appropriate language for it. Debian GNU/NetBSD is the name of the only GNU/kNetBSD system »existing«. It is sad they didn't take the »k« in the name but Wikipedia can do nothing about it. The GNU/*BSD are obviously derived from GNU. Please revert to my last version or allow me to do so. -- mms 20:52, 8 January 2007 (UTC)

I'm sorry for some confusion--I shortened my editing summary for brevity. I guess that should have been a hint to post here first.

I meant that "GNU" is not unambiguously the same as "GNU/HURD." Even the FSF and gnu.org call it "GNU Operating System" or "GNU System" in preference to just "GNU."

I don't see a reason to debate the GNU/Linux naming controversy here--WP has a fairly consistent style. Why break from that style here?

I don't know what GNU/*BSD variant you are referring to.

GNU/*BSD has parentage in both BSD and GNU. Hence their names, and hence their appearance at multiple places in the list.

In short, I think all of your changes are contentious & should be discussed before revert. --Karnesky 21:33, 8 January 2007 (UTC)

Your revert also renamed Debian GNU/NetBSD to Debian GNU/kNetBSD and GNU/Solaris to GNU/OpenSolaris. Wikipedia shouldn't invent new names. Have a look at http://www.debian.org/ports/netbsd/ and http://www.gnusolaris.org. In this context we should really call the Linux based GNU operating system GNU/Linux because it is consistent. After all we do link to the naming controversy. The GNU/*BSDs have the kernel of some BSD but I thought we would list every operating system only once. If I choose between the GNU/*BSDs beeing a variant of GNU/Hurd or of BSD, I choose it is mainly a GNU variant. -- mms 00:53, 9 January 2007 (UTC)

Sounds good. I'm fine with your partial revert. (Do observe the title on Nexenta's page, though: "Unix Portal:Nexenta OS - Nexenta GNU/OpenSolaris" & note that the 'k' in 'Debian GNU/K*BSD" stands for "kernel of" & has been used by Debian developers to describe the system (including in the initial project announcement. I'm fine with your changes--I suppose they're more popular). I'm fine with listed the GNU/*BSDs twice, but would suggest putting them in a separate section if they were only to be listed once. --Karnesky 01:21, 9 January 2007 (UTC)

Taxonomy

The indented list indicates some form of official taxonomy. In the interests of trying not to fan the flames on the GNU/Linux debate I'm flattening this. Chris Cunningham 09:19, 10 January 2007 (UTC)

Well as I explained elsewhere, you can't avoid to take a POV. Wikipedia should reflect every POV to become NPOV. But in things like a list it is hard not to make a statement. Whether we refer to GNU/Linux as Linux, a self-contained operating system which happens to use some GNU software, or we acknowledge that GNU/Linux is a GNU variant which uses Linux, a kernel. So I suggest to revert your changes in Unix-like#Current examples. -- mms 10:06, 10 January 2007 (UTC)

You don't understand WP:NPOV, but regardless of that you haven't presented an argument in favour of a revert (other than the logically fallacious "by not presenting my point of view you're presenting your point of view"). No official taxonomy of late (especially free) Unixen exists. The information should not be presented in a misleading form. Chris Cunningham 11:17, 10 January 2007 (UTC)

»Everybody has a point of view. Though 99% of the world may see something exactly the way you do, still your view is just one of many possible views that might be reasonably held.« WP:NPOVT

There is an official taxonomy if you ask the projects on which system theirs is based on. I think there were only 3 questionable classifications in my last version:

1. Debian GNU/kFreeBSD and Debian GNU/NetBSD are not descendants of 386BSD as the are GNU variants
2. GNU/Hurd is a new name invented by the Debian developers for GNU
3. GNU/Linux is considered by some (or most) to be Linux, an operating system initiated by Linus Torvalds which happens to use some GNU software

-- mms 11:44, 10 January 2007 (UTC)

That isn't an "official taxonomy". Every variant using the GNU stack is questionable on the grounds you've given. Most normal people classify OSes based on the kernel, not on the toolchain. Rather than doing that (which is heresy to the GNU people) I simply eliminated the hierarchy. This is the least contentious way of presenting the information. Chris Cunningham 12:15, 10 January 2007 (UTC)

I thought it would provide useful information if we grouped the systems. But if we can't agree on an "official taxonomy" we should also sort the entries alphabetically. The main reason for me to think of GNU variants as of GNU variants is not a technical consideration but the history of computing. First a program was considered to be a piece of source code. It was shared, sold and modified. Even without licenses. Then there came proprietary software to the world. RMS founded the GNU Project and the GNU Project developed the first free operating system after there was proprietary software invented. When Linux came out, it was clear that it was a kernel which was used by GNU — making it a variant of GNU as Linux is not part of the GNU Project. Even if a few years have passed since then this is a historical fact and I don't want Wikipedia to support widespread misbeliefs held as truths. -- mms 12:58, 10 January 2007 (UTC)

That mantra has quickly gotten tiresome, and I'm sick of hearing it from the same three or four people. Alphabetical order is fine. Chris Cunningham 13:02, 10 January 2007 (UTC)

While I disagree with mmm on particular points of the taxonomy (especially that toolchain is the one and only thing we should trace), I do agree with him that the list should be classified. The list contains less information now, there are enough entries that an alphabetical list is unwieldy for an article, and the article still has the potentially contentious Unix-history.svg. If enough people agree with Chris that there is no way to make a hierarchical list, my vote is to remove the list completely & link to a list or comparison article or a category.

I believe it is possible to clearly define the taxonomy (either through direct parentage of the kernel or toolchain (as it was before), or of only one of these). --Karnesky 13:46, 10 January 2007 (UTC)

Isn't MS Windows becoming increasingly Unix like ?

Over many years I have seen Microsoft add features to their main consumer operating systems products, MS-DOS and Windows which were previously present in Unix to the extent the differences are much less than previously. These features include:

• Windowing based on the X desktop model.
• Inclusion of BSD TCP/IP stack within the Windows kernel
• Inclusion of BSD command-line networking utilities.
• Pre-emptive multitasking with protected memory.
• Multiuser authentication and user-based privileges, and ability to restrict certain privileges to administrator.
• Support for Kerberos network authentication and authorisation.
• Support for Unix-style pathnames in respect of remote object access using HTTP

While differences remain - in particular concerning developer/user access to source code and local pathnaming - in a sense these differences are dwarfed by the growing similarities. The microkernel Windows NT/XP architecture is present on other Unix-like systems. I therefore think it could be useful for this article to include a comparison showing the extent to which Windows has become derivative of Unix and consequently is Unix-like. Copsewood 11:20, 24 January 2007 (UTC)

This smacks of original research. If others have made this claim, it's worth referencing. For what it's worth I agree that NT made specific steps towards POSIX in a good number of areas, though this isn't the same as becoming a "Unix derivative". Chris Cunningham 11:38, 24 January 2007 (UTC)

Adding "pre-emptive multitasking with protected memory" and "multuser authentication and user-based privileges, and ability to restrict certain privileges to administrator", hardly makes an OS qualify as "Unix-like", as both of those characteristics pre-date Unix (yes, as in "existed in some OSes before 1971" - see, for example, Multics). Using Kerberos isn't sufficient to make something "Unix-like", either, nor is the inclusion of the BSD TCP/IP stack (Network Appliances Data ONTAP operating system's TCP/IP stack is BSD-based, but it's hardly "Unix-like" - it doesn't have pre-emptive multitasking, it has cooperative multitasking, and all processes run in kernel mode in the same address space), and if "support for Unix-style pathnames in respect of remote object access using HTTP" makes an OS "Unix-like", then any OS that happens to be a Web browser would become "Unix-like", which would render "Unix-like" not a very meaningful adjective.

Frankly, the one characteristic of recent versions of Windows that might render it "Unix-like" in some meaningful sense would be the Subsystem for Unix-based Applications, as that provides a Unix-compatible set of APIs and commands; however, if that renders Windows "Unix-like", it would render z/OS even more "Unix-like", given that z/OS's Unix-compatible subsystem has been certified as conforming to the UNIX 95 specification[1]. Guy Harris 11:32, 25 January 2007 (UTC)

z/OS? I think that should be AIX.

Most such systems are commercial derivatives of the System V code base in one form or another, though a few (such as IBM's z/OS) earned the trademark through a POSIX compatibility layer and are not otherwise inherently Unix systems.

z/OS is a mainframe operating system. It does have something called UNIX System Services internally, but I suspect that the correct reference is "... (such as IBM's AIX) ..." Can anyone verify? —Preceding unsigned comment added by 69.109.233.83 (talk • contribs) 05:24, 9 February 2007

According to http://www.opengroup.org/openbrand/register/ , z/OS is UNIX 95 branded, so I think the current wording is correct. AIX has a reputation for being a somewhat funky UNIX implementation, but it's still UNIX 03 branded without resorting to a compatibility layer.--NapoliRoma 12:28, 9 February 2007 (UTC)

OpenVMS

I have removed OpenVMS as it has nothing to do with UNIX. It is a far descendant of RSX-11 and RT-11. The only thing it may share with UNIX is X-Window-compatible windowing system. But it was introduced only for easier support for UNIX terminals. Command line language, device names, file naming conventions, internal calls, system utilities and development tools all completely different.--79.120.8.182 14:52, 6 November 2007 (UTC)

NeXTSTEP

No mention of NeXTSTEP, even though Mac OS X is referenced momentarily - seems a bit odd, given the historical importance of the system. It could at least be included in the time-line chart, perhaps? 90.205.92.86 (talk) 05:57, 10 December 2007 (UTC)

OS X

Does Mac OS X fall under Unix-like? Ham Pastrami (talk) 06:46, 6 February 2008 (UTC) Ah, I see there is some ongoing discussion about this, so allow me to rephrase: does Mac OS X run binaries that are for Unix-like systems? Ham Pastrami (talk) 06:50, 6 February 2008 (UTC)

Mac OS X is Unix certified. So it's not Unix-like, it's Unix. -- AdrianTM (talk) 06:51, 6 February 2008 (UTC)

Any Unix system is about as Unix-like as it's possible to be, so it's not "it's not Unix-like, it's Unix", it's more like "it's not just Unix-like, it's Unix".

As for binaries, there's no guarantee that one Unix-like system will be able to run binaries for another Unix-like system - even if the two Unix-like systems use the same processor (which would typically mean x86 or x86-64, although various free-software Unix-like systems run on the same processors as the ones from the suppliers of proprietary Unix systems such as HP-UX (PA-RISC), AIX (IBM POWER/PowerPC/Power Architecture)) and semi-proprietary systems such as Solaris (SPARC)), there's no guarantee that they use the same object file format, much less a guarantee that they're binary compatible. What matters is more whether they can run a binary built from the same source code as the binary for another Unix-like system (or the same source code except for a few platform-dependent tweaks). Mac OS X is reasonably good about running programs built from the same source that runs on other Unix-like systems, as far as I know. Guy Harris (talk) 07:51, 6 February 2008 (UTC)

I agree that technically OS X is "not just Unix-like, it's Unix" (nice phrase :-)), but generally my feeling is that Unix-like is a legal category (OSs like Unix which are not certified) rather than technical, and certified versions of OS X should be left out, if only to avoid repeating content. NicM (talk) 08:16, 6 February 2008 (UTC).

Unix versus Unix-like - II

I don't agree with the way the proprietary Unix-like list is made. It lumps together Unix-like (non-Unix) systems like QNX and MacOS X with IRIX and XENIX which atually ARE Unixes, not Unix-like systems. --Arny 20:52, 17 March 2007 (UTC)

What criterion do you use to distinguish between "Unix" and "Unix-like"? Descent from the AT&T code? General UN*X compatibility? (If Apple succeeds, Mac OS X 10.5 will be a Unix(R) system.) Guy Harris 01:20, 18 March 2007 (UTC)

I define Unix as genetic Unix systems, so yes, it would mean descent from AT&T code. (Note that BSD's are also considered genetic Unixes, even though they're formally free of AT&T code...) Things get a little blurry with non-descendants that got a UNIX certification... however, I'd be quite reluctant to call, z/OS for example, a Unix. arny (talk) 12:26, 4 August 2009 (UTC)