“What’s really needed to settle this is a definition of Unix. Unfortunately, there’s no obvious definition against which we might compare Linux. The systems we all accept as unambiguously Unix — that is, BSD, HP-UX, Solaris and Tru64 — have more functional differences than, say, Linux and Unixware.” Read the article at LinuxInsider. Apparently, you also need to have the sign of approval to be called a “Unix” by passing some tests.
Is Linux Unix?
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Definitely YES, if we’re talking technical, not trademarks.
Its a clone of Unix and even among “official” SVR4 Unixes there are variations.
Clone implies exactly or nearly exact copy.
Since that’s what Linux is all about.. cloning…
Strictly speaking, no, because the Open Group owns the UNIX trademark and manages the Single Unix Specification, yet Linux is not compliant with the specification, and therefore does not meet the criteria to validly use the trademark. The trademark is somewhat diluted though, so there is a grey area that probably allows use of “Unix” as an off the cuff reference, but I think you need a trademark attorney to tell you whether it is allowable to use “Unix” as part of descriptive/marketing/etc material. Certainly the use of “UNIX” in capital letters would be even darker shade of grey.
There is no official Unix other than “The Single UNIX Specification”, because otherwise Unix just does refer to collective “style” of software (BSD, SVR4, and other implementations have clouded a strict heritage). Many vendors spent years harmonising on POSIX UNIX. Unfortunately, Linux seems to want to press ahead with standardisation of LSB at ISO, claiming it to be “sufficiently different” and other sorts of bull head closed minded attitudes. The POSIX guys are trying their best to help: recently they relaxed rules on allowing POSIX copyright information to be used in man pages, and they’ve made it clear that harmonisation is always a push and pull, and that POSIX will probably have to make some changes to work with the Linux way of things. The Linux guys don’t seem to want to budge though.
Anyone who spend years porting software between SVR4 and BSD will tell you that standardisation is a good thing. However, now that we’ve solved the big SVR4 x BSD, we seem to have a new POSIX v ISO-LSB coming up on the horizon. Now is the chance to stop it all before we’re 10 years down the road. However, people maintain bull headed ness.
First this guys article says that becuase the orgninal Unix promoted cooperation amoung developers and users and Linux does the same that they are the same, which is bull. Just because they are thought of the same way doesn’t make them the same, similar yes, functionally alike yes, but not the same.
Linus took a good idea, and made his own version of it, using tools that were available that were similar to the tools that he was already used to using.
just cause I use a hammer to make another hammer that looks different but is functionally the same, does that make it any less orginial?
There is no doubt that “Linux” is “Unix”;
However, “Linux” is not “UNIX(R)”;
Other “Unix”‘s are “UNIX(R)” though;
You do _not_ need a “sign of approval” to be called “Unix”, but you do need to pass a suite of tests to be “UNIX(R)”. This is a good thing. Just about every other sensible standard defines itself this way, allowing manufacturers free run to do as they please, so long as they meet the specific requirements of the standards. That makes for good certainty, interoperability, etc.
You may soon find that not all “Linux” distributions are “ISO-Linux”, in the same way. This will resolve some of the fragmentation in the many “Linux” distributions.
just cause I use a hammer to make another hammer that looks different but is functionally the same, does that make it any less orginial?
That’s somewhat of a fallacious argument. A more accurate representation of reality, still making use of your analogy, would be along the lines of the following. Assume X invented the hammer. The hammer is X’s innovation, as the hammer had not existed before X came along to create it–simple enough. Some time later, Y comes along and creates a new hammer that looks almost identical, except its handle is made of a different type of wood, the wood from the type of trees that were growing in Y’s backyard. In no way did Y innovate, and in no way is Y’s hammer equally original.
Nick
Before I say anything, I’m sure someone will drag TOG into this. I could really care less about TOG’s definition of “UNIX”.
To me Unix implies something which was derived from The Unix Time Sharing System, whose code structure and interfaces internally map back to the original Unix code from which it’s derived.
While BSD Net/2 saw the elimination of all functional USL source code, one of the big issues broached in the trial was the fact that the internal interfaces (and in many cases, the parameter names) were the same.
One project which really pointed out the differences between the Unix-derived operating systems and Linux was SGI’s port of XFS to Linux. This required SGI to cobble their own page buffer mechanism onto the Linux VMM through which XFS would attach, rather than using the Linux VFS itself. This was due to how discontiguous the Linux VFS interface is with other Unix operating systems. FreeBSD developers noted after looking at the huge list of hacks necessary to port XFS to Linux that virtually none of those would apply to FreeBSD, as the FreeBSD and Irix VFS interfaces were quite similar in comparison to the Linux VFS.
This has had some disasterous consequences. XFS is still dangerous to use on IDE drives with write caching enabled as the Linux VFS does not currently provide a mechanism for ensuring that journal data is synchronously written to disk. Consequently from the XFS implementation’s perspective, a journal entry will have been written to disk when it’s actually still in the drive’s write cache. Any files which have been altered prior to a power outage may be zero filled after the journal is replayed. Attempts have been made to solve this problem with things like the IDE barrier patch, but unfortunately the barrier patch isn’t well maintained and probably won’t see integration into the mainline kernel sources for some time.
That’s not to say that moving away from the original Unix design is a bad thing. Dragonfly will soon sport a radically different VFS interface from FreeBSD, which, from the description I read, seems much cleaner. Unix was written around some relatively ancient design requirements, and operating systems need to modernize to stay useful on modern hardware.
Semantically, is Linux Unix? Well, to you, is Unix an idea of how programs and users interact with the system, a name (owned by TOG), or a program system (The Unix Time Sharing System). As I said earlier, I prefer the latter. Those who prefer the former will undoubtably say yes.
can be called UNIX?
Do UNIX programs run on Linux? No … then no. Inspired by doesn’t mean same. If it did, the DOS is UNIX, which it isn’t.
You are correct in your working, but the answer is wrong.
“Unix” is a very loose term, and it does apply to “Linux”.
Analogies are dangerous, but some relevant ones are that you can classify many cars as “SUV”, despite the fact they are all not-interoperable (roof racks, parts, etc). It still fits into the “SUV” mold. For this reason, “Unix” itself refers to the broad category.
Now, _within_ “Unix”, you can get a more more detailed: SVR4-style, BSD-style, Linux-style, GNU-style, UNIX(R)-style, etc.
What TOG holds is only “UNIX(R)”, not “Unix”, or any of those other categories.
If you were to try and describe what makes a “Unix”, well that would be fun, a number of datapoints about how the file systems and structure are organised, the common daemons and startup processes, similar tools such as cat, ls, etc. I’m not sure where “cygwin” would fit in here, it’s not strictly “Unix”, but it’s more like “Unix compatible”, and if it adopted the right standards, it could be “UNIX(R)” compatible too!
In your argument about XFS, the value statements about compatibility are moot: the fact that a lot of work is required to port to one style over the other says nothing more than which style the original design of XFS was built for, it doesn’t really say anything about how “Unix” like one is. The fact is that despite the differences in kernel makeup and other factors, Linux is still “Unix” in its way of being.
The Open Group (or was it USL?) threatened Apple with a lawsuit for using the trademark “Unix” in one of their advertisements. That having been said OS X is clearly related to Unix, especially FreeBSD.
…but Unix is becoming Linux
By that logic, OS/2 is Windows and MorphOS is AmigaOS.
This page is consistently locking up Firebird (0.7), I had to use IE to read it…
It all depends on whether you believe in Unix the trademark or unix the functionality (or philosophy, if you prefer).
Linux is a unix clone in that it tries to duplicate unix functionality. Hence the term “unix like”.
Linux is not a Unix(R) because, well, it isn’t a Unix(R).
For all practical intents and purpose Linux is a unix (not an especially good unix, but a unix nonetheless).
njm (IP: —.mn.astound.net)
In no way did Y innovate, and in no way is Y’s hammer equally original.
We are not talking about the innovation. Isn’t Y’s hammer still a HAMMER? There are many variants of hammer but still all of them are called hammer aren’t they?
Sabon (IP: —.ci.seattle.wa.us)
Do UNIX programs run on Linux? No … then no. Inspired by doesn’t mean same. If it did, the DOS is UNIX, which it isn’t.
What da h*ll are you talking about? Even among Unixes they cannot share programs in binary form. In you login, then ones that we call Unixes are not Unixes! However programs in source form which is well implemented can be compiled and run on virtually any *nixes.
These guys make me to realise that Linux is Unix technically. I’m serious.
MobyTurbo: I don’t know the lawsuit, but Apple would not be able to label OS X as UNIX(R) unless it met the criteria of TOG. That’s probably not what Apple tried to do, but it illustrates the extreme. What Apple probably tried to do was use the word “Unix”, and TOG likely claimed that was a trademark infringement. Now that’s a grey area relying upon trademark law aspects of confusion, deception and related issues “in the course of trade” that don’t strictly correspond to the colloqual term of the way we use “Unix”. For example, hoover is diluted and generic, so it would be hard to get burnt for trademark infringement for writing an article referring to “the guy hoovering down the hall …”, yet if you were selling vaccuum cleaners and you use the world “hoover” anywhere near your material, then you’re going to get in trouble.
We are not talking about the innovation. Isn’t Y’s hammer still a HAMMER? There are many variants of hammer but still all of them are called hammer aren’t they?
No doubt, and I couldn’t agree with you more. Moreover, I definitely agree that Linux is deserving, at least colloquially, of being known as a Unix. I was simply highlighting a problem in peragrin’s argument.
In you login = In your logic
Technically, Linux is NOT UNIX, it’s a MINIX-based kernel. As it behaves as UNIX in many areas, you can say it’s UNIX-like. On the other hand, FreeBSD is UNIX but not UNIX(R).
Is Linux UNIX?
Well, what is UNIX? Someone has to come up with a singular definition by which we can compare the two. Crawl before we walk.
No…onto something completely different.
“that’s somewhat of a fallacious argument. A more accurate representation of reality, still making use of your analogy, would be along the lines of the following. Assume X invented the hammer. The hammer is X’s innovation, as the hammer had not existed before X came along to create it–simple enough. Some time later, Y comes along and creates a new hammer that looks almost identical, except its handle is made of a different type of wood, the wood from the type of trees that were growing in Y’s backyard. In no way did Y innovate, and in no way is Y’s hammer equally original.”
Actually, your analogy misses a huge point.
Check it.
X’s hammer was made out of steel, weighed 10,000 pounds, cost $10 million dollars and could only be run by certain equipment on certain job sites. Y comes along, sees how freakin’ cool X’s hammer is and is like “Man, I could really use a hammer like that, but I can’t afford that and I really think the design of the hammer should be so that everyone could use it…for all those big heavy things they are doing now or for the little things that I want to do.”
So, Y goes and builds his freakin’ hammer. Next thing he does is say “you know what, I think this should be free…and I think other people would want to see my exact approach to making this hammer….then they could make MY hammer better.”
What happens? Y’s hammer ends up being better than X’s hammer because it can do all of X’s work, but it can also do Y’s work. Y’s hammer isn’t a clone of X’s, and it isn’t built upon (derived from) X’s design. Y’s hammer is just a better hammer…but it is still a hammer
njm (IP: —.mn.astound.net)
Well the difference of the wood that the handle was made of doesn’t disqualify it as a hammer. Still X could make some handles out of the wood Y used (by, well, buying the wood from Y after X investigated that the wood is better and cheaper than the wood X originally used.) But still the one X made would be called hammer even tho the handle changed. Or in other way around Y could use the wood X used to make the handle after recognised that the wood X used is better.
The important thing is that a hammer is “A spherical weight attached to a flexible handle and hurled from a mark or ring. The weight of head and handle is usually not less than 16 pounds.”(from Webster) Anything that satisfies the purpose of the hammer and figures similar to the hammer, can be called a hammer. Nothing to do with the wood handle is made of or something. handle is there to just let people to hold it in comfort and concentrate on work.
>>>>
Technically, Linux is NOT UNIX, it’s a MINIX-based kernel. As it behaves as UNIX in many areas, you can say it’s UNIX-like. On the other hand, FreeBSD is UNIX but not UNIX(R).
<<<<
Incorrect there is no “UNIX” itself, the only combinations are “Unix”, “UNIX(R)” as “UNIX” implies “UNIX(R)” by trademark law.
Linux is not “UNIX(R)”, but it is “UNIX(R) like”; yet it could not actually use that term because of trademark infringement. Linux is “Unix” though.
FreeBSD is not “UNIX(R)” nor “UNIX” (since the terms are equivalent), yet it is “Unix”, and it is “POSIX(R)-compliant” (which is a subset of “UNIX(R)”)
They are both “Unix” rather than “Unix-like”: because if you say “Unix-like”, then you must imply there is a “Unix” somewhere: so please define what “Unix” is (is it an abstract concept that nothing can meet, or is it only claimed by the original AT&T 3B2 Unix ?). All “Unix”‘s draw a heritage (even if in different implementations) from that original AT&T code.
I thought Linux is part of the GNU software stuff. Therefore, but definition Linux is not Unix because GNU’s Not Unix. Right?
Besides, I would consider Linux to be more of a work-a-like or a clone of Unix. Work-a-like would probably be the best term to describe it.
On the other hand, I would consider BSD to be Unix because at one time it did contain Unix source code. Although, that Unix source code was removed because of the lawsuits from AT&T and Novell back in the early 90’s. Which is why SCO *probably* can’t touch BSD, but I’m sure they’re trying.
By n0dez (IP: —.58.11.45.proxycache.rima-tde.net)
Technically, Linux is NOT UNIX, it’s a MINIX-based kernel. As it behaves as UNIX in many areas, you can say it’s UNIX-like. On the other hand, FreeBSD is UNIX but not UNIX(R).
What? Linux kernel is MINIX-based? Can anyone confirm this? Then why on earth did Tanenbaum argue Linus about the design of Linus’ kernel?
Bah, you people are just plain impossible.
t3RRa: I agree, and have made no claim to the otherwise.
JasonW: your point make sense, but doesn’t change what I said. Even considering everything you brought up, ask yourself whether any of that “makes [the hammer] any less orginial?” Yes, it does.
That is all. Move along.
Maybe the author then would like to clarify what OSs are not “Unix” then. Based on his logic, DOS with networking, AmigaOS and whatever else is out there be considered Unix – even if they don’t have POSIX compilance – since there all would promote a communal communication and comradre.
-D
You originally said:
“In no way did Y innovate, and in no way is Y’s hammer equally original. ”
I pointed out that Y did in fact innovate. Y innovated by building a BETTER hammer, a more cleanly designed and implemented hammer. This in addition to pricing it differently and using it for a different purpose.
Are you seriously going to argue that the stealth bomber is not an innovation from the original bi-plane? No, because you aren’t stupid. Same thing here.
GNU is unique in that it is not a UNIX. It is something different. Better perhaps, but certainly not stagnantly traditional.
Linux is not minix-based, but minix-inspired.
It was due to dis-satisfaction with Minix that Linus started Linux. There is a classic newsgroup/email debate between Andy and Linus. I think Andy was the right one in fact and degree, but Linus had all the youthful rebellious attitude :-).
@n0dez: Linux is not in the least MINIX-based. Linux uses certain bits of MINIX (for example, various magic numbers for certain named constants) but the kernels share no code and are vastly different — Linux is a monolithic kernel, while MINIX is a microkernel, for example.
@Anonymous (freedom2surf.net): As I remember, it wasn’t just about Apple used the word “UNIX” but Apple clearly trying to leverage the UNIX trademark to market OS X. Apple’s OS X pages use the word “UNIX” anywhere from 4 to 10 times each. Before the lawsuit, their main OS X page used the word “UNIX” 7 times, and had a big metal “UNIX based” logo. Apple has not paid for the right to use the UNIX trademark, and has not certified OS X as a real UNIX by passing the conformance tests, so The Open Group was justified in sueing them for it.
@Anonymous (freedom2surf.net): The Linux developers refusing to change things in the system isn’t necessarily about bull-headedness. They are pragmatic. They care about choosing what they feel are the best interfaces, not necessarily the most compatible or widely used. For example, Linux 2.6 uses a new mechanism called epoll instead of the traditional /dev/poll, because its faster and scales better. They also are on good political footing for trying to get things done their way. Going forward, Linux will be the most widely used UNIX. SGI is already moving over to Linux on Itanium, and has declined to develop an Itanium port of IRIX. IBM has already stated that in the long term, Linux will replace AIX on POWER.
GNU is really a “Unix-compatble”, and isn’t really the whole of “Unix” but an equivalent of a subset because GNU is just the tools and everything else: nothing about the kernel and the overal operating system lifecycle.
In the same way “Linux-kernel” is not “Unix”, but is really “Unix-compatible” in the same analogy to GNU.
When you put “GNU” + “Linux-tools” together to make “GNU/Linux”, _then_ you have something that is “Unix”.
Frequently, when we say “Linux”, we really mean an informal definition of “GNU/Linux”. You can also have “foo/Linux” as well, and what the LSB is trying to solve is this problem of so many “wibble/Linux”‘s all floating around doing a different thing.
Rayiner Hashem:
Bad argument about LSB. It’s not able pragmatism. The POSIX folk are just as pragmatic, they are saying that they may have to adjust the POSIX standard to include some of the Linux specific things. They certainly did that for other Unix’s during the harmonisation project. Your use of the term “pragmatic” is an excuse for “we don’t want to change”.
The problem is that there is a huge base of software built on compatibility with POSIX, and without coming to some consensus about equivalence in the standard, then it’s going to cost manufacturers to port this from POSIX style to Linux. Now, if both Linux and POSIX can share some give and take to agree to move the POSIX standard forward so that it harmonises, which will involve a bit of pain for the different camps, then it will pay off significantly for the manufactueres (of software) as they do not need to go through (as) painful conversion processes.
As you say, Linux will be the most prevalent Unix, but the Linux camp – and you too – seem to be pulling a Microsoft argument by using that fact to justify remaining in an entrenched position and effectively telling the manufacturers of software to “stick it and wear the cost of coming over to our way of doing things”.
Equally, the POSIX community could use the same argument because the sheer base of POSIX compliant applications is huge: yet POSIX are mature enough not to through their way around, but try to work and mediate and build consensus on the issue.
Are you seriously going to argue that the stealth bomber is not an innovation from the original bi-plane? No, because you aren’t stupid. Same thing here.
That is a ridiculous straw man. I’ll let it slide, since continuing this discussion will only lead us further off topic (if OSNews had threading, things would be different!).
Anonymous (IP: —.dsl.sbndin.ameritech.net):
You have a good point, we can’t just use the term “Linux”, we always need to qualify it, either “Slackware Linux”, “Gentoo Linux”, “GNU/Linux” (more abstract than concrete), “Linux-kernel”; and then when it’s expressed like this, then it can be measured against “UNIX(R)”, “Unix”, “VMS”, “Windows”;
so at least _at a minimum_ we could say that “Linux-kernel” is “Unix-like”, and then as “Slackware Linux” it is “Unix”, but not “UNIX(R)”, and who knows if someone comes along with “WibbleDistro Linux” that can be called “UNIX(R)” because it meets the certification criteria.
Then, to introduce LSB into the equation, we might have any of those combinations _also_ being “LSB(R)” because it meets LSB certification criteria.
Then we have “NewWare Linux – UNIX(R) LSB(R)” for dual compatibility.
While that’s all very nice; it smells of the old USR days of “HST”, “V.32”, “Dual HST/V.32” until finally we ended up with “V.34” (admittedly, fdx rather than HST style asymetric).
So _at least_ if the LSB guys aren’t going to move, then they should move “just as far” to support dual “UNIX(R) LSB(R)” which may involve a bit of pain in both camps. Perhaps then in the future, both LSB(R) and UNIX(R) will converage and become equivalent – that would be a great thing too.
ME:
Are you seriously going to argue that the stealth bomber is not an innovation from the original bi-plane? No, because you aren’t stupid. Same thing here.
YOU:
That is a ridiculous straw man. I’ll let it slide, since continuing this discussion will only lead us further off topic (if OSNews had threading, things would be different!).
ME again:
Here, let me rephrase that for ya.
Would you argue that the stealth bomber is an innovation from the original bi-plane (the humble beginnings for the stealth bomber)? See, no more strawman.
So, what counts as innovation in your book? does it have to have a pretty gui? does it have to say “Hey, I am not a copy, I am a new thing…no really” at boot? What counts? Does an alien have to come down and just hand something to you that has never been on earth before?
Basic principle for you: innovation is not the same thing as invention. They are two different words.
Is Win+SFU UNIX but not UNIX(R).
Honestly I find these discussion really silly.
So, what counts as innovation in your book? does it have to have a pretty gui? does it have to say “Hey, I am not a copy, I am a new thing…no really” at boot? What counts? Does an alien have to come down and just hand something to you that has never been on earth before?
Come on! I use Linux and BSD exclusively at work and home, and have for years. For God’s sake, my GUI at home is Motif, which I use for fun! By no means will I require something to have a “pretty gui” to be considered innovative.
What do I consider innovative, then? I’ll entertain you, now. Not Linux. Allow me to state something with which almost everybody, including almost certainly Linus himself, will agree: Linux is a reimplementation of Unix. The two are so blatantly similar that countless utilities can be ported back and forth with almost no effort. Linux has in fact explicitly sought interface compatibility (e.g. POSIX) with Unix in many regards. This is obvious.
Now, a stealth bomber is most definitely not a reimplementation of a biplane. Do you consider a Ford V6 engine to be an innovation over a Chevy V6 engine? No, both are simply reimplementations of the same concept, just as Linux is a reimplementation of the original concept of Unix. I’ll continue: is a jet engine an innovation over a piston? Certainly, but they operate in such fundamentally different ways–in terms of both internal (of course, as any two engines will differ in terms of internal details) and external interfaces. Allow me to ask you a question now: name one technological aspect in which Linux is innovative over Unix? For clarity, I’ll offer my working definition of the concept of innovation: “A change effected by innovating; a change in customs; something new, and contrary to established customs, manners, or rites.” And, I’ll go ahead and outright grant you the libre software side of things, an aspect in which I genuinely do think that Linux is innovative.
Without specific evidence, this discussion is pointless. You claim that the Linux people are being bull-headed about compromising with the POSIX people. My point was that often, the refusal to compromise is not a result of being bull-headed, but being pragmatic. The Linux developers are by and large *not* diplomatic. If they don’t feel that something is the best technically, they just won’t do it, whether or not its a good move from a compromise standpoint.
Oh, and one more point: Until very recently (a matter of a month or so), the POSIX standards were not free. Thus, Linux was forced to be designed somewhat independently. You can see why this would make this less amenable to compromise this late in the game.
I classify something as being UNIX if it is based off the original UNIX code from AT&T. If it is not based on that code, then it is not UNIX. Just because it looks like UNIX doesnt mean that it is.
Even WinXP has some POSIX functionality, should we be calling that a UNIX derivitive?
Re: Pretty gui –
I was just askin!
Re: reimplementation
Wha? because something seeks to play nicely with something else it is now a reimplementation? Is that what you are saying? And reimplementations are not innovative?
You are going to have to come up with a better argument than that.
Re: chevy v6 vs ford v6
Well, considering you setup a question that is in no way related to the topic….since a ford v6 is based upon ford specs and a chevy v6 is based upon chevy spec. I would have to say that if the engines are differently designed, the one that functions the best is indeed an innovation over the other.
What about cold air intake? how about modding a chip on an onboard computer? Is that innovation? Of-freakin’-course!
I think you are hung up on the term, for some strange reason.
Re: What has linux innovated?
1. the culture of computing. Uhm…free vs not free etc etc
2. bringing servers to the masses
3. computer system interopability (GNU software to make linux bases systems talk to everything else in the world)
I mean, you ADMITTED it…you said linux innovated the pricing structures…what else does it have to innovate to be “innovative”?
And yes, I like to argue
And yes, I like to play devil’s advocate
I work as an AIX administrator and the skills I use in my job on AIX I can transfer to FreeBSD and Linux with no more (or less) difficulty than it would take to move from AIX to Solaris. So IMHO by any practical definition all of the above are unices (or unixes, whatever)
> Without specific evidence, this discussion is pointless.
The specific evidence is found in TOG mailing lists where the issue was debated. I assumed it was common ground that if you continued this debate you were aware of assumptions, otherwise one of us is participating in something they don’t know anything about.
>You claim that the Linux people are being bull-headed >about compromising with the POSIX people. My point was >that often, the refusal to compromise is not a result of >being bull-headed, but being pragmatic. The Linux >developers are by and large *not* diplomatic. If they >don’t feel that something is the best technically, they >just won’t do it, whether or not its a good move from a >compromise standpoint.
Well, that’s an argument that doesn’t pull much weight, as it seems a poor excuse to force excessive costs onto the community as a result of not compromising and making the portability path less expensive. In all the arguments about the “community feel” of open source, it’s a responsibility on engineers to work with the community. Linux doesn’t just exist in a vacuum. Again, this point has been discussed by on TOG lists (including LSB representatives who engaged to put forward the Linux persective) by people more aware than I am, and the consensus opinion is clearly that benefits would accrue from working towards harmonisation. It was acknowledge that some Linux LSB folk have “ego programmer” issues happening, which is not conducive consensus oriented future where there are many stakeholder interests to represent.
>Oh, and one more point: Until very recently (a matter of a >month or so), the POSIX standards were not free. Thus, >Linux was forced to be designed somewhat independently. >You can see why this would make this less amenable to >compromise this late in the game.
Well, lets look at the facts: implementing and using the standards is and always has been free (no royalty etc), it just cost money for the paperwork and cost money if you wanted to license the specific copyrights in the works to reproduce them in man pages. They didn’t prevent independent construction of man pages. This is an approach used elsewhere.
Naturally, the problem is that it costs time and effort to do that independent construction (we ignore the cost of buying a copy of the standards because in the scheme of things, it is really negigible, and also I don’t know whether this was a revenue stream by TOG to pay its way).
Now also, are there any cite’s for previous compaints to TOG about this being a problem ? If so, tthen it’s an issue that TOG took so long, but as far as I understood the activity, TOG allowed reproduction of the standards in man pages. Now this actually cost TOG something in lost revenue, and it’s a pretty good concession on their part towards harmonisation. Kudos to them for offering a leaf to make it easier to harmonise.
Anyway; I think it non-productive to continue this line of debate, I do my best in the interests of harmonisation to advocate where I can, but it falls on deaf ears.
Well, considering you setup a question that is in no way related to the topic….
And your stealth bomber vs. biplane argument was more related how? =) Surely you can share with me the right to use analogy, especially when my analogy is at least somewhat more closely applicable to the topic at hand.
Anyway, I’m going to throw in the towel, as it’s clear that we’re both arguing fundamentally unrelated points–I’m discussing technology, you’re discussing culture. I certainly agree with you about the culture side of things (hell, I even said so in my last post), but I’m not going to concede much if anything within the sphere of technology.
There!
If Linux is Unix then Linus and Co. have been completely wasting their time all these years, and doing nothing but cutting and pasting code. You “Yes” people go and tell Linus that.
>Even WinXP has some POSIX functionality, should we be >calling that a UNIX derivitive?
Don’t be silly – but you can call it “POSIX compliant”, and “UNIX(R)-compatible”, and even “Unix compatible”, but it is definitely _not_ “Unix”, or “UNIX(R)” or “Unix-like” or “UNIX(R)-like” either.
You don’t call a station wagon an SUV derivative simply because they have the same petrol engines.
> Silly question, really
No it’s not: commercial decisions and choices about spending a lot of money depend the extent that things are compatible. And being able to use the right terminology to represent a product certainly makes it easier to make those decisions without having to compare all the details up front.
>If Linux is Unix then Linus and Co. have been completely >wasting their time all these years, and doing nothing but >cutting and pasting code. You “Yes” people go and tell >Linus that.
Huh ? “Unix” is now simply a generic term to refer to a class of product, not a specific concrete product, which is why we refer to “AT&T Unix” when we talk about the “original thing”, like “asprin”: people use “asprin” interchangably for products of the same basic composition, and even for products that achieve the same end use “to cure headache” (paracetamol, etc). In the same way you say “I want to buy an SUV”, you say “I want to buy a Unix OS”, and out comes FreeBSD, OpenBSD, Distro/Linux and so on.
Well, here’s the test actually: I challenge someone to walk into numerous computer stores and say “I want to a Unix OS, what do you have a available”, and just tell me how many people come back with a copy of red hat / mandrake / etc :-). I reckon that this would prove the case that “the Unix term has diluted to become a generic word for class of good”, and “Linux is a Unix” in this case. Of course, Linux is not “the Unix” – there is no “the Unix” anymore.
Linux is definately not Unix. For example, we humans can claim that we are evolved from one original humanaoid creature. During those times, we have evolved and changed and today there is a wide variety. However, we still share that common root and we all are still homo sapiens. On the otherhand, if some biological engineers got together and created some artificially intelligent creature, regardless how close it may look like us and act like us, it can not be considered a homo sapien. The creature will clearly be a different species. The classification is in the genetic details not in the overall picture. It isn’t important that something works similar conceptually but what is important is the exact details of how it was made. The exact DNA sequences creating the protein are far more important than having proteins with similar functions. Similarly, Linux has a lot of system calls/functions that have the exact purpose and approach to solving a problem that Unix has, but the fact that how the actual code is different clearly means it is not Unix. I’m not saying that Unix is better because it is Unix, nor am I saying that Linux cannot replace Unix. But Linux is definately not Unix.
Rubbish: that’s a bad example. A human and a robot are as distinctly different a Linux and OpenBSD, no one would ever use the terms to refer to the other. However, both a human and a robot are likely to be equally referred to as a person, or a guy or girl (even if they are approximate only because of artificate construction) – “check out the girl over there, she’s a robot/android”; “hey, I hate Linux”, “hey, I hate robots”, “hey, I hate Unix”, “hey, I hate people”.
My challenge remains: take a poll of how many people use the term “Unix” to describe “Linux”, try it either in a technically savvy community, or elsewhere (e.g. consumer stores with less than savvy sales people), and I’m betting that if you asked for “Unix”, you’d get presented with a Linux, BSD and related distros. Hey, my local bookstore even has a “Unix” sign on the cabinet where all the Linux, BSD and related books are kept. They certainly think Linux is Unix. On amazon, I find “Software > Categories > Home Computing > Operating Systems > Linux & UNIX”,
This article is short on history and long on blather. Linux is not Unix. It is Unix-like. It was not based on Minix, either, as this author states.
No. Linux (and the BSDs) are UNIX-like.
1) The term “UNIX” comprises the kernel *and* the userland, however, the term “Linux” describes the kernel only.
Hence Linux is no UNIX.
2) Linux was not, is not, and probably never will be certified by the Open Group, mainly because it costs a LOT of money.
Hence Linux is no UNIX.
3) Linux’ source is not directly derived from any AT&T UNIX source, but was and is written from scratch.
Hence Linux is no UNIX.
It’s as easy as that.
“Is Linux UNIX-like?” is a different matter entirely.
And the answer would be yes.
The irony of it all?
Microsoft sent in an early version of NT (that would later become NT4) for certification — and it passed…
Well, if the Name LINUX (Linux Is Not UniX) means anything 😉
There’s one point missing, Linus never used the MINIX kernel he only used the MINIX filesystem on which Linux booted with.
Linux never came from Minix at all only the filesystem.
That article made a nice philosophical point that I hadn’t considered. I can think of dozens of other points of similarity that are also in the philosophical vein. However, most of those would focus on the details of the system. This articles showed me one of the overarching goals of unix.
Nice to know where the original unix developers were coming from.
First of all, we have an uncomfortable problem: We have to accept that due to legal complications, there are no less than THREE ways of defining what Unix is.
First of all there’s the lawyer’s definition. This is the one where something is Unix if the Open Group takes your money and sprinkles you with holy water and says, “You’re okay with us, sucker.” This definition has very little to do with software. In fact, some years back, they had a little section on their web site that said that if Microsoft were willing to put up the money, they’d be willing to certify NT as “UNIX”. That’d be really scary. Obviously, this isn’t the definition of Unix we’re looking for….
Secondly, there’s the code heritage definition. This is a commonly used one, and there’s something to be said for it. By this definition, Linux is most definitely *not* Unix, since it was coded from scratch under the GPL through a totally transparent developer process open to peer review. This is being contested by SCO, but as we all know, they’re just folks that have suffered the migration of a small woodland creature up their collective butts, and the animal’s movements are causing them to behave in peculiar ways that no doctor can really explain. By the code heritage definition, Linux is not Unix. Also by this definition, BSD is, since despite the removal of the offending AT&T code after the lawsuits, BSD’s code heritage is very definitely steeped in the ancestry of Unix.
And finally, we have the third definition, the functionality-based definition of Unix, which is the one I use myself. In this definition, we say if it walks like a duck, and it quacks like a duck, then either it’s a duck, or it’s the best darn duck impersonator we’ve seen in years. This is the same type of definition that allowed people in the PC industry to start calling clones such as those made by Compaq, HP, Dell, Gateway, and so on just “PC’s”, and not “PC compatibles” as they had always been obliged to say during most of the 80’s back when “PC” was not yet considered a generic term. The reason those companies can all just call their machines PC’s now is because despite the fact that it was originally the name of an IBM product, it has become so generic in common usage that it would be onerous and absurd to make people say that any machine that conforms to the IA-32 architecture isn’t one. By the same logic, any operating system that gives you TTY’s, POSIX behavior, vi, emacs, the X-Window system, shells, and the whole nine yards, is probably Unix in any way that matters to 99% of tech weenies.
There are some gotchas that keep people from seeing this. One is the idea that if Linux were Unix, you would be able to take a program executable off of a Solaris Sparc and run it on Linux. That’s a specious conclusion; you can’t even take a Solaris executable and run it on an HP9000 running HP-UX either. Does that mean that HP-UX isn’t Unix? In Unix, software compatibility is determined at the source code level: Can you *compile* a Solaris program on Linux? Of course you can. (Which just goes to show you that there’s usually no such thing as a “Solaris program”….)
Another gotcha is GUI’s: Linux systems typically come running KDE or Gnome by default. “Unix” systems such as Solaris and HP-UX typically come running CDE or Motif by default. It’d be another specious conclusion to infer from this that it’s some kind of real difference though. Linux is perfectly capable of running CDE or Motif, and vendor Unix systems are perfectly capable of running KDE or Gnome, or Afterstep or any other window manager desired. And the reason for this is because of what we just covered above: The compatibility is determined by the compiler, and since any of these systems can *compile* any of these GUI’s, any of them will run on any of them.
And then there’s always the good old, “But there are DIFFERENCES between Linux and Unix!” This one is almost always from people who have never had any experience with any other other such system besides Linux itself. Anyone who’s been around Unix for any time at all realizes that there are boatloads of differences between *all* of them, going from Solaris to HP-UX, HP-UX to AIX, AIX to Tru64, and so on. They’re all totally different. What do they all have in common enough to make us think they’re all “Unix”? They all compile Unix software. Note, not *execute*, since as we saw above, there is no expectation of binary compatibility. But they COMPILE the same software. Linux is no outsider on that criteria. It compiles anything the rest of them can, and runs the same programs. By the functionality-based definition, Linux is Unix.
Linux was created to make an open source version of Unix. It’s ment to “Clone it” however, the code isnt the same but alot of it is actually, as you know BSD code is in linux- ALOT of it because the BSD license is compatable with the GNU License. It’s not an exact copy but linux was made to meet the UNIX specifications but since linux has grown wild it may create its own specifications that unix may have to follow… So.. Linux is a UNIX-LIKE System. yup, thats it.
For the millionth time, there is not a lot of BSD code in Linux. Oh there is, is there? OK show me.
One other thing, Linux is intended to implement the Unix API – ie the syscall layer. Call that cloning if you like. Yes, I suppose it cloned the system call semantics.
“Rubbish: that’s a bad example. A human and a robot are as distinctly different a Linux and OpenBSD, no one would ever use the terms to refer to the other. However, both a human and a robot are likely to be equally referred to as a person, or a guy or girl (even if they are approximate only because of artificate construction) – “check out the girl over there, she’s a robot/android”; “hey, I hate Linux”, “hey, I hate robots”, “hey, I hate Unix”, “hey, I hate people”.”
Unix is not a generic term like person. That is the reason why Unix is capitalized and person isn’t. I think the generic term you are looking for is operating system. Just because Linux smells like Unix, walks like Unix, acts like Unix, doesn’t mean it is Unix. Look at the underlying code and compare it to Unix. I can assure you that the similarities are almost nill. Otherwise, SCO would have a pretty good lawsuit on their hand.
“My challenge remains: take a poll of how many people use the term “Unix” to describe “Linux”, try it either in a technically savvy community, or elsewhere (e.g. consumer stores with less than savvy sales people), and I’m betting that if you asked for “Unix”, you’d get presented with a Linux, BSD and related distros.”
Actually if I were to take a poll, I guarantee you that the vast majority would disagree with you. The ignorant will be clueless as to what I am talking about, the Unix ™ admins will be offended for suggesting that Linux can even be considered Unix and point to their proud heritage, and the Linux admins will be insulted for trying to tie in 30 years of legacy code and thinking to their baby. The only people who would agree with you and say that Linux is Unix are the Windows admins. But to them whatever isn’t Windows and requires typing is considered Unix.
” Hey, my local bookstore even has a “Unix” sign on the cabinet where all the Linux, BSD and related books are kept. They certainly think Linux is Unix. On amazon, I find “Software > Categories > Home Computing > Operating Systems > Linux & UNIX”, ”
Well, if Linux was really UNIX… isn’t saying Linux and UNIX a little redundant. If not, why doesn’t the sign read, “Linux, Unix, Solaris, AIX, HPUX, IRIX….?” The reason why Linux and Unix are usually bundled together in bookstores is because they way they are administered is very similar and they are usually used by the same target audience. In addition, there aren’t too many books on Unix or Linux (as to compared to other sections). Finally, they usually are catering to a lot of Windows admin who want to learn Linux but think they will find the books in the Unix section.
I swear, sometimes I think it means Linux Is Not Useful….
But only sometimes ;o))))) (allright, fireproof suit on)
Very well argued, better than I could do myself. The answer really depends on what you define “Unix” to be, and this seems to vary quite a bit from person to person. It seems clear to me that there is a colloquial use of the word “Unix” denoting a class of operating system giving similar functionality and operating in the same spirit. How popular this usage is, I don’t know. Certainly the code heritage definition is popular, and UNIX == UNIX[TM] is used if for no other reason than people don’t want to get sued by The Open Group (TOG).
For several years I have seen people use “*nix” or “free *nix” when talking about linux and the BSDs. These constructions, based on shell globs, seem to be meant to match “Unix” or “Linux” equally well, putting the two terms into a sort of equivalence class.
Linux Is Not Useful … Xenophobia?
Whats the debate about? UNIX is a specification, you either conform to it, or you don’t. Want to call yourself a UNIX, submit your implementation of the UNIX specification to the OpenGroup to get checked out. Once you’ve paid the fee and your implementation has passed, then you can scream that your implementation is a UNIX, until then, at the most it can be called a “UNIX-like” operating system.
Why are there so many people who are ignorant. It isn’t as though this is something new. Its like Java, you can call your implementation Java once it has passed the test and you’ve paid the fee, until then you can call it anything other than Java, heck, you can call it “coffee maker” if you want, just don’t call it Java.