Talk:Operating system/Archive 2

From Wikipedia, the free encyclopedia
Jump to: navigation, search
Archive 1 Archive 2 Archive 3

Windows is not a OS, the Microsoft's OS "first name" is DOS.

removing Windows from "OS names" —The preceding unsigned comment was added by Krauss (talkcontribs) .

You're correct that the Windows layer atop DOS wasn't an operating system, but that ship has sailed. Most people are probably now using "Windows" to refer to the new-generation Windows products including Windows/NT, Windows/2000, and Windows/XP, and there, the operating system is most definitely not DOS. I don't think we need to get so pedantic at this point in the article.
Atlant 14:50, 17 April 2006 (UTC)
Because "Windows" usually now means the Windows NT and it's newest versions like XP (NT5.1) and Vista (NT6) we have to tell what it has actually meant and what does it mean now. I have heard that on Vista, Microsoft disconnected the GUI from the OS, but problem is that Windows does not longer mean GUI or the OS. First Microsoft did have only a OS, then OS & GUI and then OS+GUI and now newest Windows NT versions has gone again to OS & GUI but Windows still does mean complete package, the software system, what is believied by wrong reasons to be the OS. And Microsoft sells other products under Windows name too, like Windows Mobile. Microsoft is using very strongly the "Windows" to sell own products. Golftheman (talk) 08:31, 14 August 2008 (UTC)


OK, I agree, it is correct. Only dicussion comments:

  1. Take care on reverting, people working with other contents in the same edit!
  2. My point... the citations on Itroduction is only for "... the most popular names when people talk about OS..."; DOS is a "very popular OS name", can cited, but Windows cause mistakes with the OS/GUI separation, it is treated on the Windows section.
Krauss 18 April 2006

Google say (the 3 most populars OS-names!):

Windows "Operating system":    165,000,000 
Linux "Operating system":      102,000,000
UNIX "Operating system":        63,400,000

"Mac OS"  "Operating system":   26,200,000
DOS "Operating system":         13,300,000 
VMS  "Operating system":         1,800,000

Come on: we can take off "Mac OS".

Two points. "Windows" historically referred to the technology of representing a user interface with segmented boxes which can present disjoint information streams; "XWindows", for example, was named for Xerox, which was innovative in this techology. Microsoft named their OS after this methodology, starting with MSWin 3.11 (a GUI application on top of DOS) and continuting past MSWin 95 (which no longer used DOS as an OS, but only as a command interpreter.) I refer to "MSWin" instead of "Windows" when I talk about the Microsoft brand name for it's operating system family, instead of the technology of windowing itself. This leads to the second point; the article says that
Operating Systems themselves have no user interfaces; the user of an OS is an application, not a person...
While this is appealing, it's not entirely true. Since MSWin 95 Microsoft has integrated the GUI with the OS; for example, you could boot Win3.11 to DOS, without invoking the GUI, but you can not effectively boot contemporary MSWin systems without the GUI. Pete St.John 19:23, 5 June 2007 (UTC)

History and OS/360

The paragraph starts by saying "several major concepts ..." and then mentions the development of OS/360, which strictly speaking is not a concept. Also it mentions hard disks without describing what concept was involved. Also only hard disks are mentioned in relation to OS/360 even though the sentence is introduced with "also" (something earlier apparently got deleted).

I don't know enough about OS/360 to fix this. Anyone? Bueller? Ideogram 15:53, 30 May 2006 (UTC)

Until TSO was introduced, OS/360 had no GUI, no user console whatsoever. It had file ownership, scheduling, protected supervisor mode, protected RAM, multitasking, multiple user scheduling, asynchronous I/O, and a Job Control Language with a "procedure library" which was a collection of macros in modern terminology. Much of the system code was written so that, when in RAM, it could be interrupted by a higher priority task, while the current task was stacked up to resume execution when the high priority task completed. In order to do this, code was written to be "re-entrant", i.e. it had to do make no changes for the current task that would affect the interrupted task. This was a pretty formidable demand, given that the RAM size might be as little as 64K. Albert 20:26 13 Nov 2006

Done. But I'm sure if we really put our minds to it, we can come up with more concepts. Certainly "computer networking" belongs in there somewhere (SNA, Bitnet, UUCP, etc.) Compatible filesystems may belong in there as well (UFS seems like the best example to me of an early, widely-compatible filesystem).
Atlant 16:17, 30 May 2006 (UTC)

OS vs Operating sytem

We should try to agree on some kind of policy for where to use OS or Operating system. Generally I think we should use operating systems where it is plural (OS's seems awkward), and try to use OS elsewhere, except maybe the first usage in a paragraph. Ideogram 16:26, 30 May 2006 (UTC)

GUI's as shells

How is the GUI in Windows NT descended OS's a shell? Isn't the GUI integrated into the kernel there? Ideogram 16:37, 30 May 2006 (UTC)

No, definitely not. The actual Windows/NT operating system is quite small, with all of the other "stuff" implemented outside the kernel including stuff (like the HAL) that we might ordinarily bundle into the definition of OS. I'm not sure there ever really was a bundled GUI/OS, although you might make the claim for MacoS pre-X and I don't know enough about AmigaDOS to really say. Certainly all the workstations (using X/windows) have the UIs (including the GUIs) decoupled from the OS as does MacOS/X as does NT/2K/XP...
Atlant 16:53, 30 May 2006 (UTC)
I could have sworn the GUI was moved into the NT kernel for performance reasons. I'll see if I can dig up a reference. Ideogram 16:55, 30 May 2006 (UTC)
At least on the W2K reference shows that window manager exist in kernel space but not in the kernel itself. And from earlier NT versions, the window manager was moved from the user space and integrated to OS services inside kernel space but not to micro kernel itself. Golftheman (talk) 07:55, 17 July 2008 (UTC)
You may be right, but as your research this, be careful to distinguish "operating in Kernel mode" from actually being a part of the kernel; they're separable concepts. I could imagine for high graphical performance you'd want your windows drivers to have access to the hardware, a privilege often reserved for code operating with at least some kernel permissions.
Atlant 17:04, 30 May 2006 (UTC)
I think there's no question that architecturally the GUI is separate from the rest of the kernel. Whether this constitutes "distributed with tools for programs to display and manage a GUI" is a matter for discussion. Ideogram 17:14, 30 May 2006 (UTC)
What does "from rest of the kernel" mean? That window manager is not part of microkernel (kernel itself) or it does not exist on kernel space, so is the kernel space always a same thing as the kernel? Golftheman (talk) 07:55, 17 July 2008 (UTC)

Actually after reading your changes closely I have no objection to the wording, since you say "sometimes". I might even say "usually". Ideogram 17:18, 30 May 2006 (UTC)

CLI's within windowing OS's

I think there's some question as to whether a CLI within a windowing system is really a "CLI operating system". Ideogram 19:08, 4 June 2006 (UTC)

In unix I start xwindows from the shell prompt, but in NTx I start the "DOS" command from within the "Windows" GUI. In Unix the GUI (X) is an application and a command-line-ineterface is the default acess by the operator; in MSWin (since about Win98) the GUI has been integrated with the OS and the CLI is merely an application. This isn't the greatest way to classify OS's but I'd say that MS/PC DOS, VMS, IBM 360 and System V are "CLI OS" and WinNT, XP etc are "GUI OS". MS is the only purportedly general purpose OS I know that integrates the GUI like that, although I think Mac intends that you not need the CLI and generally boot directly into the GUI. (But in NT you *must* boot directly into the GUI.)Pete St.John 16:17, 5 September 2007 (UTC)

The notion that an operating system includes any kind of user interface is not found in the formal understanding of operating systems taught in computer science curricula. Operating systems classes do not teach UI design; they teach about multitasking, virtual memory, permissions, device drivers, and such.

That said, the academic notion of operating system is not the same as the notion used by companies that have products which they call "operating systems": products such as Mac OS X, Windows XP, or Ubuntu Linux. These "operating systems" do come with user interfaces; in fact, all of them default to a GUI but also offer a CLI. However, from a computer scientist's understanding of what an "operating system" is, the GUI in each of these cases is simply an application that runs on top of the "real" OS. --FOo 03:00, 6 September 2007 (UTC)

The Canonical what use operating system term of ubuntu, just lies for customers so they would buy ubuntu, and this is because microsoft use "operating system" so much and people gets information that they need operating system to computer before they can use it, so competitors make same. It is all marketing and not true. Ubuntu comes with GUI and CLI but those ain't operating system parts, and it is new information what ubuntu users gets when they read books from operating systems (or computer technology at all) or they go to study it that Linux is the operating system and gnome . There is differences on Windows/NT and Linux but that does not mean that distributions like Ubuntu is the OS or it would be different OS than Linux. The _real_ OS definition is what is academic, currently Linux or Windows/NT. And it is the reason why Windows/NT and Linux are two different operating systems while Ubuntu and Fedora are same operating system but different software systems, so called different distributions because of license. Because OS is science, the academic definition is the correct, while the marketing "Ubuntu is different OS than Fedora" is marketing and not right on operating system definition and so on false and misleading what just spread misinformation and makes science to look wrong one. Golftheman (talk) 09:16, 14 August 2008 (UTC)

copyedit request

I have gone over this article very closely. Can you be more specific about what needs to be done? Ideogram 22:19, 5 June 2006 (UTC)

You've done a great (and fast) job. I think you've covered it all. Remove the tag if you think it's good. -- Steven Fisher 22:30, 5 June 2006 (UTC)

This article needs a great deal of expansion

This is not even close to being even considered complete. What about scheduling, memory management, kernel mode vs user mode, processes and threads, a fuller discussion of micro-kernels vs monolithic kernels, etc, etc? - Ta bu shi da yu 14:38, 19 June 2006 (UTC)

Great idea! Go ahead! Ideogram 14:40, 19 June 2006 (UTC)
Yeah, and what about a hospital as an operating system?? --Blainster 21:28, 19 June 2006 (UTC)
And we now have a woefully-inadequate section on security, written purely from the perspective of IP ports and completely ignoring every other aspect of system security.
Atlant 17:57, 29 June 2006 (UTC)
I disagree that the information on kernels, scheduling, memory management, and processes threads, etc. should be expanded. As long as those technical concepts are briefly mentioned here, that is sufficient. Readers with the curiosity to find out what those things are can follow the links. Keep in mind that most people do not find kernels or big-endian v. little-endian to be very interesting (and I have tried to explain the importance of such technical concepts to many acquaintances over the years, with mixed results).
Operating systems are huge, complex beasts. The article should maintain a high-level overview of the subject with sufficient links to in-depth information for those who are interested.
I do agree that the information on security is grossly inadequate and should be expanded. --Coolcaesar 19:32, 29 June 2006 (UTC)
I think that the information about threads at the threads on computer science are more like programming like than the thread manipulation by the os. -- SiegeM 23:25, 09 August 2006 (UTC)
for those topics there's some coverage in the kernel article. I'm not sure what could be included here.--BMF81 22:43, 8 November 2006 (UTC)

Suggestion

It will be a good idea to add about a minimun of 3 pictures, maybe from Windows, Mac OS X and Linux.

chickens

Who vandalised this page with "i like chickens"? WHO? —The preceding unsigned comment was added by 81.76.113.125 (talkcontribs) .

In the opening lines "higher level functions" are not explained. I htink they should be replaced with something more clear.

The article history page says it was 206.207.175.170. Complain to them. ;-)
Atlant 19:11, 29 August 2006 (UTC)

Most popular before Microsoft?

Please, tell me the name of the most popular Operating sysytem before MS-DOS became the most popular. You can answer right here as soon as you can Moscvitch 16:14, 27 October 2006 (UTC)

The world was fragmented.
In the PC space, CP/M was popular, but PCs were still not very common.
Meanwhile, every vendor had their own operating system(s) so MVS was very popular in the IBM world, VMS was very popular in the DEC world, and so on.
I doubt we could figure out (let alone agree upon) which O/S was most popular before MS/DOS.
Atlant 16:18, 27 October 2006 (UTC)
Thanks. Yeah, I mean O/S for IBM-PC/ Moscvitch 16:39, 27 October 2006 (UTC)
IBM PC debuted with PC/DOS (which was actually MS/DOS). So there wasn't any OS for the IBM PC before MS/DOS.
Atlant 16:42, 27 October 2006 (UTC)
I assume that on "Home Computers", it was Commodore's KERNAL and Atari_TOS before QDOS came up. I remember having used GEM on TOS when I got some brand new Intel 80C86 with MS-DOS 2. On huge systems (not specifically meaning mainframes, but systems taking a lot of space ;)), Unix and VMS were quite common. Depending on the price, use cases were very widespread. In Fortran 4 times and with needs of every bit performance, the interest in common operating systems and their clean separation of tasks was not always an issue. But that's only a personal memory. 80.108.61.230 03:13, 20 December 2006 (UTC)

Really sorry

I just want to apologise for my blatant re-write of the operating system definition, i'm just trying to help, but next time i'll do a bit more research before I think about trying it again, sorry again. - Mc hammerutime (talk)

Windows CE (+), a descendant of Windows?

In marketing and some aspects of usability, it's a part of the product family. But technically, it's neither a relative to DOS nor to the NT Kernel. So it could be misleading to imply a relationship besides having the same vendor. It's similar to calling the Apple OS on their iPod being a descendant of Mac OS 80.108.61.230 03:28, 20 December 2006 (UTC)

TSR programs not too easy

The article states: "although DOS itself featured TSR as a very partial and not too easy to use solution" The problem with writing TSR programs was, that it was too easy. From a programmer's view, making a program TSR was easier than forking and deamonizing on a modern Unix system because of its limitations. You didn't have to care about signal handling - and that's the point. They were limited and lacked IPC. But saying it would have been hard is not exactly what the problem was about. What should be said that forking was not possible and the need to handle parallelization for yourself by vector-swapping and interrupt-violation mania. When a TSR compares to MySQL's auto-increment and daemonizing to Oracle's sequences+triggers- which one is easier and which one is more powerful to work with? ;) 80.108.61.230 03:50, 20 December 2006 (UTC)

What are you talking about? Your comment makes little sense —The preceding unsigned comment was added by 58.166.146.197 (talk) 16:09, 18 January 2007 (UTC).

What about the other MAJOR OSs of our time?

Novell? Solaris? Symbion? Windows CE? —The preceding unsigned comment was added by 203.97.20.142 (talk) 22:40, 26 January 2007 (UTC).

Market Share

The market share statistics that are being quoted here apparently come from [[1]]. These statistics are for computers that surf the internet, and do not include web servers, database servers and the like.

The statistics at [[2]] tell quite a different story. In the realm of web servers (computers that run the internet), Apache/Unix machines are 60% of the installed base. There are 63,800,000 such machines. While this article does suggest that the 94% statistic is for desktop computers, there is no mention of Unix's dominance in the web server world.

The mainstream press got it completely wrong, as usual. In this article, [[3]], it is stated that "Windows runs on more than 95 percent of the world's computers." Not possible.

I added a citation for the 94% figure in this article. Maybe someone could put something in the Unix section about web server market share? Robertwharvey 03:18, 31 January 2007 (UTC)

You are right when you say that Internet-connected desktop computer operating system market share quotes are being misused to discuss operating systems in general.
However, you're doing the same thing with the Netcraft numbers. Netcraft tracks particular types of Web servers, not operating systems. Apache runs on a lot of other systems besides Unix; it is erroneous to conclude that the proportion of Apache Web servers is the proportion of Web server systems running Unix. (Notably, Apache can run on Windows.)
What's more, market share is not really a very useful figure for people who aren't in the sales business. For ordinary users, installed base is more interesting, because network effects (like, can you get support for it?) depend more on installed base than on market share. Netcraft tracks installed base, not market share ... but of the wrong thing for this discussion. --FOo 09:42, 31 January 2007 (UTC)
The term "Market Share" here was making me nuts (for reasons similar to the above). So I've written "Note on Market Share" as a subsection under "references". Pete St.John 16:07, 3 July 2007 (UTC)
"Market Share" Tug of War. I see that we have a small edit war, maybe an edit skirmish, over the "market share" tables. What I wonder is what is so attractive about it, and to which demographic of editors. To Computer Scientists and related types (computer hobbyists, software engineers, whateve) Operating System is a scientific and engineering topic and the large amound of space in market share tables is inappropriate...even if it were meaningful data, which is disputed. I'm thinking that to undergraduate business majors, "operating system" is a commercial product category of special significance (like "oil" or "commodities"). I don't think the tug of war will go away until we figure out what's motivating whom, so we can have a dialog with them. Pete St.John 18:30, 23 August 2007 (UTC)

Hi, I was the one of the participants in the tug of war over market share tables. I didn't see the note at the bottom of the page at first, which notes the statistics' dubious reliability. I guess that's okay, so I'll stop reverting it, but I'd really like to see "personal computer" market share specified in the tables as well ... maybe I'll add that later. As for my motivation, well, not only do I think the statistics don't really belong here, but I think that HitsLink statistics in particular are horrendously flawed. In particular, I think that they overcount the number of Apple computers by mistaking Konqueror on Linux for Safari on Mac -- Konqueror has a very significant share on Linux, and if you look at the browser table, Konqueror only shows up with a .01% share, so something's fishy. I have no idea how they get their statistics, and I think that quoting such unreliable statistics doesn't really add to the article. I'd still like to see the tables removed, but I've made that known and spoken my piece, and others obviously disagree about their utility, so I won't take them out anymore. 192.94.94.105 21:21, 23 August 2007 (UTC)

I have to agree. Any way to measure any OS is going to be very unreliable and biased-- if I have my user agent set to Vista IE to access certain pages, even though i'm using firefox on linux, or especially if browsers are misrecognized, how is that fair? In addition, it's biased to the site that measures it-- a site in english might be biased towards newer proprietary OS, such as OSX and Vista, whereas less needy or open source OS such as BSD and Linux might have more usage in countries that don't access that page, or the internet at all. Wikipedia's aim is to provide an encyclopedia that people can use to learn about things, putting up these figures is incredibly misleading and gives the impression that there is a clear cut way to measure OS popularity. Saying that "estimates of users are between 10%*REF* and 50%*REF*, although there is no accurate way to measure market share for OS" in the text might be ok, but having the tables there as if they are a fact is incredibly misleading --lucid 21:31, 23 August 2007 (UTC)

Future of operating systems?

This page would benefit from a section detailing what the future holds for operating systems.

As with most engineering the future looks exceedingly bright due to SCIENCE FICTION writers and movies. We are where we are today with thermo-chemical warfare, political-industrial complexes, popular consumer gadgets and global commerce largely due to national policy makers. So the future of computing will address those aspirations, such as global warming carbon reduction, agro-business famine farming, disaster relief (except the FEMA trailers), cad driven manufacturing and robotics, and new warfare weapons touted to exceed "War of the Worlds": Tasers and lasers, shockingly awful assault weapons, and billions of dollars for munitions, soldiers, and modern war toys. I would imagine that if the USA/UK/AU had not spent trillions in Afghanistan and Iraq (I & II) we would already have green lawn golf courses, mineral mining, and vacation cruises all on the Moon and Mars via Virgin Atlantic. Engineering and robotics must vastly improve before there will be anything close to photon torpedos, warp-speed, and worm-hole travel thanks to Star Trek, Contact, or Stephen Hawking which will hopefully happen much sooner than 75 million years, give or take a few red or blue shifts. :User:bwildasi Fri May 16 20:53:09 UTC 2008 —Preceding comment was added at 23:55, 16 May 2008 (UTC)

Unix based?

"Free software Unix variants, such as Linux and BSD, ..." Linux is not a Unix variant, but I don't know what to put in its place that would maintain the way the Unix-like section of the article without giving linux its own section. Can someone give me advice as to how I could do this or fix it for me? Zarathrustra 18:47, 29 May 2007 (UTC)

I don't mind calling Linux a (family of) unix variants. Actually, Red Hat was able to join the Unix Consortium at one point, so legally, Linux is a Unix. "Unix" had been an AT&T brand name, and "Linux" is a brand name that started with Linus Torvalds, who wrote a kernel for unix to port it to the Intel platform. All of the Open Source Foundation, Torvalds, GNU, Berkeley, numerous authors sharing source code, and AT&T (which gave a source license freely to Berkeley) deserve credit for making unix accessible to the public at large, enabling cooperative development across platforms, continents, and generations. Pete St.John 19:32, 5 June 2007 (UTC)
Linux was one of many ports of UNIX to the PC and was always describe as such until it became legally necessary to claim it is a completely different OS. —Preceding unsigned comment added by DonPMitchell (talkcontribs) 03:07, 23 January 2008 (UTC)

External Security subtopic Network Address

I've added the hyperlink for network address, but don't know what would be a good topic to link it to. It goes to a disambiguation page. Any help is appreciated. Thanks. --Bookinvestor 21:16, 5 June 2007 (UTC)

Is TRON really important?

I find it difficult to believe that the TRON Project deserves to be mentioned alongside Unix/Linux, Windows, and Mac OS X as a major operating system with its own heading. The TRON section is pasted from its Wikipedia article, a bunch of the links are broken or go to outdated, abandoned-looking pages. Can anybody more qualified than me verify that TRON is important enough to merit an entire section of its own in the OS article? Even if it does, it probably shouldn't be the same text simply pasted from the opening of its own article. --Skyfaller 22:26, 15 July 2007 (UTC)

I agree. I've removed that section. There's about a dozen other OSes that are much more notable than it that aren't even mentioned in the article much less have their own section. --Android Mouse 19:40, 11 August 2007 (UTC)

Spam - 'Iranian OS'

In the first paragraph, it says:

Windows, Linux,Iranian OS ( Code name SAM ) , and Mac OS are some of the most popular OSes.

Iranian OS ( Code name SAM ) should be removed, not being a popular operating system

Never heard of it, and the contributor was anonymous, so yeah, I revereted it. Please feel free to revert stuff that looks patently self-serving (in this case, nationalistic). The burden is on that fellow to cite a reference to the significance of the thing. Pete St.John 17:02, 9 August 2007 (UTC)

OS and memory

Since when do OSes manage registers or CPU cache? Compilers do, but AFAIK, OSes don't. They do need to know about 'em in order to perform context switches (multitasking) but in general, I don't think they make the decisions WRT whether to use registers, cache or RAM. --Elvey 22:35, 22 August 2007 (UTC)

Processors can do alot of stuff autonymously, but whatever a processor can be told to do, is defined exactly by it's assembly language. Specifying a register is typical of assembly languages. The C compiler let's the programmer *request* a (non-specific) register, and FORTRAN does not, but either compiler will let you link assembler; so any program (including the OS itself) can in principle micromanage registers in the processor (through native assembler macros or libraries). If you wanted to write a small linux kernel to optimize the use of cache for a beowulf, please, feel free :-) Pete St.John 18:12, 24 August 2007 (UTC)
That doesn't answer the question. The answer is: since never. The first two sentences of the "Memory management" section are simply wrong. The whole section is poorly written. KenshinWithNoise 03:39, 24 March 2008 (UTC) —Preceding unsigned comment added by KenshinWithNoise (talkcontribs)