Talk:Comparison of operating systems/Archive 1

From Wikipedia, the free encyclopedia
Jump to: navigation, search

FAT16 and FAT32

FAT16 and FAT32 are very much a Windows/DOS terminology, so simply indicating "yes" for other platforms is misleading. These platforms have equivalent, but not identical file systems. As it stands the table seems to suggest that they all run the exact same file system, which just ain't so.Graham 09:18, 18 Sep 2004 (UTC)

On "n-bit" filesystems

The article should try to be clear about what it means by 16-bit, 32-bit, or 64-bit filesystems. For the FAT16 and FAT32 filesystems this seems to be simply the bit-width of the cluster number. For the inode-based UFS-like filesystems, I'd guess 32 vs. 64 bits would either be the maximum useful offset size within a file, with the acid test being whether a single file can exceed 2GB. Or, maybe it would refer to the maximum number of clusters in a single partition, with the 32-bit cutoff being in the low TB depending on cluster size. (I've never used any UFS-like filesystem that could be called 16-bit, but Sun recently announced a 128-bit version[1].) If these are vague marketing terms that just don't have the precision I'm demanding of them, then IMO the article should come out and say so (or better yet, should just use more precise terms like "maximum file size" if that's what's intended). -- Saucepan 08:01, 21 Sep 2004 (UTC)

Improving rapidly

The article is improving far more rapidly than I ever expected it to, and so far has utterly failed to attract religious warriors as I had feared. I withdraw my earlier misgivings. Looks like Wikipedia wins again. Saucepan 03:24, 23 Sep 2004 (UTC)

Yep, it's dragging itself up!! I have a few small issues with the Mac entries (non-religious, I assure you). The kernel type - I added that OS X was based on Mach, is that monolithic or microkernel? What was wrong with a link to the Mach article? For the classic Mac OS, there are some problems there because there is some ambiguity over which version is being discussed, if any - it could be that the entire line of classic Mac OS is meant. The introduction date indicates we are talking about the entire line (which I agree with really, since it has been superseded), in which case the kernel type is "none", because until version 8.6 or so there wasn't a kernel, and even then it wasn't a kernel in the usually accepted sense of the word.Graham 07:02, 23 Sep 2004 (UTC)
Well I too thought Darwin had a Mach microkernel, with VFS, UFS, networking and other kernel services taken from FreeBSD 4.x, and most user-mode utilities taken from NetBSD. But it looks like Dysprosia is absolutely right in that XNU is based on Mach but isn't being used as a microkernel. OS X does unfortunately seem to have more technical misinformation swirling around it than any other OS: I had a heck of a time finding out if Mac OS X Server's journaling feature works with UFS, or if Apple has yet merged in the UFS soft updates feature from FreeBSD 5.x. The best answers I could find after half an hour of searching were "maybe" and "probably not" respectively, and the pages on Apple's own site were among the worst offenders as far as contradicting each other. Now I'm wondering what else I thought I knew about Mac OS X that is completely wrong.
As for kernels, most of the definitions kicking around seem pretty agnostic about whether OS code needs to reside in protected memory in order to be called a kernel. I'd tend to agree with the definitions that call the portion of the OS that stays resident and manages access to the hardware the "kernel," regardless of whether other hardware-dependent features like memory protection or virtual memory are present. Saucepan 07:52, 23 Sep 2004 (UTC)
The journaling feature is implemented in the VFS layer on OSX. Although Apple only added journalling support to HFS+, in theory any filesystem on OSX could make use of it. [2]
I know this is irrelevant to the article but any excuse to talk about filesystems...  :-) AlistairMcMillan 20:28, 23 Sep 2004 (UTC)


I changed FreeBSD section:

  1. stable version is 4.10. Next stable version will be 5.3.
  2. default file system is UFS (in 5.3 it will be UFS2)
  3. there is by definition no default Window Manager, thus changed question mark to N/A (not applicable). Przepla 21:54, 23 Sep 2004 (UTC)
If you install X, what is the default window manager used? Dysprosia 22:15, 23 Sep 2004 (UTC)
Last time I did this in FreeBSD it was the XFree86 (well, startx) default of twm, unless/until you installed something else. But I'm pretty sure I remember sysinstall making it just as easy to pick a GNOME or KDE desktop during the install as a headless system, so I'm not really sure if it's meaningful to speak of a default desktop in this case. Saucepan 22:49, 23 Sep 2004 (UTC)


I added the Solaris public release date as the release date of the first version of SunOS. SunOS was a BSDsh variant, which was used by Sun for several years until they switched to the AT&T based Solaris. Should the release date for Solaris reflect the Solaris or the SunOS release? ElBenevolente 08:44, 26 Sep 2004 (UTC)

Using the Windows XP/2k columns as a guide, I'd guess the date that should be used would be the first time the code was released as "Solaris". Saucepan 08:48, 26 Sep 2004 (UTC)
Then we'll have to fix that date. ElBenevolente 08:51, 26 Sep 2004 (UTC)
Done. ElBenevolente 08:56, 26 Sep 2004 (UTC)

OpenBSD creator

The OpenBSD column credits Theo DeRaadt as the creator of OpenBSD. If we want to be consistent, we should treat OpenBSD as we treat FreeBSD and attribute the creation of OpenBSD to the University of California. ElBenevolente 08:53, 26 Sep 2004 (UTC)

well, Theo was the one to start the fork from NetBSD. I dont know the circumstances around FreeBSD's genesis, whether they forked 386BSD or merely inherited the project, though... Dysprosia 09:05, 26 Sep 2004 (UTC)
To maintain consistency with the FreeBSD entry, I changed the creator of OpenBSD to University of California. If someone wants to research who forked FreeBSD from 386BSD, they can change both entries to maintain consistency.
Yeah, that's kind of inconsistent, isn't it. I did some looking into what the organizations are called and found that the FreeBSD project refers to itself on its web site and in its copyright statements as "The FreeBSD Project" while the OpenBSD project refers to itself as "OpenBSD" or "the OpenBSD project". I've changed the table accordingly. Saucepan 09:12, 26 Sep 2004 (UTC)
Good call. ElBenevolente 09:12, 26 Sep 2004 (UTC)

Mac OS and Lisa OS

Hard as it is to believe, it looks like Lisa OS may not have shared any significant code with early Mac OS in spite of the obvious resemblance. It'd be nice to find out for sure, if anyone can find an authoritative reference one way or another. (Or we could add a row like "Influences" which could include Lisa OS for Mac OS, Mac OS for Windows, Mach for FreeBSD and OpenBSD, and so on.) Saucepan 17:39, 26 Sep 2004 (UTC) Of course they were developed kind of simultaneously (although the Lisa popped out first), so you would expect some crossover. But I don't think you could say that Mac inhertied from Lisa, in the same sense as OpenBSD inherited from NetBSD or WindowsXP inherited from Windows2000. AlistairMcMillan 19:08, 26 Sep 2004 (UTC)
BTW FreeBSD and OpenBSD and Mach all share an ancestor in 4.xBSD. I don't think either of the current BSDs inherited from Mach though. [3] AlistairMcMillan 19:29, 26 Sep 2004 (UTC)
There was lots of back-and-forth, but I think 4.4BSD borrowed from Mach more recently than Mach borrowed from BSD. An ex-coworker walked away with my copy of The Book so I can't quote chapter and verse, but I'm sure I remember (and a search appears to confirm[4]) that the 4.4BSD VM was based on Mach (the mmap semantics, the unification of file buffers and VM pages, and the vm_map structure[5], among other things). But I guess this is getting a bit too nitpicky -- even for this article -- so I'll stop now.  :) Saucepan 21:24, 26 Sep 2004 (UTC)

Your right! I have my copy of The Book right here. I'd never picked up on line of inheritance that before. Although (and I hope I'm not being naive here) you could make a case that the focus of Mach development now rests at Apple with Tevanian amongst others. So the inheritance from BSD continues today. AlistairMcMillan 22:01, 26 Sep 2004 (UTC)

Missing info including Linux, RTOS

By the way, Linux should be here. Also, what about the various popular RTOS and embedded OS that are very widely used (but not necessarily very visible). I'm wondering if this page is likely to have much value within WP, this info might be better obtained elsewhere. Graham 09:24, 18 Sep 2004 (UTC)

I assume this will be expanded significantly in the future... this appears to be a very depressing comparison (you might as well put it under "file-systems of 5 select operating systems" since that's pretty much the only significant info it has). --Ctachme 20:29, 18 Sep 2004 (UTC)

If it isn't expanded significantly in a couple of months, you can put it on vfd as "non-encyclopedic and not likely to expand in the future".
It is encyclopedic, it's a table of facts like in other print encyclopedias they have tables of numbers and things... there already had been a simlar comparison article which was VFD, but after work on it... I prove all the opposition all all (at time of writting). Before making such comments, gets your facts right... squash 05:24, Sep 21, 2004 (UTC)

Further ideas to expand the table

Possible new columns: Windows Me; GNU Hurd; VxWorks; QNX; Minix; NetWare; EROS[6]; Palm OS; Symbian; IOS; ProDOS

Possible new rows:

  • theme support (yes / no / optional);
  • virtual memory support;
  • Integral "start" button (yes / no / optional / nonstandard position or behavior);
  • filesystem metadata updates (sync / ascync / soft updates / journaled / log-based / not applicable);
  • persistence (application-managed / continuous-automatic);
  • TCP fingerprint;
  • command line completion (yes / no / not applicable);
  • stackable virtual filesystems (yes / no / not applicable);
  • concurrency models (none / process / thread / co-op / state machine);
  • thread models (user / kernel / hybrid / not applicable);
  • accessibility features (left-handed mouse / screen reader);
  • kernel isolation (none / monolithic kernel / modular kernel / unprotected microkernel / protected microkernel);
  • default desktop background has clouds on it (yes / no);
  • Major ABIs supported (MS-DOS / Win16 / Win32 / Linux / iBCS2 / Mac OS / other) (integral vs. optional, level of support);
  • Major APIs supported (Win32 / POSIX / GNOME / Carbon / other); Integral web browser;
  • capabilities-based
  • source code availability (included / extra charge / NDA / freely redistributable)
  • security (hardening / code audits / services running by default / firewall support / event auditing)
  • networking (protocols supported / benchmark performance / features like T/TCP)

..and lots more, but I'm getting sleepy. To be honest I'm a bit skeptical that a single overview article will ever be able to provide much in the way of useful comparison info. The topic is just too broad, and I expect the result would be like a single chart comparing motor vehicles with vastly differing requirements, from scooters to tanker trucks. Restricting the comparison to "viable" or "mainstream" desktop OSes would inevitably invite argument from partisans about how unfair it is to exclude OS/2 or CoCo OS-9 or whatever.

But I do like the idea of a table comparing obscure technical features from an eclectic variety of OSes, if for no other reason than it would be a great jumping-off point for people interested in exploring the subject. Heck, it would be glorious for the sheer geekery alone! Anyone else want to brainstorm? -- Saucepan 08:22, 21 Sep 2004 (UTC)

I agree with much of this. As it stands the article has limited merit, but it could become useful. One thing that is offputting is the table editing though, not that I can think of any better way to do it. It's very easy to lose track of which row/column you're editing and with these extra OSes added it would become even harder. So I'm not saying it shouldn't be done, but that I'm a tad reluctant because of the work involved! Another row we need if these others are added are its general application area - RTOS/embedded/desktop/palmtop/server, etc.. otherwise it will be too easy to start comparing apples and oranges.Graham 00:03, 22 Sep 2004 (UTC)
Nice work, Saucepan... squash 23:13, Sep 22, 2004 (UTC)

I've added Symbian OS to the list, as it's comparable in signficance to Palm OS and definitely more widely used than, say, Minix or EROS. I'd include a couple of the commercial Unices, OS/2, BeOS/Zeta/Haiku, RiscOS, and AmigaOS ahead of those two as well, since they each have both noticeable installed bases and ongoing development of some sort. If we're going to include ProDOS, we should include MS-DOS and other obsolete OSes (which is to say: "Forget ProDOS").

The list of possible new rows is definitely over-ambitious. Command-line completion, for example, isn't even an OS feature; it's a command-shell feature. And frankly, not very signficant. Likewise, accessibility features has little to do with the OS itself. Source code availability is implicit in the licensing. And desktop has clouds demonstrates how silly most of those suggestions are.

I have some ideas for operating systems:

  • Add info for Red Hat Enterprise Linux and merge Fedora Core and RHEL into a single column (differences between the two would be contained within one column)
  • Add Windows Me or another Windows 9x OS to show differences between it and the Windows NT line (Mac OS Classic is there, why not windows 9x?)
  • Possibly add MS-DOS and/or FreeDOS
  • Info on Windows Mobile, Palm OS, and other mobile operating systems
  • Mandrake may or may not be worthy of being added

That's all I can think of right now. --Evice 05:42, Dec 29, 2004 (UTC)


Should NetBSD be included on the list? OpenBSD is listed, and NetBSD is derived from OpenBSD. ElBenevolente 08:44, 26 Sep 2004 (UTC)

If you feel strongly that a particular OS deserves a place in the table then by all means add it. But just keep in mind that there are about 200 other OSes in List of operating systems (including 4 BSDs and a dozen versions of Windows, and not including the hundreds of Linux distros), and that each new column that gets added to the comparison makes it harder to argue against adding in the next one. Saucepan 08:56, 26 Sep 2004 (UTC)
It looks like this will ultimately come down to as issue about which operating systems to include. I only brought this up because OpenBSD forked from NetBSD, and NetBSD is a large product that is compatible with dozens of architectures.
In the long run, I see the possibility that this page will have far too many columns. ElBenevolente 09:00, 26 Sep 2004 (UTC)
Yes, things could get ugly indeed if the current approach is continued forever. Hopefully before then meta:Wikidata will be online and we'll be able to move this kind of info there, and generate product comparison charts dynamically showing whatever columns one wants. Saucepan 09:31, 26 Sep 2004 (UTC)

Mac ignorance

I've corrected a number of Mac OS errors. First, Carbon is the preferred API for mac OS 9, because by adopting it you get to run on OS X too, which is what Apple really want. Carbon is supported back to, I believe, OS 8.6. Carbon is derived from the classic APIs that existed on the Mac since 1984, though significantly extended over the years. Carbon itself adds a few new things, but essentially it's the Mac OS classic API. It's fully supported on OS X, and is NOT deprecated in favour of Cocoa - they sit in slightly different areas of the system and represent two different programming paradigms - carbon is mostly a flat "traditional" API (though with newer parts adopting object oriented structures), whereas Cocoa is higher level, fully OO. While I'm sure Apple would, in an ideal world, prefer to have dropped Carbon in favour of Cocoa only, it isn't going to happen because of the very large power wielded by certain corporations. Having put Carbon into OS X, it's become very entrenched, and could not be removed without a significant redesign of the OS. Cocoa now relies on it for much of its lower levels. Installer - Mac OS 9 and earlier does have an official installer, Apple Installer. It's been around since the System 6 days at least. I never saw that much software use it apart from Apple's own, possibly because setting it up was a very difficult job. Most third party software that needs an installer uses Installer VISE from Aladdin.Graham 03:14, 27 Sep 2004 (UTC)

Is it called Carbon in OS 9 and before, however? Dysprosia 03:20, 27 Sep 2004 (UTC)
Yes, the OS 8.6 -> 9.2.2 implementation resides in a library called CarbonLib. Programmers have called it Carbon since it was first announced. The pre-carbon APIs however, do not have a name that I know of - just "Mac toolbox" or somesuch. Again this goes back to whether the classic Mac OS column is referring to the specific version - 9.2, or to the whole line going back to the beginning. However, since Carbon largely embodies the original APIs, even referring to the whole line's APIs as "carbon" is not as wrong as it might seem! Graham 03:27, 27 Sep 2004 (UTC)
Ok, thanks. Sorry bout the mixup. Dysprosia 03:34, 27 Sep 2004 (UTC)

Could someone point me to the discussion that ends with OS X deriving from OPENSTEP rather than FreeBSD? Somegeek 09:20, 2005 Feb 1 (UTC)

Well, I would if I could... except that OS X is not derived from one OR the other, but from both. FreeBSD is the basis for the kernel at the heart of the system, and is codenamed Darwin. This is an open-source kernel/OS. OPENSTEP is the thing that sits on top of that to provide the application programming interfaces, I/O driver layer, Quartz and so forth. OS X as a whole is a marriage of the two - you can't have one without the other. You could boot a Mac or other hardware using the Darwin kernel alone, but you wouldn't have OS X, you'd have Unix, basically. Graham 22:40, 1 Feb 2005 (UTC)
Thanks for the explanation, that jives with what I have since learned. I'd like to add FreeBSD as "a" predesessor to OS X. I notice that FreeBSD's /usr/share/misc/bsd-family-tree claims that pre-10.0 Darwin Mac OS X included code from FreeBSD 3.2, Rhapsody, and NetBSD 1.4, but I guess NetBSD will have to wait for the next fanboy to come along.  ;-)

Linux licenses

The Fedora Core project is released under the GPL license. [7]. Same for SUSE. [8] Some specific components (for example the xorg fork of XFree86) are released under the BSD license. AlistairMcMillan 11:33, 29 Sep 2004 (UTC)

I was thinking about changing it to "GPL and compatible", but what about, for example, binary drivers that may be shipped with Linux distributions and not be GPL'd? Is this the case for Fedora or Suse? Dysprosia 12:16, 29 Sep 2004 (UTC)
Just change the row to "Primary license," IMO. Lots of free distros include bundled software (I remember when even some of the most idealistic of the GPL-based Linux distros bundled the still-closed-source Netscape Navigator since at the time that was all there was), but it's a judgement call whether you want to consider this bundled stuff part of the OS. Saucepan 14:28, 29 Sep 2004 (UTC)
If we are going to be that obsessive about it, then we better include the BSD license under Windows XP and Windows 2000. I say keep it as it is, the vast majority of a Linux operating system (kernel, libraries, compiler etc) is under the GPL library. The largest component that comes to mind that isn't is whichever version of XFree86 which is under the XFree license (doh, just realised I put BSD by mistake on the page). AlistairMcMillan 15:06, 29 Sep 2004 (UTC)
What you're doing is highly confusing, the fedora project uses the GPL for its own stuff, while bundling vairious other software in its distribution. However fedora is primeraly a distribution so most of the stuff they distribute is not made by them save for some system management shellscripts and other distribution related stuff.
Saying that Fedora Core is under the GPL is a wrong statement of extreme inaccuracy, it is under various other licences including the apache licence, x11, gpl, lgpl, ...
Furthermore what do you mean by BSD and compatable [Free/Open]BSD? have you even looked at the BSD licence (almost) everything is compatable with it so this statement is too vague to be of any use.
You may be right that it is better to change it to something like Primary licence - then you'll be faced with another problem however: what do you define as primary?
As the page stands now i'm changing it back.
And what BSD licenced software in windows? There is no BSD licenced software in windows, what software exactly are you referring to? -- Ævar Arnfjörð Bjarmason 16:24, 2004 Sep 29 (UTC)
Run strings against "ftp.exe" in Windows XP. You'll get a bunch of stuff, but the second last string is "@(#) Copyright (c) 1983 The Regents of the University of California." Same for nslookup.exe except different dates. As I understand it a bunch of the little network related command line utilities are derived from BSD. AlistairMcMillan 16:48, 29 Sep 2004 (UTC)
So? it is still not under the BSD licence, it _used_ to be BSD licenced, since the BSD licence is not viral Microsoft can use BSD code, however that does not mean that it's still BSD licenced. -- Ævar Arnfjörð Bjarmason 17:01, 2004 Sep 29 (UTC)
Sorry but that isn't how the license works. You can take the code and do whatever you want with it, but you cannot just decide to take the license off. As you can see because the license is still there in the code. When people say the license is "not viral" they mean putting BSD code inside another application does not mean the application as a whole is now under the BSD license, which is what happens with the GPL. Which is why they created the LGPL. AlistairMcMillan 17:21, 29 Sep 2004 (UTC)
Granted. -- Ævar Arnfjörð Bjarmason 20:47, 2004 Sep 29 (UTC)
Anyway my point was that the GPL covered code makes up the VAST majority of any Linux distribution, not ALL but most. So although it is a generalisation we should just say GPL. If we are going to obsess about every single license, then we may as well just put "various" under every operating system, since even Microsoft uses open source code (however minor that use may be). AlistairMcMillan 16:48, 29 Sep 2004 (UTC)
See above, they dont. -- Ævar Arnfjörð Bjarmason 17:01, 2004 Sep 29 (UTC)
See above, they do. AlistairMcMillan 17:21, 29 Sep 2004 (UTC) AlistairMcMillan 16:48, 29 Sep 2004 (UTC)
BTW Did you read the two sites I linked above. From Fedora...
Fedora™ consists of hundreds of software modules, some developed by Red Hat and many developed by other members of the open source community. Those authors hold the copyrights in the modules or code they developed. At the same time, the combined body of work that constitutes Fedora™ is a collective work which has been organized by the Fedora™ Project, and the Fedora Project holds the copyright in that collective work. The Fedora Project then permits others to copy, modify and redistribute the collective work. To grant this permission the Fedora Project usually uses the GNU General Public License ("GPL") version 2 and the Fedora Project's own End User License Agreement. AlistairMcMillan 16:48, 29 Sep 2004 (UTC)
See above, they use the GPL for _their own stuff_, just how much is that? The translation of the installer? Some management shellscripts? Certanly nothing more than 1% of the software. -- Ævar Arnfjörð Bjarmason 17:01, 2004 Sep 29 (UTC)
Read what they are saying. "Fedora is a collective work" "Fedora Project then permits others to copy, modify and redistribute the collective work" "To grant this permission the Fedora Project usually uses the GNU General Public License ("GPL") version 2 and the Fedora Project's own End User License Agreement." They are talking about "the collective work" not just their work. AlistairMcMillan 17:21, 29 Sep 2004 (UTC)
From SUSE...
The Linux kernel and the major part of all Linux applications are subject to the "GPL", a license that guarantees free availability and the disclosure of the source code.
AlistairMcMillan 16:48, 29 Sep 2004 (UTC)
SuSE is overgeneralizing, just because they make wrong statements doesnt mean that we should. -- Ævar Arnfjörð Bjarmason 17:01, 2004 Sep 29 (UTC)
I agree they are overgeneralising. That doesn't change the fact that the VAST majority of a Linux distribution is covered by the GPL. Not Apache. Not XFree. Not the binary-only drivers that you mentioned. But the vast majority.
Look at the Cost part of the table. You can undoubtedly get Windows or MacOS cheaper elsewhere on Amazon, or from some special offer on Apple's site or wherever. But we have the price that people pay the majority of the time. Not ALL the time but the majority of the time. We don't have to be absolutely 100% specific with the licenses either, because we'd either end up with a massive unwieldy table or just "various" under each OS. Just enter the license that the majority of the OS comes under. AlistairMcMillan 17:21, 29 Sep 2004 (UTC)
I never disputed that the majority was, even if just one line of code was under another licence it would still be wrong to say that it is under Licence Foo while some of it is under licence Bar.
People will assume that the price in dollars is an approximation, however they will not assume that the licence information is an approximation especially if you write GPL & X11 which relays the false information that they only use those two.
I belive the best thing is to write "Various free software licences" for fedora and "Various free software licences as well as proprietery licences" for SuSE. -- Ævar Arnfjörð Bjarmason 20:47, 2004 Sep 29 (UTC)
Re "BSD and compatible", BSD distributions can't package with it GPL software because that breaks the BSD and GPL licenses, which is why BSD has to be released with a BSD license and only BSD licensed or compatible licensed software can be released with the main distribution. It doesn't restrict users from installing/downloading GPL licensed software themselves, but we are talking about the baseline distribution. So it's not quite accurate to say "preferred" license for the BSDs... Dysprosia 22:41, 29 Sep 2004 (UTC)
That is just plain wrong, you can package them togather just not link from the BSD code to the GPL code, you may however link from GPL code to BSD code. How else would you explain these operating systems being packaged with GCC for instance? -- Ævar Arnfjörð Bjarmason 00:05, 2004 Sep 30 (UTC)
I must be thinking of the kernel. Dysprosia 02:03, 30 Sep 2004 (UTC)
There is no special case for kernels in either licence, they just talk of linking in general. -- Ævar Arnfjörð Bjarmason 03:15, 2004 Oct 1 (UTC)
There's been a contrib/ directory in the BSD source tree since possibly the 4.4BSD days (I remember it was in BSD/OS from the start, and it's present in FreeBSD today). It contains software with various licenses, including the GPL[9]. Saucepan 23:12, 29 Sep 2004 (UTC)
See also GPL "mere aggregation"[10] and the acceptable licenses listed at [11]. Saucepan 23:25, 29 Sep 2004 (UTC)

Features Section

The current features section is confusing, and doesn't work at all for Mac OS X and Mac OS. For one, it implies that the Finder is a Desktop Environment, when it reality it is just a filesystem browser application. Also, why are the default theme category and the window manager category the same thing? I would edit the page directly, but frankly, I know so little about the technical details I couldn't do a decent job. Here is what I'm suggesting for now:

Operating System: Windows XP Windows 2000 Mac OS X Mac OS Fedora Core (Linux) SUSE Linux FreeBSD OpenBSD Solaris
Graphics software and capabilites3 kernel based kernel based Quartz (based on PDF) kernel based User program:
X Window System
User program:
X Window System
User program:
X Window System
User program:
X Window System
User program:
X Window System
Default Filesystem Browser Windows Explorer Windows Explorer Macintosh Finder Macintosh Finder Nautilus Konqueror N/A N/A CDE file manager
or Nautilus
Default desktop environment Windows Explorer Windows Explorer Macintosh Finder & Dock Macintosh Finder GNOME KDE N/A N/A CDE or GNOME
Default window manager Integrated (win32k.sys) Integrated (win32k.sys) Quartz Compositor  ? Metacity kwin twm
(not installing GNOME/KDE at install)
fvwm dtwm (for CDE), Metacity for GNOME
GUI style Luna-styled interface Classic-styled interface Aqua Platinum Bluecurve theme plastik theme  ?  ?  ?
On the subject of Windows Explorer and Finder, they are a wee bit more than just file browsers. They maintain the desktop (which okay is itself just a directory/folder with files in it), but the Windows Explorer also maintains the Start Menu. Basically on Windows the whole interface you are presented with is basically maintained by Windows Explorer, on the Mac it is slightly different with the Dock being a separate application and the menu at the top of the screen (as far as I understand) being maintained by some combination of the WindowServer and whichever application currently has focus.
I agree with you on the Window Manager/Theme split though. AlistairMcMillan 21:01, 1 Oct 2004 (UTC)
The current one is also wrong, the desktop enviroment in OSX is finder+dock+aqua+graphic api's ( cocoa,carbon ) and probably something more. -- Ævar Arnfjörð Bjarmason 19:55, 2004 Oct 2 (UTC)
That is taking things a little far. (1) It's a bad idea to put Aqua in because Aqua is an idea or design or set of guidelines (or whatever you want to call it) that is implemented by the WindowServer (which we already mention). There is no specific piece of software called Aqua. (2) Cocoa and Carbon are sets of APIs (which we already mention), they are not actual programs that display windows or icons or whatever?
Going from the Desktop environment page, the Windows are draw by WindowServer, the Icons are draw by the Finder, the Menus are drawn by a combination of the WindowServer and whichever program holds focus and the Pointer is drawn (I think) by the WindowServer. Also the Dock is a separate program called Dock. Drag and drop is performed by the Finder. Toolbars are drawn by the WindowServer. I think we have everything covered. AlistairMcMillan 20:33, 2 Oct 2004 (UTC)
On Finder and Windows Explorer, they are still file browsers, even though they do other things. That's like saying that Safari shouldn't be listed under a Web Browser column because it is a RSS reader too. On OS X's Desktop environment, that's why i said Standard Mac OS X, because it DOES include all those things. --Ctachme 21:36, 9 Oct 2004 (UTC)
Sorry I wasn't being clear. I wasn't saying that we shouldn't have them down as file browsers. I was saying that we should ALSO have them down under the "Desktop Environment" bit as well. Instead of just saying "Standard Windows" or "Standard Mac OS X", because that doesn't really tell anyone anything.
I've amended your table, with what I was trying to suggest. AlistairMcMillan 02:11, 10 Oct 2004 (UTC)


Quartz is the graphics layer that sits on top of the Darwin core of Mac OS X, (see Quartz (graphics layer))

suggests Quartz is a layer that is not in the kernel, and is thus a user-level program. Dysprosia 02:09, 11 Oct 2004 (UTC)

I think that the way it is now is pretty good, don't you? --Ctachme 02:55, 11 Oct 2004 (UTC)

Desktop Enviornments

I've been thinking and I don't really think it is appropriate to call the Desktop Environment of Mac OS X Finder and Dock. I could quit the finder and dock, and everything still works fine. The dock is nothing more than an application switcher and the finder is just a file browser. I've always thought that the DE is more than just the application switcher and file browser. That's why I said "Standard Mac OS X" in my original table, because there are so many aspects to the DE, it isn't just the finder and dock. --Ctachme 02:55, 11 Oct 2004 (UTC)

Do you have a machine running Mac OS X handy? Try quitting the Finder and see what happens. No more desktop. The Finder is more than just a file browser.
If you want to change DE that much, then go ahead. Just please (I'm begging) don't just put "Standard Mac OS X" because that doesn't tell anyone anything. Since we are supposed to be comparing the standard components in a default install we could put "Standard Windows" or "Standard Mac OS X" or "Standard FreeBSD" or "Standard X" under every component of every operating system. If we are going to go to the trouble of building this table we have to be more explicit than saying the standard filesystem in FreeBSD is "Standard FreeBSD filesystem" or the standard desktop environment in Mac OS X is "Standard desktop environment used by Mac OS X". AlistairMcMillan 03:12, 11 Oct 2004 (UTC)

Split tables

I just noticed that Ctachme has split and rotated the table. I didn't like it when this happened on the Comparison of web browsers page and I don't like it here. I think it makes the page more confusing for readers not less. It may appear simpler to editors who see this page regularly, but I think people who come here simply to read the content would find a single table (even if it does spill out of a browser window and require scrolling) easier to read. AlistairMcMillan 04:57, 18 Dec 2004 (UTC)

Ctachme also removed the helper comments, which is very annoying. I don't mind the split, but the rotation is confusing also. I'd also like them rotated back in the logical fashion, with the helper comments back. Dysprosia 08:50, 18 Dec 2004 (UTC)
The reason I rotated the tables was because otherwise they would not fit onto the screen. As per wikipedia guidelines, content should be geared, when possible, to fit within a reasonable screen width. If the tables were re-rotated, they would certainly be too wide to fit on a screen (and my screen is 1024 wide... the most popular screen width). Additionally, it really isn't any more confusing this way than it was the other way, except you read across instead of reading down (which actually, now that I think about it, is probably better this way since most english speakers read from left to right, not top to down). The reason I removed helper comments is because the name of the OS is right there are the top of what you are seeing. You *do* have to glance up a couple of lines, but it really isn't all that inconvenient. And, like I said, this method is no more or less "logical" than the other method. --Ctachme 00:22, 30 Dec 2004 (UTC)

Do we need more than one Linux distro?

People have added very incomplete Debian and Gentoo streams to the table. Linux is Linux is Linux - the kernel supports the same filesystems on all distros; X is X; they nearly all have GTK+ and QT for graphical API's, and so forth. Theres no need to have more than one distro listed. Kiand 15:31, 19 Dec 2004 (UTC)

Distributions aren't the same according to their defaults. NTFS and MP3 aren't supported on Red Hat distributions because of patent issues. Some distributions come with GNOME, some with KDE, and are therefore built around gdm or kdm. And so on. [ alerante | “” 23:15, 19 Dec 2004 (UTC) ]
I've happened to notice this too and fixed the Debian entries now. It should be fairly obvious that there is sufficient variety contributed by the existence of that particular entry. --Joy [shallot] 08:33, 20 Dec 2004 (UTC)
We now have FIVE Linux distros. This is way, way too many for the same OS. I think the example set by Windows should be kept, with two, and no more. Kiand 12:55, 22 Dec 2004 (UTC)
Note that this is after UltimaGuy's addition of Slackware. Gentoo's entry is existent but sub-stub... --Joy [shallot] 18:22, 22 Dec 2004 (UTC)

Fork of Linux distributions to Comparison of Linux distributions

Nowadays, there can never be a day where a comparison of operating systems will not be influenced by Linux supporters. It is also a burden for editors to edit an article that is largely dominated by Linux distributions. I know that much information can be added about these distributions but are kept due to the reason of not bloating this comparison of operating systems article. For the future of this series of articles, I forked the Linux distributions article to Comparison of Linux distributions. I hope you will understand this decision. Squash 03:15, 3 Jan 2005 (UTC)

This makes sense, but - writing as someone who just saw the article for the first time - I have to say that it looks a bit odd for the tables to contain NO information about Linux, even for those aspects (like the kernel type) that really don't vary between distros. Might it be worth including some "typical" Linux data in these tables for basic comparison purposes, with N/As for the really varied stuff like update tools and default file browsers?


I have gone ahead and created an entry for Netware. It is based on the published specs that are out there and personal experience. The analysis of the Netware kernel cited in the reference field that led me to call it a Microkernel/Monolithic hybrid is by one of the software architects responsible for Netware. It is somewhat dated, referring to Netware 4.0, but my understanding is that it has not changed sufficiently for this information to be completely outdated. Thanks MARQUIS111, Feb 3, 2005 12:00 EST

Including Linux's most common traits, leaving others open

I think it's bad for the article to not have Linux in the tables at all. Could we not include the most common things (like kernel structure, file systems etc.) leave out those disputed, with a link to hte Linux article? I think it could look like this:

Kernel type Default file system Other major file systems supported Architectural support
(first, primary platform)
Default online OS update tool Default package management/
software installation tool
Main API and language
for GUI applications
Main API and language
for textual/CLI applications
Linux Microkernel/
Several ext2, ext3, reiserFS, FAT, NTFS, UDF, ISO 9660, HFS x86 Several Several Several Several

What do you say?