Talk:Unix/Archive 5
This is an archive of past discussions about Unix. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 | ← | Archive 3 | Archive 4 | Archive 5 | Archive 6 | Archive 7 | Archive 8 |
"Unix was unique in its time"
Should the line be something like it was the first to have online documentation? Unix isn't dead is it? --Gbleem 01:32, 12 November 2006 (UTC)
- I agree that it should be a little more specific, but I'm not knowledgable enough on this subject to comfortably fix it myself. --Merovingian ※ Talk 01:36, 12 November 2006 (UTC)
it definetly isnt death, Mac osx uses UNIX as kernel —Preceding unsigned comment added by 77.251.99.102 (talk) 13:46, 24 August 2009 (UTC)
- UNIX lives on in very hale health, through MAC OS X, the many BSDs, Solaris and trademarked UNIX vendor versions, along with its kissin' kin Linux. Unixes have a much deeper market depth in "mission critical" stuff like commercial/business servers and secure desktops, where high security and steadfastness are needed. Also, much of the web is still served by unixy OSses and many Windows users would likely be startled at how advanced and media-friendly X Window GUIs (desktops) have become. Gwen Gale (talk) 14:49, 24 August 2009 (UTC)
UNIX had MAN pages which might be considered as online documentation. Using a 3270 on VM/CMS required a printed manual. About as useless as "photographic memory" on an open book test: still have to read and comprehend the material. MAN pages were unorganized and cryptic. Shjacks45 (talk) 22:18, 16 March 2012 (UTC)
BSD TCP/IP stack claims
I don't see the source for the claim that most TCP/IP stacks today are descendants of the BSD TCP/IP stack. I've read that Microsoft licensed a BSD-derived stack for Windows NT 3.1 (and also Windows 95, I believe) from a company called 'Spider', but Microsoft claim to have put a 'completely rewritten' TCP/IP stack into Windows NT 3.5 (1994, I think): http://www.microsoft.com/technet/network/deploy/depovg/tcpip2k.mspx. There have also been various claims that Microsoft wrote yet another TCP/IP stack for Vista.
On the whole, any claim about 'most systems' is going to effectively mean whatever Microsoft are using, given the large market share of Microsoft Windows. As such, I'm inclined to remove this claim unless there's some supporting evidence from a reliable source (e.g. not a forum posting by a random individual). The network interfaces defined by BSD (BSD sockets) have unquestionably become the standard on most systems, but that does not imply the implementations are descendants of the BSD code itself. The limited evidence I've found seems to indicate, at least in the case of Microsoft Windows, the latter does not apply. Shalineth 22:01, 3 March 2007 (UTC)
- The claim was correct as of a few years ago; I don't know how accurate it is today. The number of TCP/IP stacks means the number of implementations (platforms), not the number of fielded units. Thus, Windows just counts as one, not as a zillion. -- DAGwyn 22:23, 6 March 2007 (UTC)
- I see what you mean, but the exact wording is, "almost all TCP/IP network code in use today", which I think can be read as referring to systems in use (and maybe also as the number of implementations). It could be clarified relatively easily, so maybe I'll do that, but it would still be ideal to have a solid source for the claim.
- I've also seen the "Windows uses a BSD TCP/IP stack" claim for a long time in forums and such, but the only documented sources I've ever seen for the NT line have referred to Windows NT 3.1 specifically (which has a totally different TCP/IP stack to NT 3.5 and later). This has sometimes been confused because later version of Windows continued to include BSD TCP/IP tools such as ftp and telnet (which contained BSD copyright notices in the binaries, whereas the driver modules that implemented the TCP/IP stack did not). Shalineth 14:15, 10 March 2007 (UTC)
- I haven't examined the Windows implementation, so the follwoing is just speculation, but it always seemed to me that in the early days Microsoft was hoping IP-based networking was a passing fancy, and when they were forced by market pressure into supporting it, the fastest way to do so was by porting the BSD TCP/IP stack. It wouldn't be surprising if they eventually replaced it by a home-grown implementation. — DAGwyn 05:14, 11 March 2007 (UTC)
In the XP EULA and licend it has the standard BSD licence disclaimer. I know utilities like FTP are currrently ported from OpenBSD so there is definately SOME BSD code in there. 134.36.93.46 (talk) 04:50, 21 May 2009 (UTC)
XP? Microsoft started using BSD stack in betas of Win95, about late 1993 (would include NT3.5), I was paid to troubleshoot Windows 3.1 network issues and Win95 network problems. NT 5.0 and NT5.2 code: Windows 2000 and 2003(XP) also. "Linux User and Developer" magazine had article(s) about SAMBA incompatibility with Vista (NT6.0) and Windows 7 (NT6.1) due to change of Internet stack. Wikipedia article Windows Vista networking technologies refers to stack "named Next Generation TCP/IP stack", or SMB2. Shjacks45 (talk) 22:56, 16 March 2012 (UTC)
Careless use of the word "modern"
I'd suggest that in general, and for computer science articles in particular, the word "modern" be avoided where possible. Is Solaris 10 "modern" and for how long will that be?
In this case, I'd suggest that a year (for example) be used.
Information such as this is always in flux, but there's no need for planned obsolescence. —The preceding unsigned comment was added by 76.18.106.66 (talk) 22:48, 29 March 2007 (UTC).
- Much Agreed. Dare to say Windows is not modern, and this is agreed. [1] and
Does Microsoft's buggy patched Windows define modern OSs?."Having studied Total Quality Management this quarter, I’m aware that Microsoft is notorious for not implementing the processes that would produce the best quality output. Problems could be always patched over anyway. Microsoft’s softwares, let alone Windows, are historically known to be vulnerable to plethora of bugs."
- Much Agreed. Dare to say Windows is not modern, and this is agreed. [1] and
- Let me ponder: Modern OSs. No such thing. Your right, ( perish the thought that Solaris 10 is "Modern"?!?!?!?! ) — Preceding unsigned comment added by 71.202.117.96 (talk) 10:02, 2 September 2011 (UTC)
OK. How about: Linux only works on Legacy Hardware because there are no drivers for new Wireless cards, Video cards, Printers, doesn't support UEFI, doesn't support Native mode SATA, ... Shjacks45 (talk) 23:04, 16 March 2012 (UTC)
Article contradicts itself
Beginning in the late 1980s, an open operating system standardization effort now known as POSIX provided a common baseline for all operating systems; IEEE based POSIX around the common structure of the major competing variants of the Unix system, publishing the first POSIX standard in 1988. At around the same time a separate but very similar standard, the Single UNIX Specification, was also produced by the Open Group. Starting in 1998 these two standards bodies began work on merging the two standards, and they are now identical.
Contradicts:
In 1996, X/Open merged with OSF, creating the Open Group. Various standards by the Open Group now define what is and what is not a "UNIX" operating system, notably the post-1998 Single UNIX Specification.
How could the Open Group have created the Single UNIX Specification "at around the same time" as 1988 if the Open Group didn't exist until 1996? —The preceding unsigned comment was added by 216.23.105.16 (talk) 14:22, 1 May 2007 (UTC).
- The second section is correct; the SUS name came later. I just changed the Standards section to reflect this; what is not clear to me is whether the Open Group's use of the SUS name predates the efforts to sync the two standards; I believe it did, so the edit I just made may need a little refinement.--NapoliRoma 15:32, 1 May 2007 (UTC)
- I think the GSA got involved in specifying POSIX in government contracts. "X11 server interface" doesn't define UNIX any more than Motif defines X11 or WOW defines Windows, or Google's Java presentation interface defines Android. Or Linux Kernel Source Code not compiled by GCC compiler is not a Linux Binary, or Unix source compiled with GCC is a Linux Binary? Shjacks45 (talk) 23:19, 16 March 2012 (UTC)
Mac OS X 10.5 for Intel is now officially UNIX
The 18th of May 2007 Mac OS X 10.5 Leopard for Intel was rewarded the UNIX'03 certificate. [2] -- Henriok 16:45, 12 June 2007 (UTC)
Could it be possible to add it to the UNIX genealogy graphic? —Preceding unsigned comment added by 200.57.2.113 (talk) 15:40, 22 February 2008 (UTC)
- The graphic on the Unix page is labeled as a "filiation of Unix and Unix-like systems", so the fact that it's now covered by the UNIX trademark is irrelevant; it's a significant enough Unix-like system (note that a licensed UNIX system is also a Unix-like system) that it arguably belongs in the graph, although if you add it you should probably add NEXTSTEP to get the full history. Guy Harris (talk) 18:51, 22 February 2008 (UTC)
Either way, MAC OS/X, NEXTSTEP or even OpenStep ought to be included ? —Preceding unsigned comment added by 90.196.3.37 (talk) 16:04, 24 October 2008 (UTC)
Avie Tevanian, was formerly head of software at NeXT, then Chief Software Technology Officer at Apple Computer until March 2006 —Preceding unsigned comment added by 90.196.3.37 (talk) 16:21, 24 October 2008 (UTC)
And I think in connect with the above it would be appropriate to mention Steve Jobs's name. AlvinMGO (talk) 16:29, 19 June 2011 (UTC)
- Then, of course, you have to name the hardware it runs on as PC Clones/MacBook Pros. ( Which I am typing this on, running NetBSD! ). See ( and really study ) the UNIX genealogy graphic. Steven Jobs called the Mac Cube a super computer, i.e. just cause the company calls it that, does not make it that.
- IMHO, Mach/NextStep/NuKernel and Mac OS X need to be mentioned as the linage of the most popular UNIX-like or variant as they were called while AT&T/BSD were the most popular.
- Mac OS X was a Mach/Acedemic variant, but continuted development with a new kernel.
- Then you have statements like this:
- "In 1991, Linus Torvalds began work on Linux that runs on IBM PC clones."
- Um. No. Linux began work on a Kernel, and File system, that was open source ( like GNU ), and ran on 386 processors. and was a Unix variant, that was based upon Minux. ( No AT or PC Support ). I read his original Usenet post, found it interesting, and waited until he had English keyboard support.
- But before you go and mention, Linux, it should follow directly with the mention of GNU. Linux would have not gained traction, without GNU. and GNU would have been nothing without the development of the GNU C Compiler GCC, which Linus used. ( Richard Stallman, got a MacArthur grant, Linus, didn't ). I have my GNU 1.0 release Mag tape. You can name this comment as a source. I was there, and saw it happen.
- I would rephrase this as follows, just to reflect actual history, and the most accurate sources:
- "In April of 1991, Linus Torvalds began work on Linux, based upon the Minux sources, that ran on 386 and 486 class processors.'
- Ok, Ill quote sources
"Hello everybody out there using minix -
- Ok, Ill quote sources
I’m doing a (free) operating system (just a hobby, won’t be big and professional like gnu) for 386(486) AT clones. This has been brewing since april, and is starting to get ready. I’d like any feedback on things people like/dislike in minix, as my OS resembles it somewhat (same physical layout of the file-system (due to practical reasons) among other things)." — Preceding unsigned comment added by 71.202.117.96 (talk) 09:38, 2 September 2011 (UTC)
- Aren't 486 computers rather ancient? You could use "Lions' Commentary" like Linus did. Shjacks45 (talk) 23:29, 16 March 2012 (UTC)
Appearance of Linux in the article?
Why is the History of Linux in the article after the discussion of SCO vs Linux? Anyone without any knowledge of Unix would not know what Linux is. --217.204.163.50 11:00, 25 June 2007 (UTC)
Similarly HP-UX, and AIX are both mentioned as doing well before any history of these versions.
- So long as there are links to appropriate articles, the reader can easily find out. It would be inappropriate to insert history before mentions of the various OSes in the introductory section. — DAGwyn 13:35, 25 June 2007 (UTC)
- Why is GNU Linux even included? After all: "Gnu's Not UNIX" equals GNU! Shjacks45 (talk) 23:33, 16 March 2012 (UTC)
No GNU in the 80s?
In the 1980s, there's no mention at all of the 1983 launch of the GNU "Free Unix!" project. I don't know when I might get time to add this, but I thought I'd leave a comment here in the mean time in case anyone else has helpful comments or if someone wants to beat me to it. Gronky 13:43, 25 June 2007 (UTC)
- Free Unixen are not mentioned in the timeline at all; they've been split into their own section. If you want to try to combine these then go ahead; I'm not sure that separating them is necessary. Chris Cunningham 13:59, 25 June 2007 (UTC)
- Hmm. I'm not sure if I'll have time to try merging that section in, but for the short term, I'll add a few mentions soon, maybe with "see below"s. Even if the timeline is to be limited to official UNIX, the work on free Unices is still a part of their environment and history. Gronky 15:35, 25 June 2007 (UTC)
- GNU never really produced an entire Unix system. The project probably deserves mention here but not much more. The GNU article should be linked to. — DAGwyn 14:15, 26 June 2007 (UTC)
- It's irrelevant that GNU failed to produce a running system: GNU software was popular on UNIX systems of all kinds Trashbird1240 (talk) 00:38, 31 August 2008 (UTC)
- A GNU article should definitely be linked to. :P ¦ Reisio 20:42, 26 June 2007 (UTC)
- The wikipedia article on "unix wars" says that "the GNU operating system was made operational with the inclusion of the linux kernel, and countless operating systems, lumped together under the label "GNU/Linux", emerged that were based on this combination." So it is not completely accurate to say that GNU never produced an entire Unix system.
Furthermore, even if you won't accept GNU/linux because it wasn't wholly from the gnu project, there is GNU/Hurd(which uses Hurd, the gnu project's own kernel). Even though it is not classified as "stable" GNU/Hurd is definitely an operational system. In fact I think GNU/Hurd should be added to the Unix evolution chart along with linux. Or maybe have the chart replaced with this more accurate one: http://wiki.riteme.site/wiki/File:Unix_history.svg —Preceding unsigned comment added by 67.243.184.122 (talk) 03:11, 6 September 2010 (UTC)
Stallman was busy fapping his beard in the 80's so GNU never realesed an OS. They stole the Kernel from Linus and now they threat people who don't call it GNU/Linux. Nowadays they spend their full time spreading FUD against Micros$ft and singing the Free Software Song, so they will never have a full OS till the next millenium. --213.138.243.107 (talk) 17:20, 10 September 2010 (UTC)
- The "n" was a typo, fixed now. — DAGwyn 21:11, 26 June 2007 (UTC)
- Well, I wouldnt say it that way, but I agree with the point mostly. GNU/Hurd did distributions that ran, and they ran as UNIX variants. ( and on a 3B2, it was stable. ) I would disagree with that FUD is against Microsoft, I think the river of stuff flows the other way, and as someone who knows or have met most of the principals of the free software song, none of them can sing. GNU didnt steal the kernel, they wrote their own, and Linus borrowed the development tools for Linux from GNU. So there is HURD/GNU and Linux/GNU. — Preceding unsigned comment added by 71.202.117.96 (talk) 09:46, 2 September 2011 (UTC)
Why is GNU included? After all: "Gnu's Not UNIX". In the Stallman/Rashid letters, Stallman indicates HURD isn't usable.Shjacks45 (talk) 04:12, 17 March 2012 (UTC)
Unix history-simple
Is there a graphics representation of UNIX history rather that Unix-simple? Because there is an awful lot of simplification in the simple box, and there are a lot of versions of the Operating system missing eg Digital UNIX both in BSD (OSF/1) and what is now Tru64 UNIX (on its last legs) but it must be more notable than Minix. As is shown in this UNIX tree there were many other UNIX types than is shown in the Unix history-simple diagram.
There are other issues with the current Unix history-simple for example it presents Sun OS and Solaris as different developments, yet the first few generations of Solaris ran BSD device drivers and the emulation of system V drivers and memory allocation were buggy. This meant that one could either use BSD subsystem calls that usually worked as reliably as they did in SunOS or SYSV calls to devices such as STREAMS that weren't even fully implemented. So at the very least there ought to be a link between them.
I think something else that would help is to draw the lines by kernel rather than shells. For example the current Unix history-simple diagram implies that the NeXt system is derived from BSD yet that is misleading as it uses the Mach kernel.
Of course I am not volunteering to do the work. I'm hoping someone else will agree with me and volunteer to do it :-O -- PBS (talk) 09:34, 8 February 2012 (UTC)
- I could quibble with some of the details, but I agree with your main points. I haven't seen a really accurate Unix evolutionary diagram since around 1980, and there have since been too many variants with too much borrowing back and forth among variants to ever come up with a truly satisfactory diagram. I suggest not worrying about it. — DAGwyn (talk) 23:12, 8 February 2012 (UTC)
- Now that Apple's Clang compiler is a part of FreeBSD, I think that there should be a line added from Mac OS X (not sure which version) to FreeBSD 9. — Preceding unsigned comment added by 46.112.124.35 (talk) 10:25, 11 February 2012 (UTC)
- I agree that a graphic diagram would be useful, they're used in Genealogy and org charts. I'd like to note that 4.4 BSD Lite is the UNIX version Apple adapted and that the "MACH Kernel" has been supplanted by a new Kernel of Apples design.
- Hard to think of Kernel as the operating system: operating systems are usually defined as programming models. Anyone who has used Cygwin or other portable unix distributions has to wonder is the kernel the host operating system or not? MACH and HURD certainly have no programming interface for commonly expected operating system presentation or other embedded or user interface functions, however they schedule memory usage and CPU cycles for an operating system to exist. Shjacks45 (talk) 03:45, 18 March 2012 (UTC)
Lions' Commentary
Should have a reference to Lions' Commentary on UNIX 6th Edition, with Source Code by John Lions (1976) which contains the complete source code of the 6th Edition Unix kernel. Undoubtedly used to reverse engineer other versions. Back in 1993 Linux kernel .91 had same software bugs as UNIX v6. Hmm. Shjacks45 (talk) 10:30, 18 March 2012 (UTC)
- So much for previous assertions that Linux is not Unix.. — DAGwyn (talk) 07:56, 22 March 2012 (UTC)
POSIX time of 2^64 -1 equals 15:30:08 Sunday, 04 December 292,277,026,596 UTC?
Over at Talk:Unix time#Year 292.2C277.2C026.2C596 is illegal as fact for encyclopedia another editor has questioned the claim that at 15:30:08 UTC on Sunday, 04 December 292,277,026,596, 64-bit versions of the Unix time stamp will overflow the largest value that can be held in a signed 64-bit number. In particular, he is questioning the conversion from Signed 64-bit time_t = 9,223,372,036,854,775,807 (264-1) to the above date/time. Is there a standard UNIX utility that I can invoke from the shell that will convert a POSIX time of 264-1 to the above date? --Guy Macon (talk) 19:19, 31 January 2012 (UTC)
- It's actually kind of silly to argue about an impossible date, especially when our time units, leap year conventions, etc. are based on certain characteristics of a solar system that will no longer exist at that time. First, that 2^64 should be 2^63. You can readily estimate the year by extrapolation: log(y-1970) = log(263-1) - log(60*60*24*365.2425). Using Windows' Calculator accessory, I got y = 292,277,026,596.927... which is sometime after noon on Dec. 4 in the cited year.
- Presumably, assuming there are no bugs in the program or in the C implementation, you could use the following on a 64-bit Unix platform (on Windows, try running the free Solaris 10 download, plus free Solaris Studio, in a VirtualBox; I'm typing this on a deficient system so I can't try it right now).
/* endoftime64 -- display largest 64-bit representable time on stdout */
#include <stdio.h>
#include <stdlib.h>
#include <time.h>
static void fatal(const char *mess) {
perror(mess);
exit(EXIT_FAILURE);
}
int main(void) {
static const time_t t = (time_t)0x7FFFFFFFFFFFFFFF; // assuming time_t is at least 64 bits wide
struct tm *tmp = gmtime(&t); // this should always work
if (tmp == NULL) fatal("gmtime");
char *buf = asctime(tmp); // watch out: implementation might fail here
if (buf == NULL) fatal("asctime");
if (fputs(buf, stdout) == EOF) fatal("fputs");
return 0;
}
- This program cannot work! It depends on struct tm, that only works until Year 2147485547 due to the fact that struct tm contains a 32 bit int for the year. --Schily (talk) 10:04, 9 February 2012 (UTC)
- What "64-bit Unix platform" are you describing? It would be kind of foolish for it not to have taken advantage of the opportunity to consistently use wider types where appropriate. — DAGwyn (talk) 07:54, 22 March 2012 (UTC)
- I recommend you to read the POSIX standard and BTW: POSIX already lasts until Earth was burnt by Sun. --Schily (talk) 10:23, 22 March 2012 (UTC)
- What "64-bit Unix platform" are you describing? It would be kind of foolish for it not to have taken advantage of the opportunity to consistently use wider types where appropriate. — DAGwyn (talk) 07:54, 22 March 2012 (UTC)
- This program cannot work! It depends on struct tm, that only works until Year 2147485547 due to the fact that struct tm contains a 32 bit int for the year. --Schily (talk) 10:04, 9 February 2012 (UTC)
Readability
I came to this article as a reasonably intelligent, general reader, not a tech expert. I hoped to find some basic information on what UNIX/Unix is. That's the purpose for which many people use a general encyclopedia. This article completely misses that boat. I have now read the entire article as well as the entire talk page, and still know little more than when I started, except that "UNIX" and "Unix" are hotly disputed minor details in an article so filled with tech lingo that it is incomprehensible to the average person. The article needs to be rewritten in language that a layman can understand. Please let me know when that happens because I still want to know some basic information on what UNIX/Unix is. Carmaskid (talk) 01:55, 17 October 2011 (UTC)
- "Unix" denotes a family of operating systems originating at Bell Labs but now widespread and considerably evolved. The best way to gain a feel for what Unix originally was, apart from actually using it for a while, is to read Brian Kernighan's beginner's guide that shipped with early releases of Unix. However, we cannot reprint that here, and it is not in encyclopedic form anyway. I agree that there is too much concentration on "branding" and too little on functional aspects of the (generic) Unix system. If you have specific suggestions concerning what needs to be clarified, post them here, and maybe somebody will accept the challenge. However, it is probably asking too much for a "layman" to really understand any OS technology in any depth based on just an encyclopedia article. — DAGwyn (talk) 05:04, 13 June 2012 (UTC)
Unix TSS and 32v License
Caldera, released the source code of 32-bit 32V UNIX and 16 bit UNIX Versions 1, 2, 3, 4, 5, 6, 7 under a BSD-style license. So does that mean we can move them from mixed/shared source to open source. Can we change the color to green. If not can anyone specifically say why not.
See, the license letter by Caldera here. [3], [4].
And here is an article written by Ian Darwin. An account of why Caldera released the early Unix source code under a BSD license. Published on-line in linuxdevcenter.com in 2002. [5]
Also can some one add OpenIndiana as a successor to OpenSolaris, and also include SGI's IRIX into this tree. --Sithjedi (talk) 09:47, 26 June 2012 (UTC)