Talk:Comparison of operating systems

Archived talk:

• Talk:Comparison_of_operating_systems/Archived talk 1
• Talk:Comparison_of_operating_systems/Archived talk 2

No mention of IRIX?

Is this list working towards an authoritative comparison of all OSes or is it just a list of current, or nearly current ones? I see HP-UX is in the table, and I beleive that it is no longer for sale and support runs out on Dec 31st of this year. IRIX will probably be on life support for a few more years. Plus it's one of the grand old OSes from one of the (formerly) big UNIX system vendors. I'd like to see it included.

Perhaps the page should be divided between legacy OSes and current OSes? --RageX 19:30, 8 March 2006 (UTC)

Perhaps it should, but, as far as I know, HP-UX isn't going away. HP says that they're adding new stuff to it. Guy Harris 20:17, 8 March 2006 (UTC)

Hmmm ... I read that HP was ending support of HP-UX on the support section of PTC.com. They link to this article on HP.com: here

It seemed to make sense, seeing that many UNIX companies are killing their own products and switching to Linux on some commodity hardware (AMD/Intel).--RageX 07:46, 10 March 2006 (UTC)

They're ending support for particular older versions of HP-UX; the article doesn't say they're ending support for HP-UX, just ening support for older versions. The note recommends upgrading to a newer version of HP-UX. Guy Harris 08:22, 10 March 2006 (UTC)

Solaris is not free

"Solaris Sun July 1992 SunOS 10 (February 1, 2005) Free CDDL Server, Workstation"

Solaris 10 is not free, it cost some dollars. OpenSolaris is free. A5b 06:45, 24 February 2006 (UTC)

Sun claims you can download it for free from:

http://www.sun.com/software/solaris/get.jsp

Guy Harris 07:44, 24 February 2006 (UTC)

Solaris is free to download. It costs if you want original media, which of course, costs money to produce. The charge with Solaris is from support. If you want anything above the freely distributed security fixes, you need a support contract. The support remains free if you have a contract for support of your hardware.The Pedant 13:53, 21 March 2006 (UTC)

• I think it would be a good idea to clarify the distinction between Solaris and OpenSolaris on the page, considering that previously Solaris was indeed a proprietary operating system. Even though a user can check the price and see that it's free, it's still reassuring knowing that it's OpenSolaris, not just some generic Solaris. Nicholasink 01:59, 28 March 2006 (UTC)

Relative Distributions

While I also concede this is one of "those" articles, what about providing an estimated installed user base for each OS? Is there even a reliable source of this information? Jeff schiller 17:22, 11 July 2005 (UTC)

That would be interesting in principle, but I think in practice it will be impossible to get reliable data. For a fair comparison, the data would have to cover not only the PC / workstation market, but also servers and ideally embedded systems. --K. Sperling 09:54, July 13, 2005 (UTC)

Cutting back

Someone half-added AIX, HP-UX and SunOS. Solaris is already here, so thats the latter out, HP-UX is abandoned really; as is AIX really.

Also, I don't think ReactOS deserves an entry, so I'm probably going to remove it to - minor Windows clone thats not yet anywhere really.

--Kiand 17:38, 15 July 2005 (UTC)

Windows ME and MacOS(classic) should probably go too, and I'm not sure Windows Server is worth keeping around, it's litle more than a repackaged XP... Lost Goblin 23:46, July 15, 2005 (UTC)

I agree about Reactos. I'm not sure what windows ME's user base is, but I suspect its not that small yet? In principle I would agree that Server and XP are basically the same OS, but then you'd have to remove XP because Server 2003 is the newer version, and that wouldn't be good considering the large user base of XP. Anyway, the Windows entries at least have pretty complete information.

With AIX and HP-UX i'm not sure, how widely are they still used? In principle I'd actually be for having more of these classic Unices and other server OSes, to avoid worsening the "home user bias".

Maybe a general guideline as to what qualfies an OS to go into the list would be good. That would save us having to take them out again (or at least avoid too much discussion of every single case). --K. Sperling 00:00, July 16, 2005 (UTC)

After some removing and reverting i've decided to leave AIX in now, apparently IBM is still selling and supporting it. I've also removed Reactos, and added a list of removed OSes to the editors' notes at the top, so far those are AmigaOS, BeOS and Reactos. --K. Sperling 12:06, July 19, 2005 (UTC)

BeOS isn't gone, its just been renamed. SkyOS was also removed (negliable user base, not even publically available) in the distant past. Theres no data for AIX, I really think it should be removed if nothing can be added. --Kiand 12:56, 30 July 2005 (UTC)

(Minor point) BeOS being renamed -- that's not really accurate. There's both BeOS proper and Zeta; one was fully created by Be and one was not. Dysprosia 14:05, 30 July 2005 (UTC)

Much as I hate having Zeta in that list (my views on the company and the product are on record), I don't think I'd get away with having BeOS in on its own anymore. However, with MacOS Classic, Plan9, AIX and OS/2 there, I think there could actually be some justification.... --Kiand 14:07, 30 July 2005 (UTC)

Partitions, zones, security contexts and jails

Every operating system seems to have a different name for this feature.  :) Perhaps we could pick a generic term to make the "Isolation mechanism" column more consistent? I use Linux, but would prefer the term "jail" as a way of crediting the FreeBSD people with the first Unix implementation. —Ghakko 09:08, 21 July 2005 (UTC)

There is a problem I think in that a zone is different from a mere jail, and if I understand you correctly, a partition is completely different as well. So a generic term would blur the distinction between the concepts. Dysprosia 09:21, 21 July 2005 (UTC)

That's a good point. Would a table like this help?

Operating system	Mechanism	File system isolation	Disk quotas	I/O rate limiting	Memory limits	CPU quotas	Network isolation
-	chroot	Yes	No	No	No	No	No
Linux	Security context	Yes	Yes	No	Yes	Yes	Yes
FreeBSD	jail	Yes	No	No	No	No	Yes
Solaris	Zone	Yes	No	Yes	Yes	Yes	Yes
z/OS	Logical partition	Yes	Yes	Yes	Yes	Yes	?

• Solaris Containers: Resource Management and Solaris Zones
• Linux-VServer Wiki
• z/OS System Programming Volume 10

—Ghakko 14:27, 22 July 2005 (UTC)

I think a comparison of these mechanisms would be useful, but should have its own article, where the mechanisms can also be described in greater detail. Maybe integrating it with Sandbox (security) would also be an option. Chroot jail is also quite weak at the moment, I think the whole subject matter could use some attention. --K. Sperling 18:18, July 22, 2005 (UTC)

I'd stick with the actual name of the mechanism as it is used in the OS. If we chose the same name, the column would just degenerate into yes/no. The name can always link to a specific article, or it can be explained in a footnote. --K. Sperling 18:15, July 21, 2005 (UTC)

add it,but why not making also a comparison page between theses mecanism,add <!-- --> at both pages and tell to update the other when you make changes213.189.165.28 23:03, 17 April 2006 (UTC)

Mac OS X and ACLs

I'm going to tentatively update the OS X entry to state that it supports UNIX permissions and ACLs starting with 10.4. While Tiger client does not have the GUI that the server version does, ACLs can still be enabled using the fsaclctl command and manipulated with chmod.

Bot to update the bug count in the security table

I wrote small bot based on the pywikipedia package that can automatically update the bug count in the Security table. I put the request to run it on Wikipedia_talk:Bots#Permission to run SecuniBot, comments encouraged. --K. Sperling 18:44, July 21, 2005 (UTC)

Out of box comments

The following was added by an anonymous editor and I have moved the comments here to discuss:

Some of their tests seem to not reflect "out of the box" status, that is, they enable all services in inetd as well as daemons that come with the base before executing the analysis software, when the "out of the box" state has them all disabled by default for several listed operating systems.

The out-of-the-box section does mention 'various configurations.' Does this not cover the issue? Uriah923 22:26, 9 August 2005 (UTC)

Hm. I'm not sure the out-of-the-box site should be mentioned in the article body at all. I think the a link in External Links (the one that is now in Reference) would be enough. --K. Sperling 12:10, August 10, 2005 (UTC)

The article does compare an important aspect of different operating systems. Why would it not be included in the article? Uriah923 13:42, 10 August 2005 (UTC)

Network Appliance

What does "Network Appliance" mean in the "Target system type" and how it's different from embedded? maybe linking the various terms might be a good idea too so people can understand the implied meaning?

When the vendors talk about a network appliance, it usually refers to a system that performs some relatively-dedicated function (but isn't "embedded" within a larger piece of equipment). The function might be web serving, file serving, a database engine, a DNS server, etc., but it's usually pretty well-constrained as compared to a general-purpose PC (connected to a human) or "server" (that may be doing *MANY* back-end tasks).

Atlant 15:20, 19 August 2005 (UTC)

Proposal of changing name and structure of this page

I propose to change this page and show in it only reference links to various comparison charts, because it fails to show (as stated in the title):

[Quote] The table only includes systems that are widely used and currently available. Due to the large number and variety of available Linux distributions, all of these operating systems are grouped under a single entry in these tables. [End quote]

It could have been called "Comparison table of OSes recent than 5 years", but it still presents MacOS and not only Mac OS X, that is the current selling version. (For example Windows'95 and windows'98 are both canceled from the table, and only Windows ME is shown. And it is correct. because both 9x versions are discontinued from market longer than 5 years.)

OS/2 is discontinued, and IBM plans to release it Open Source, but it is still on the table.

Linux OSes are shown apart. Their invoice is no longer (?) necessary.

It lacks of OSes still on the market, such as AmigaOS, which recently started also 2 split versions, AROS and MorphOS (and also MorphOS it is currently on the market). [I wrote AmigaOS invoice in the table and I planned to insert MorphOS later].

It lacks of very interesting OSes currently on the market such as RTOS (Real Time Operating Systems), which major representative is QNX.

Also it could have been called "Comparison Table of Operating Systems with a million or larger userbase", but then, FreeBSD and Solaris are far from reaching this record of installed releases.

So I propose to create multiple pages of comparison charts, and this page pointing with respective links to each of them (Historical chart ones, OSes actually on the market, Linux Only, etcetera).

But one chart amongst the others must be a very extensive comparison table showing all existing OSes, present and past ones, to be of reference for all readers, and mainly students and scholars.

Criticisms and advices are welcome.

Insert your Sign and say why, if you agree or refuse to change this page.

--80.104.191.169 01:37, 21 August 2005 (UTC) [Whoops. Sorry. I am Raffaele_Megabyte, and I failed to log, becase I reconnected internet after a brief pause]

In my opinion the tables should list OSes that are widely used. However, use isn't the same as number of installed systems; that number just boosts desktop operating systems out of proportion. If, for example, a Solaris system runs an application server used by 100 users, that would really have to count as 100 users, not just one. A good measure might be how often an OS or services running on an OS are directly or indirectly used by how many users. Or to put it another way, the amount of "real work" done using that OS. (Of course this cannot really be measured, only estimated).

Basically I'm all for including OSes like QNX and similar systems, as right now the embedded operating systems are very under-represented. However, if we were to go down that road all the way, then we'd have to start including e.g. Symbian OS or Windows CE, too, and there isn't really much point in comparing e.g. Symbian OS to Solaris.

So if there was to be a split at all, I'd split it by target system type, i.e. server OSes, workstation/desktop OSes, embedded/realtime OSes. With that kind of split, many systems could be clearly assigned to one of the types, instead of having to maintain the same data multiple times. Also, splitting into multiple tables doesn't necessarily mean splitting into multiple articles. For example, it would be possible to simply have one section per target system type.

In either case, I'm very much against doing this kind of change in the "reorganize first, fix later" style, since a lot of changes tend to stop after the "reorganize" phase and never get around to actually filling in all the new content (E.g. I've seen a fair amount of people who insist that OS xyz has to be in the tables, but then can't actually be bothered to add more than a sketchy version of the "general information" data.)

Instead, I suggest you create the reorganized version as a page in your user namespace, and then post a link here for discussion. --K. Sperling (talk) 21:21, 21 August 2005 (UTC)

Value of ON content and quality of reference

The content added from the ON reference remains in this article, but the reference has been removed. This action is disputed and a conversation is ongoing here. Uriah923 06:14, 2 September 2005 (UTC)

Unfortunately, a recent renaming of {the thing that shall not be named, lest it drag search engine references to it} to "ON" changed references to User:Uriah923/{the thing that shall not be named} to User:Uriah923/ON, which means that the Wikilinks no longer work. Should User:Uriah923/{the thing that shall not be named} be renamed to User:Uriah923/ON? Guy Harris 05:32, 24 March 2006 (UTC)

Should the link be replaced?

The following was recently removed from the article as spam contingent on an agreement here that it is not: "See (see article history for the link. Link removed to reduce SEO attempt) for a summary of vulnerability scans produced by Nessus against operating systems in various states of configuration."

I think the content is valuable and of high quality and, therefore, that the link should be replaced. Uriah923 16:54, 12 September 2005 (UTC)

I'll say pretty much the same thing here as I did at Talk:ITunes and all the other pages Uriah is spamming this message to in order to divide the issue up. This is just one out of a SEO campaign by the ON people to get links to that site from Wikipedia. (Just as we don't need the above link since it can be seen in the diff where it was removed) The same thing was done with multiple Wikipedia articles until people cried foul, and the consensus was that ON articles do not make the type of quality references Wikipedia needs. That said, if a consensus forms here (with a reasonable minimum of 5-6 people involved) that the article is valuable enough to justify an external link despite the linkspam implications, I certainly wouldn't stand in the way. That said, our article would be much better off using highly regarded networking textbooks and other high quality resources as references than linking to a website with no inherent credibility. Wikipedia:Verifiability is the goal, not seeing how many links we can get to ON. - Taxman ^Talk 18:33, September 12, 2005 (UTC)

This is not about references, this is about adding an external link. It would serve everyone best if you would adress the actual issue at hand. Uriah923 20:04, 12 September 2005 (UTC)

I did, and it would be a lot better if you would spend more time actually trying to improve wikipedia, than trying to get links to ON. If you were trying only to improve WP you would have ended this campaign long ago and started looking for actual high quality material to reference, not ON. - Taxman ^Talk 20:19, September 12, 2005 (UTC)

Again, I am not trying to reference anything. This is not about references, this about adding one specific external link to one specific site. You have not yet addressed that topic. Instead you continue to talk about SEO and linkspam. How about you take a look at the article that could be linked to and the Wikipedia policy on external links and then give your opinion on if it's a good match? Uriah923 20:30, 12 September 2005 (UTC)

No ON link, dang it, and a thought: clearly Uriah is trying to wear us down, by adding ON links to everything he can, and then demanding that we consider each link on its merits. I suggest that just as we have blocked users, we have blocked sites. Once a site is convicted of linkspamming, no links to that site can be added for, say, a year. I'd say forever, but the remote possibility exists that the site could improve and attain high enough quality that someone OTHER than the proprietor would want to link to it. Zora 21:16, 12 September 2005 (UTC)

You mean like someone did here and here? Also, are you meaning to suggest I am approaching this in the incorrect manner? ON gains nothing from holding a discussion asking for a consensus on a talk page. Why do you refuse to let the other editors look at the merits of each individual page and article and judge for themselves if an external link is appropriate? Uriah923 21:43, 12 September 2005 (UTC)

Both before it was made clear that adding links to ON and the discussion about it is only in the name of SEO spamming by you. If it wasn't you would have given this up long ago and looked for better sites to link to, and not just a token effort. Thank yourself for the fact that people don't want your links anywhere. I was willing to be reasonable until you made it so obvious that getting your links into articles was the only goal and that improving Wikipedia wasn't. - Taxman ^Talk 23:12, September 12, 2005 (UTC)

I'm sick of Uriah's SEO spamming. As it stands, his account seems little more than a Wikipedia:Role account, with the purpose of pimping traffic to his blog. He seems to have put a link to ON on every single topic he's ever written about. He presents his arguments as an "objective" look, but he is supporting the inclusion of all ON links on all of the articles so far! I oppose all ON links at this time. --DropDeadGorgias (talk) 19:55, 13 September 2005 (UTC)

I wasn't particularly hot about the link in this article when it was added in the first place, and I don't much care if it would be removed now. However, the ON article is vaguely related, so I'm not too bothered by keeping the link for the time being. In general I do agree with the critics, Uriahs pattern of editing is just too clear to be just a coincidence. First, he tried to pitch the ON links as references; now that a consensus has been reached that ON isn't high-quality enough for that, he reiterates the whole debate with "external link" instead of "reference". Part of that may be my fault for suggesting on User:Uriah923/ON that an external link may be viable in SOME (rare) cases. However, Uriah seems to have taken it as some sort of support or easy line of reasoning for adding an external link in ALMOST EVERY case, which I neither said nor implied. --K. Sperling (talk) 01:28, 14 September 2005 (UTC)

"Project Tiger OS" ???

What's that supposed to be? The link redirects to Windows Vista, the added info claimed it's something to do with BeOS, searching on the web yields almost nothing; the only thing I've found is that a Sun Java OS was/is using that project name. In either case, it doesn't seem exactly notable. --K. Sperling (talk) 10:57, 2 October 2005 (UTC)

Random vandalistic crap, its not even the name of one of the vapourware projects that are like flies around BeOS (the only legitimate one thats a 'project' and not corporate being Haiku (operating system)) --Kiand 18:41, 16 November 2005 (UTC)

Predecessor

Maybe we should clarify what we mean by predecessor. Is it necessary for the later to OS to be actually based on the code of the former, or is it enough if it shares a lot of concepts? For example I'd consider Unix to be a predecessor of Linux, even though the relevent field has recently been changed to "None". --K. Sperling (talk) 02:44, 8 November 2005 (UTC)

Windows 2003/XP?

It's pointless to put in both OSes. They should be merged into one Windows 2003/XP entry, IMHO.

WinME and MacOS should stay, since they (or their OS lineage) both still have significant userbases.

Thoughts?

I'd disagree. Even though they're fairly similar, the respective rows in the tables do actually differ. I don't see why we should ignore those differences just because they're theoretically almost the same OS. We might get rid of XP when Vista comes out, though. --K. Sperling (talk) 00:46, 12 November 2005 (UTC)

I think that we could easily take care of the security vulns. thing by specifically stating "x for XP, z for 2000." The pricing, well, we already have seperate prices for the two editions, so why not go further? Vista is basically going to be a rename, since we aren't covering too many of the changed areas.

Generally I'd have thought it always makes sense to only list the latest version of any particular OS, but as these are both currently available products, I guess it makes sense to list them both.

I think WinME and (classic) MacOS should go. Or if not, then the heading needs to be renamed. Is this a history of popular OSs, or a list of those currently available? Mdwh 22:26, 4 December 2005 (UTC)

Currently available and widely used, I believe. OS9 is still available and easily found, as are 98 and ME. MS may have dropped them from support, but it hardly means that they aren't available. -- original poster, 11 Dec 2005

Still available for sale? And yes, they are still used quite a lot, but I imagine DOS is still in use more than classic MacOS and that isn't mentioned. Mdwh 23:01, 1 February 2006 (UTC)

AmigaOS

The AmigaOS is still widely in use in Europe, and serves as web servers and script servers in the U.S. Version 4 is currently being developed for the new PowerPC AmigaOne computers, and as such is a "Current OS" and should be included. Naidim 17:16, 16 November 2005 (UTC)

A few thousand users is not "widely" used - along with RISC OS, AmigaOS is a seriously marginal OS used mainly in one country - Germany (RISC OS being English through and through and only really used there) --Kiand 22:31, 4 December 2005 (UTC)

I agree with including AmigaOS - AmigaOS 3.9 is also still available (last version dated March 2002, I believe). But I guess we need to clarify what this list is really about - admittedly AmigaOS is most well known in a historical context rather than it being currently available, as Kiand points out.

But then by this logic, classic Mac OS and Windows ME shouldn't be listed either.

Perhaps it would be better to have a similar separate list elsewhere for such OSs that are no longer widely available? Mdwh 22:35, 4 December 2005 (UTC)

As a further note - even if classic Mac OS and Windows ME are included on the basis of still being used, I also wonder about the inclusion of OS/2 and YellowTAB ZETA - are these widely used these days? Mdwh 23:05, 1 February 2006 (UTC)

Amiga OS 4 is being finished with last stable realise Update 4 and Web sites online - for developers, applications being developed etc.. This cannot thus be considered as dead OS in terms of OS/2, GEOS or maybe Windows 95/98 (very little support and availiablity). Since this is a major, rewritten OS, which handles both upgraded OLD Amigas with PPC boards and new Amiga hardware (AmigaOne) I see no reason why it cannot be listed. At the end, its about OS and not how widely it is used. And AmigaOS pages links to this. If you don`t know much about AmigaOS 4 let people who know write it

Reasons for AmigaOS listed

Community existing in:

GERMANY

ENGLAND

POLAND

FRANCE

HUNGARY

SERBIA, CROATIA

USA, AUSTRALIA

OTHER COUNTRIES WORLDWIDE

AOS 1.x-3.x was bundled with Amiga, most popular 16/32-bit personal computer about 1987 to 1993. It boosted multimedia, computer productivity and home gaming. Amiga personal computers were invovative, take for example worlds first CD based computer, Amiga CDTV, worlds first CD console - Amiga CD32and first complete non linear editing system, Video Toaster, just to mention few. Besides great hardware and software, OS itself developed a lot and inspired several projects and has 20+ years history itself, longer history and development than most of OSs listed.Since 1997 its been in transit to PowerPC, but until Amiga OS4 there was no real solution that would completely port it, due to owners change. So wherewher these computers and OSs are, there is AmigaOS too. New AmigaOne and AOS4 is emerging very slowly, and altough contemporarized OS, can be seen a minority OS. However people who use MorphOS, UAE, AmigaXXL/Amithlon, AROS ... can also be viewed as AmigaOS users.

This would be AmigaOS work in progress, please add. Without much arguing, there is:

Amiga OS Wiki page

It is listed as OS at Wikipedia

Amigans create whole Amiga Wikipedia

and altough proclaimed to be Zealots in depth pages from this one, deserve at least a display.

--Rastavox 01:20, 4 May 2006 (UTC)

It would be (Amigans and ex-Amigans would help about section below:

Lets try to list AmigaOS

'General Info'------------

Name: AmigaOS

Creator: Amiga Inc (trademark), Hyperion Entertaiment (development), OS 3.5 and OS 3.9 Haage & Partner, up to 3.1 Commodore International

First public release: Summer 1985 with Amiga 1000, OS4 PreRelease since june 2004.

Predecessor: TRIPOS, since OS4 TRIPOS Free

Latest stable release: 4.0 Prerealise Update 4 (08.02.2006)

Cost (USD): Unknown because bundled with hardware AmigaOne. Approx. US$110.

Preferred license1: Proprietary(Closed Source)

Target system type: Home Desktop

'Technical Info'------------

Name: AmigaOS

Supported arhitectures: PowerPC, 68k emulated by petunia JIT

Supported file systems: FFS, PFS (contributed), ISO 9660 UDF and FAT(via CrossDOS), handlers expandable

Kernel type: Microkernel

GUI on by default: Yes, Workbench

Package management: Amiga Installer

Update management: update packages, downloadable by registered users, Updated version shipped with AmigaOne

Primary APIs: MUI, ReAction, MiniGL, Warp3D

'Security'------------

Need help, I do think that AmigaOS lacks all these mentioned features. Worth of mentioning is that because its little used AmigaOS, there are no viruses, spywares, netbuses or any other software that can externaly harm AmigaOS 4. No x86 code works As it is a PPC driven OS, so maybe all answers are YES :-)

Check this for more info and support:------------

AmigaOS in brief (not very updated)

OS4 Web Site

OS4 Prerelease Update 1 review

First impressions – AOS4 Update 4

Amiga Community

OS4 Free Files Archive

Yes, yes, about 2006. Amiga is slowly coming back. Marginalized, but coming back. As Jamaicans say: "Dont call me duppy, if me nah dead" (Dont call me ghost, if I aint dead) Excuse my English and clumsy Wikipeia edit.

One Love, --Rastavox 01:17, 4 May 2006 (UTC)

OS X and BSD/POSIX as APIs

Should BSD and POSIX be listed as primary APIs, given that a lot of free software written to those APIs runs on OS X (and that much of the non-GUI parts of Carbon and Cocoa, such as the file system and networking APIs, are built atop the BSD/POSIX APIs)? Or should they be listed in a footnote or a parenthetical note, given that most commercial apps for OS X are written with Carbon or Cocoa? Guy Harris 00:47, 11 January 2006 (UTC)

• I would think they'd be considered first-class APIs just like Carbon and Cocoa. I mean, there's plenty of stuff on there that uses the BSD layer -- the entire command line userland comes from FreeBSD, for example, and X windows is getting to be pretty important as well. And I believe Firefox, to take one specific example, has an Aqua front end and a Unix backend. (Don't forget, not all important software is commercial anyway; I do believe there's a good amount of GUI stuff in Fink that requires X.) Haikupoet 01:28, 11 January 2006 (UTC)
  • As a member of of the core team for Ethereal (which might well be as much used as any of the commercial network analyzers on OS X; it's the only one I use on OS X), I'm quite aware that not all important OS X software is commercial. :-) I think they should be listed as primary APIs, but I figured the only reason they weren't is that somebody figured they weren't as important, so I wanted to give them the chance to defend not marking them as primary APIs if they'd explicitly decided not to do so. Unless somebody can convincingly defend not listing them as primary, I'd list them as primary. Guy Harris 01:36, 11 January 2006 (UTC)
    • Done and done, because I entirely agree with you. I think part of the problem is that there's a metric assload of people who really don't have a full grasp of what the Mac platform is all about (with today's news about the new Intel iMacs and the horribly named MacBook all the Mac ignorance is crawling out of the woodwork on the chat sites). Haikupoet 01:44, 11 January 2006 (UTC)

X11 as a primary API?

With X11 listed as one of the primary APIs for OS X, should it also be listed for the other UN*Xes, which also have an optional X11 - or which might install with X11 standard in desktop configurations, as they, unlike OS X, have only an X11-based GUI? Guy Harris 01:53, 11 January 2006 (UTC)

• I have no objection myself, though it occurs to me it should be listed as "optional" on the Linux entry. Haikupoet 02:13, 11 January 2006 (UTC)
  • On what UN*Xes would it not be optional? You can also install the BSDs without X11, I think; I don't know about Solaris, AIX, HP-UX, etc., and I forget whether I had to install X11 separately with Tiger on my PowerBook. Guy Harris 03:34, 11 January 2006 (UTC)
    • A fair point. Hey, I can be wrong about things :-) In any case, virtually all Unix GUI software requires X except for the odd embedded system. Haikupoet 04:04, 11 January 2006 (UTC)
      • I'd in fact remove it from the OS X row, since it's not exactly a primary API of OS X, or any other OS where an X server is an optional component. --K. Sperling (talk) 07:53, 24 January 2006 (UTC)

it good to include it as api,just add a note at the end explaining why...and telling that most unixes have X11

i don't know well but you should check if the 10.3 is the first version where X was ported or where X is integrated by default

Order?

The tables list the operating systems in alphabetical order -- except for Linux, which is two items higher than it should be. Is there a reason for this? 203.211.79.208 09:55, 31 January 2006 (UTC)

• Ah, of course; someone must have changed the name from GNU/Linux. 203.211.79.208 10:00, 31 January 2006 (UTC)

Fixed it. Dysprosia 10:01, 31 January 2006 (UTC)

Inferno

Vita Nuova made Inferno - Bell Labs made plan 9, Inferno is derived from plan 9. Much like HP-UX is derived from Unix SysV, that does not make AT&T the creator of HP-UX. For this reason I have put Vita Nuova back as the creator of Inferno. Janizary 04:42, 1 March 2006 (UTC)

No, that is not correct. Vita Nuova obtained the rights to Inferno from Bell Labs. Bell Labs originally created Inferno. Some evidence: a paper on the design of Inferno's VM by Winterbottom and Pike from Bell Labs, a press release from Lucent stating that "The Inferno software was developed by researchers Rob Pike, Phil Winterbottom and Sean Dorward of the Bell Labs Computing Sciences Research Center". You may also want to use the Wayback Machine on http://inferno.bell-labs.com . Dysprosia 06:31, 1 March 2006 (UTC)

See also The design of the Inferno virtual machine, by Phil Winterbottom and Rob Pike; they're listed as being at "Bell Labs, Lucent Technologies", with Bell Labs e-mail addresses. Guy Harris 07:38, 1 March 2006 (UTC)

That's the paper I quote above ;) Dysprosia 07:42, 1 March 2006 (UTC)

Yeah, you can go ahead and fix that then. Janizary 03:16, 2 March 2006 (UTC)

Done. Guy Harris 03:27, 2 March 2006 (UTC)

Supported file systems

Are the filesystems to be listed there just disk file systems? I'm wondering because Mac OS X supports a bunch of network based filesystems natively. Here is the list from /System/Library/Filesystems/:

• afpfs.fs
• AppleShare
• cd9660.fs
• cddafs.fs
• ftp.fs
• hfs.fs
• msdos.fs
• nfs.fs
• ntfs.fs
• smbfs.fs
• udf.fs
• ufs.fs
• URLMount
• webdav.fs

afpfs, AppleShare, ftp, nfs, smbfs and webdav are all network based filesystems and are currently not listed. I am assuming msdos refers to the FAT filesystem. Dread Lord CyberSkull ✎☠ 12:15, 11 March 2006 (UTC)

No, the file systems listed aren't supposed to be just local disk-based file systems; NFS, for example, is listed in most of the entries - including the OS X one.

From your list of network file systems:

• afpfs.fs is a symlink to AppleShare/afpfs.kext, so they're not separate; that's the AFP client code, which is listed for OS X as "AFP";
• ftp.fs isn't listed;
• nfs.fs is listed as "NFS";
• smbfs.fs is listed as "SMBFS";
• webdav.fs isn't listed;

so the only network-based file systems not listed are FTPFS and WebDAV.

Other entries from your first list:

• cd9660.fs is for the standard ISO 9660 CD-ROM file system, so it's listed as "ISO 9660".
• cddafs.fs is for mounting audio CD's (stick an audio CD into your Mac and it gets mounted as a file system); it's not listed. (Its name is parsed as "CDDA FS", where "CDDA" or "CD-DA" is "Compact Disk Digital Audio", referring to the Red Book CD standard for audio CD's.)
• msdos.fs is, indeed, for various versions of the DOS FAT file system, so it's listed as "FAT".
• URLMount isn't a file system; it's a collection of userland modules to support mounting various file system types by URL.

For some other OSes, the list has "and others" at the end, because an exhaustive list of the file systems would be a bit long (especially for Linux); perhaps the major ones should be listed, and "and others" added if there are other minor ones, so we'd add "and others" to the OS X list (and perhaps add SMBFS to the Linux, FreeBSD, and NetBSD entries). Guy Harris 20:57, 11 March 2006 (UTC)

I'm wondering if we shouldn't construct a new table with file system support. I also forgot to mention disk image support. Below is a table I've cobbled together:

Operating System	Physical file systems												Network file systems					Disk images
	ext2	ext3	FAT	HFS	HFS+	ISO 9660	JFS	NTFS	Red Book	UDF	UFS	XFS	AFP	FTP	NFS	SMBFS	WebDAV	.cue/.bin	.dmg	.iso
Mac OS X	No	No	Yes	Yes	Yes	Yes	No	Some read-only	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	No
Windows XP	No	No	Yes	No	No	Yes	No	Yes	Yes	Yes	No	No	Yes	No	Yes	Yes	No	No	No	No

Dread Lord CyberSkull ✎☠ 12:14, 12 March 2006 (UTC)

UFS1 vs. UFS2? ReiserFS (and particular versions of ReiserFS)? XFS? JFS? Etc., etc., etc.. I.e., which file systems should be included? File systems supported by more than one OS? That'd add XFS and JFS, for starters. Guy Harris 00:15, 13 March 2006 (UTC)

"Disk images" aren't a type of file system in OS X, in the sense that HFS+, UFS, XFS, JFS, etc. are - they're just a mechanism by which a file can be made the backing store for a pseudo-disk device whose driver offers the same API/ABI as real disk drivers, so underlying file systems (e.g., HFS+) can use them in the same way that they use real disks. Guy Harris 06:45, 10 April 2006 (UTC)

Guy is right. disk images are comparable to Zips, floppies, CDs and DVDs. They provide a (virtual) storage medium but they aren't in themselves file systems - just empty bits waiting to be formatted with a file system. I added AFP to the list for classic Mac OS since it was missing.

--Thepreacher 14:18, 22 May 2006 (UTC)

.iso isn't a filesystem

simply(under aa unix OS) type dd if=/dev/cdrom of=image.iso and you'll have an iso image

=>the filsystem is iso9660

remplace it by loop

because i think that windows hasn't got any loop by default

some filesystem:

download a linux kenrel and type make xconfig(under an unix with X)

there are realy a lot of filesysytem

by the way we must add rw and ro instead of yes

we could include all the "compatibility" filsystems suported by linux

Integrated GUI

So what makes a GUI "integrated"? If an "integrated GUI" means the system can't function without a GUI, I suspect Solaris doesn't count, just as OpenBSD doesn't count. If it means the GUI is bundled with the OS distribution, that'd mean most if not all of the systems listed have an "integrated GUI", but the footnote for "integrated GUI environment" says "Operating systems where the GUI is not integrated into the core OS are often bundled with an implementation of the X Window System. However, installing X is usually optional.", which seems to imply that you need more than bundling to make a GUI "integrated".

Hmm. I'll have to see if I can tweak Mac OS X to boot without the WindowServer and without loginwindow on /dev/console (probably doable); if so, perhaps that means OS X's GUI isn't integrated, either. Microsoft Windows might be a bit more of a stretch, although there was a boot mode for Windows 2000 that, as I remember, just put up a console window. Guy Harris 09:08, 12 April 2006 (UTC)

I believe the interpretation is that if the graphics primitives are in the kernel, then the GUI is "integrated". But I suppose a looser interpretation could consist of whether the system is still as usable without using the GUI -- this isn't the case for Mac OS X, but of course the 'nix variants are as usable without a GUI as without. Yet then we have the subjective interpretation of usability. Dysprosia 11:30, 12 April 2006 (UTC)

What does "graphics primitives" mean here? In OS X, there are low-level graphics drivers, but a lot of the graphic work is done by the WindowServer and libraries, so most of the graphics work is done outside the kernel.

As for usability, yes, it's probably a matter of opinion of whether Unix-like systems other than OS X are as usable without a GUI as without; I suspect those who would find a Linux PC as usable via a remote login in a terminal window or on a raw console as it is in some X11-based GUI might well find OS X as usable via a remote login in a terminal window or on a raw console. People who find the UN*X CLI difficult to use would probably find it difficult to use regardless of whether uname -s says "Darwin", "Linux", "FreeBSD", "SunOS", or whatever. Guy Harris 19:35, 12 April 2006 (UTC)

Since we're talking about UI things here, these graphics primitives would be window compositing and drawing software, etc. So, it is true, with the mentioned definition, that OS X does not have an "integrated GUI", but then that's a real stretch in some senses, because the amount of effort to separate X from 'nix is not really comparable to separating the WindowServer from OS X, plus Mac OS X wasn't exactly designed to be run without the GUI subsystem in any case. I guess the issue is a more complex one than was first thought... Dysprosia 10:05, 14 April 2006 (UTC)

Comparing GNU/Linux to Windows or OS X in this regard is probably not ideal. I think that's where the confusion comes from — we're comparing kernels to complete OS's. Warrens 18:38, 12 April 2006 (UTC)

If you're comparing GNU/Linux with Windows or OS X, you're comparing complete OS's to complete OS's, at least as I understand what Stallman means by "GNU/Linux" - it's not just the Linux kernel, it's the Linux kernel plus all the additional GNU software running atop it. It doesn't necessarily include GUI code, however. Guy Harris 19:35, 12 April 2006 (UTC)

Defining "integrated GUI" as "cannot be used without a GUI" is misleading, if the latter is what the column means, that is what the column heading should say. To me, "integrated GUI" means it is a part of the base OS, not a additional package with other third-party software. This means that Windows, OS X and OpenBSD are in (in the same way OpenBSD has integrated Apache) and FreeBSD and NetBSD are out (Xorg is on ports/pkgsrc). NicM 15:09, 14 April 2006 (UTC).

It would also mean that many of the commercial UN*Xes are also in, as they come bundled with X and, often, a desktop environment. Guy Harris 16:08, 14 April 2006 (UTC)

Probably. Is it part of the full basic OS install? Then I think it is integrated. Is it a package of third-party software? Not integrated. I do have a nasty feeling there is a huge gaping flaw in my definition that I can't see though :-) NicM 16:40, 14 April 2006 (UTC).

In Solaris, the X server and basic X libraries, and CDE, are definitely not a package of third-party software; they're "third-party software" in the sense that a lot of it comes from X11, but Sun maintain and develop them, and ship them as part of the OS. I think the Java Desktop System is more than just a package of GNOME, etc..

The GUI stuff is probably an optional install, as a Solaris box might be a server, but if you can say "install everything", then, as far as I know, it'll install the GUI stuff.

I suspect at least some of the other commercial UN*Xes are similar to Solaris in this regard. Guy Harris 17:19, 14 April 2006 (UTC)

That is my point. OpenBSD uses Xorg, the bulk of which is third party, but it is maintained as a seperate version in the OpenBSD CVS repository with a number of OpenBSD-specific modifications, and is shipped as part of the base system, not as a package in the ports system where external software resides. Does that not count as integrated? You can omit X by not installing the sets, but I don't think that means anything: you can omit gcc (and some other bits) too if you leave out the compiler sets, but one wouldn't say it wasn't integrated. NicM 17:41, 14 April 2006 (UTC).

If that's the definition of "integrated" for this table, then Solaris has an integrated GUI, as do at least some other commercial UN*Xes - and so do, for example, FreeBSD "distributions" such as PC-BSD DesktopBSD in the Comparison of BSD operating systems, and so do a number of Linux distributions (depending on what you consider "integrated" vs. "third-party" in Linux distributions). That's fine with me - as long as the pages using "integrated GUI" have a note saying what it means, and as long as more than just OpenBSD gets flagged as having an integrated GUI (OpenBSD isn't any different from, say, Solaris here). Guy Harris 18:05, 14 April 2006 (UTC)

Well, that is what I think, I don't know if anyone else agrees :-). I agree that whatever is decided it should be applied consistently. NicM 18:20, 14 April 2006 (UTC).

To me, "integrated GUI" means it is installed and turned on by default. This is, I think, largely consistent with what is already in the table of this article. By that definition, OpenBSD does not qualify, as it's installation is not even "recommended", and if you do install it, it does not automatically start up. Nevertheless, footnotes should explain about this. Armedblowfish 18:43, 14 April 2006 (UTC)

I think that is a terrible definition. Apache, OpenNTPD and many other things are integrated into OpenBSD but not turned on by default. As far as I am concerned, your criteria are fine for a yes/no in the column, but the heading should not be "Integrated GUI", it should be "GUI by default" or similar. NicM 19:49, 14 April 2006 (UTC).

That would be far less ambiguous. Changed. Armedblowfish 20:30, 14 April 2006 (UTC)

You should probably change the footnote for the column title here, and in Comparison of BSD operating systems, as well - the current footnote applies to "Integrated GUI", in the sense of "the GUI code is part of the core OS installation by default", not to "GUI is installed and enabled by default". Guy Harris 21:06, 14 April 2006 (UTC)

There. meep. Armedblowfish 21:54, 14 April 2006 (UTC)