Talk:Comparison of operating systems/Archived talk 2

From Wikipedia, the free encyclopedia

Archive This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Contents

Solaris license

I've changed the Solaris license from the CDDL to Proprietary because specifying the CDDL is technically incorrect. OpenSolaris will be licensed under the CDDL, but commercial Solaris uses a proprietary license, the Solaris OE Software License Agreement. --ndc 04:15, 12 May 2005 (UTC)

68.161.176.231 reverted my changes with no explanation. I believe the text and link above show that listing the CDDL as Solaris' license is inappropriate. To --K. Sperling June 28, 2005 15:40 (UTC)clarify, claiming that Solaris and OpenSolaris are the same (and hence share licenses) would be the same as claiming Mac OS X and Darwin the same OS. Note that the OS X entry lists both a proprietary license (for OS X) and the ASPL (for Darwin), so perhaps doing the same for Solaris is suitable. If I'm wrong, please correct me. If not, I'll change it shortly. --ndc 01:54, 18 Jun 2005 (UTC)
Thinking more about this, perhaps we need to start (another) new discussion about how that column works. Fedora has it's own RTU-ish license1 similar to Solaris'. It relies on the GPL but is not itself the GPL. Does the license column refer to the license which the released (e.g. distribution format, be it CD, DVD, disk images, slack disk sets...) version is shipped as? In this respect, even Fedora has a proprietary license. Unfortunately, you're talking about 2 different things here - the distributed version of the OS and the source code that makes up the OS. In the case of Solaris, the source is under the CDDL license, but the compiled, packaged, and distributed OS is under the proprietary license. Does anyone else have any interest in this topic? --ndc 21:07, 20 Jun 2005 (UTC)
Maybe saying "proprietary (source CDDL)" would make sense then, if the source released under the CDDL actually contains the source to everything included in the binary release. Expanding the footnote to explain the realse vs. source license issue would probably be good, too.

"Target Audience"

For AmigaOS I would say that "Enthusiast" is whay more correct than "Home Desktop, Media creators"(at least for the last 10 years). For WinXP, how it can be targeted for "software development" when it dosn't even include a compiler? (and yes, I know you can download compilers for free, but that is not the point, you can get compilers for any OS)

I made those two corrections, if someone disagrees please coment here. Lost Goblin 09:53, 2005 May 17 (UTC)

XP target audience is 'Creative professionals, Scientists, Business/Home Desktop' instead of 'Home and Business Desktop'? uh? If no one objects I will revert that.

Windows OTT-ness

If theres no objections, I'm going to kill the Windows 3.1, 95, 98, NT4 and 2000 entries - it is completely over the top to have them, especially as they're all in the same series as ones listed there. --Kiand 17:04, 2 Jun 2005 (UTC)

Go for NT desktop, NT server, and 9x. « alerante ✆ ✉ » 13:58, 3 Jun 2005 (UTC)
Which is what I've done - XP, 2K3 and ME. --Kiand 14:05, 3 Jun 2005 (UTC)
Ah, good, although I'd like to see those as the actual row headings, and XP/2003/Me under the version. Objections? « alerante ✆ ✉ » 14:12, 4 Jun 2005 (UTC)

A number of changes

I think the article needed a number of changes, which I made just (bold editing and all that). However, I wanted to provide a number of comments on the edits:

  • Target Audience: This is a quite fuzzy column, it listed people as well as system types. Especially "enthusiast" doesn't really give any information at all; there are enthusiasts for any operating system. "Scientists" is very vague, too. What OS a scientist uses would obviously depend on what kind of software she needs, which of course depends on the science in question. All things considered, I think it's much better to give a broad indication of what type of system the OS is designed for (server / workstation / home desktop / embedded), and leave any more detailed elaborations to the individual articles (and the companies' marketing departments).
  • Version numbers: It really doesn't make sense to specify the version number down to the patch level here. It's not like one could buy or even download a specific patch level of most OSes. If somebody is interested in the exact latest version of an OS down to the patch level, they'll use the OS's website. The major (and maybe minor) version is the relevant level of detail here.
  • Cost: This column is meant to give a general idea of how much the operating system costs, there is no need to list a large number of options.
  • Removal of AmigaOS: Much as I like Amigas, I don't think AmigaOS belongs in this table, as it is only of historical interest. If you include AmigaOS, dozens of other OSes would also have to be included.
  • Removal of BeOS: BeOS doesn't exist as a product anymore. There are some efforts underway to reimplement a BeOS-like system as open source, but these projects are nowhere near the state that would warrant an inclusion in this table.
  • Linux (GNU/Linux): These systems have quite a lot in common, so its not necessary to leave every column empty. Since we're really talking about the whole OS here and not just the kernel, I think GNU/Linux is the more accurate name here. Or maybe "GNU/Linux based systems", but thats a bit lengthy.
  • Lots of the technical features information was wildly inaccurate or even plain wrong. Just to give a few examples
    • VT100 is not an API, its a protocol for character-based terminals
    • While mosts OSes specify their APIs for C, that doesn't mean they can only be used with that language. In the end, the actual APIs are binary and usable with many languages.
    • MFC and .NET aren't system APIs.
    • x64-64 aka AMD64 and IA-64 aren't the same thing. And just because AMD64 is x86-compatible doesn't mean any OS that runs on x86 automatically has AMD64 support (in the sense that it can actually use the 64bit features).
  • Default and other file systems can easily go into one column, by just putting the default file systems first. It is not necessary to list ALL file systems supported by an OS, just the major ones. E.g. NTFS support in the free unices is quite limited, due to the fact that its not publicly documented.
  • The Features table: I dont think this table conveyed much useful information.
    • All OSes that have integrated GUIs have some sort of desktop/file browser; what name the particular vendor chooses to call it isn't really the type of information you need a table for. For example, the file systems column is useful. You can use it forwards, to see which FSes an OS supports, but you can also use it backwards, to see which OSes support e.g. UFS. On the other hand, asking "Which OSes use Windows Explorer?" doesn't make much sense, since obviously the answer is Windows.
    • All OSes running X can run lots of different window managers and desktop environments. Which one is installed by default doesn't have much more significance than what background color the desktop is set to by default.

This should summarize most of the changes I've made and the reasons for them. I'm of course open to discussion on all of these points, so please consider posting here or fixing the specific problems instead of just hitting "revert" :-) --K. Sperling 18:40, 2005 Jun 14 (UTC)

I'm putting BeOS back in. Main reason - yellowTAB. Its a BeOS. Its commercial. Its a v1.0 level. Its on sale. Its still a product. Its still in use. --Kiand 18:42, 14 Jun 2005 (UTC)
Under the 'its not a product' logic, if you deny BeOS, take OS/2 out also. Its not being maintained by IBM, but by Serenity. --Kiand 18:43, 14 Jun 2005 (UTC)
Fair enough. I was actually wondering about OS/2 and Plan 9, too. Maybe calling it yellowTAB BeOS or something like that would be more appropriate, though? --K. Sperling 18:46, 2005 Jun 14 (UTC)
I've put it in as yellowTAB Zeta, much as I hate calling it anything other than BeOS, as that is nothing but a tradename on top of BeOS R5.1. Same as Serenity Systems eComStation is OS/2 4.5 rebadged. --Kiand 18:52, 14 Jun 2005 (UTC)
Does anyone have any objection to me reinstating AmigaOS? It is still under active development (not just "historical") and while obviously its not exactly popular, compared to Windows etc it's probably as (in)significant as Zeta.--Alex Whittaker
I've no objections, but theres no way that AmigaOS is as commonly used or as commercially viable as Zeta at the moment - yT are shifting about 2,000 copies a week; whereas eyetech might sell 2,000 AmigaOne's a month if they're extremely lucky. --Kiand 21:12, 20 Jun 2005 (UTC)

I agree with some of your changes and disagree with some of your changes:

  • Target Audience to Target System type was good
  • Version numbers... what? Detail is good, the more the better. I say keep as much in as possible--within reason. For example, saying 10.4.1 "Tiger" is >> than saying 10.4 "Tiger" because it is more accurate, plain and simple.
  • Cost: good change
  • Removal of rare systems is a good idea
  • Linux... whatever
  • Fixing technical information is good :)
  • I disagree with combining the default and supported filesystems because it is not clear at all that the first is the default, it was much better with two separate columns
  • I also disagree with your removal of the features table. Frankly, it doesn't matter if you didn't find it very facinating, other people might and there's no harm having additional information. The names of file browsers, for example, are useful because that offers links to articles on those specific file browsers.
  • I would support a column on the default color of the background, yes... or even better... thumbnails of the default background images! As always the more info the better. --Ctachme 04:01, 21 Jun 2005 (UTC)
More information is only good if it's also accurate. "10.4.1" vs "10.4" I don't really care much about, at least it's still a release number. Adding build number/patch level is plain nonsense. This table gives a broad overview over the listed operating systems, and in thus necessarily a bit coarse in places. E.g. "Target System Type", "Cost", "Predecessor" (and probably most other columns) are obviously simplified, too. And if I'd really need to know the version of an OS down to the latest build number, I'd not trust the table to be up to date anyway, since they change quite often.
The problem with the features table (and tables in general) is this: A table only provides very limited space in each cell, so the columns have to be chosen so that it's actually possible to represent that piece of information about the OS in question somewhat accurately in that little space. I.e. the columns have be applicable to at at least most (better all) of the OSes being compared. The question "what's the default file browser" may be (somewhat) sensible for a desktop OS, but is not really applicable for a server OS, because the GUI is usually an optional part there, and a file browser doubly so. You'd have to write something like "If you install a GUI, then the default FB is ...". And even the "default" doesn't really make much sense there: If there are multiple options that are equally easily installable, then the question which of those is the default is fairly meaningless. If you go the other way and instead write "n/a" in the "default file browser" column for an OS, that sounds like the OS doesn't ship with a file browser at all, which is probably not the case.
Let me try to put it differently: If you put the information I about operating system O into a column C of a table T, then the statement "in the context T, I is the C of O" (e.g. "speaking of OS features, the default background color of NetBSD is orange") should be
  1. True and reasonably accurate
  2. Relevant to a description of O in the context T
If (1) is not the case, then the table is inaccurate and needs to be fixed. If it's not possible to give a reasonably accurate answer in the space available in a table cell, then that's probably a case of (2). And if (2) is not true for all but a very few items in a column, then the table, while not necessarily wrong, is a misrepresentation of the facts.
About "default filesystem": First, it's actually inaccurate to say that e.g. NTFS is the default FS for Windows XP, because for floppies the default is FAT and for CDs it's ISO 9660. And in general, the default chosen by the installer could easily be different according to harddisk size, selected system type (workstation vs. server) etc., making it impossible to accurately represent in the table.
I generally question the usefulness of most "default ..." columns. It's not clear wether it means that the default choice is somehow better supported than the optional ones, or if it's really just a default in the sense that using any of the other options requires an additional click or two. In the first case, it would probably better be called "primary ...", and in the latter case it's a trivial bit of information not worth wasting an extra column for, though displaying the default item in a list of "supported ..." in bold may make sense in some cases.
I hope I have managed to clarify my problems with some of the columns in these tables general, and the features table in particular. If you really want a features table, then I suggest having a separate table for different target system types, e.g. "Server OS Features" "Home Desktop OS Features" or something like that, and only put those OSes into these table where these are applicable. --K. Sperling 10:40, 2005 Jun 23 (UTC)

Security Table?

Is that really useful? Not even considering the fact that it lists a rather different set of OSes than the main tables, I doubt the usefulness of manually mirroring changing data that is readily available on the web in a wikipedia article. Information like that is only useful if it's current. Is whoever added it actually willing to update it every day to ensure it stays up to date? Probably not. So why not just link to the relevant sites; anyone interested gets the same information just as quickly (and can actually be sure that is IS in fact up to date).

Sometimes it helps to keep in mind that Wikipedia is meant to be an encyclopedia, and not a repository of everything. That's the Web ;-) --K. Sperling 10:57, 2005 Jun 23 (UTC)

Dump it. We only have "Linux" as a general concept anyway, most distros are usually carrying so much software than the combined vunerabilites would outweigh Windows, and I know the entire point of that table was someone trying to make Windows look bad. We don't have specific distros, it'll be hell to update, Secunia, etc don't cover OS/2\BeOS\whatever vunerabilities anyway. --Kiand 11:16, 23 Jun 2005 (UTC)
"Is that really useful?" This topic (and related) was discussed at http://en.wikipedia.org/wiki/Talk:Comparison_of_operating_systems_%28security%29 and http://en.wikipedia.org/wiki/Wikipedia:Votes_for_deletion/Comparison_of_web_browsers_%28security%29
  • There are general consensus that "statistics are human knowledge, encyclopedic", "This is a lovely, concise and useful article", useful source of information that belongs in an encyclopedia and so on.
  • "Information like that is only useful if it's current. Is whoever added it actually willing to update it every day to ensure it stays up to date?. ... " Trends and general comparison are mostly same for a long time. This can provide general idea about care for customers/users security of each monitored systems. Microsoft release vulnerability fixes continually every second Tuesday at month, and other systems with steadily low number of unpatched vulnerabilities are easily updated.
  • "So why not just link to the relevant sites; anyone interested gets the same information just as quickly (and can actually be sure that is IS in fact up to date). ... I doubt the usefulness of manually mirroring changing data that is readily available on the web in a wikipedia article." You are wrong. The problem is that this security sources don't contain well arranged data about various operating systems, nobody has a chance to quickly compare various products. This comparison also have direct relation with topic such as operating systems, security or critism of Microsoft for security of products. They are based on facts and reports from renowned and highly regarded security firm - Secunia and SecurityFocus and can bee anytime monitored and easily updated.
  • "We only have "Linux" as a general concept anyway" Yes, I understand that is hard to add all linux distributions between other systems. But purpose of this comparison is comparise main, most-used operating systems and distributions.
  • "most distros are usually carrying so much software than the combined vunerabilites would outweigh Windows" This is big FUD as you can see at comparison. Most distributions have much less known unpatched vulnerabilities (including all distributed packages and programs) than Microsoft Windows with Notepad, Calc and several other primitive programs.
  • "I know the entire point of that table was someone trying to make Windows look bad." I, You or somebody else normal user don't take responsibility for creator ability to fix known public vulnerabilities as soon as possible. Many users people want to have as secure experience as it is possible for them. This comparison based on vulnerability reports by two independent, renowned and highly regarded security firms give visitors basic data. I can't for that Microsoft made worse quality software, while other creators can better respond to security vulnerabilities, as anyone can see in that comparison.
  • "Secunia, etc don't cover OS/2\BeOS\whatever vunerabilities anyway" 1. BeOS doesn't exists now (see table please). 2. OS/2 is continually monitoring by SecurityFocus. But we should primarily orient to most popular and most used systems (Win,Mac,Lin,FreeBsd) and then for something else.
  • So I think this type of comparison should stay on Wikipedia, becouse they are interesting and important for many users. I blow a hole in every your argument. Either anybody should present another argument or somebody should revert comparison table in article. Thank you.
--Ptomes 17:04, 23 Jun 2005 (UTC)
I'm in a hurry and may go over this again later, but Zeta is to all intents and purposes BeOS. Do Secunia carry vun's for it? Nope. No need to be pedantic to prove a point. --Kiand 17:22, 23 Jun 2005 (UTC)
Yes, You are right here. But how many people use it? Newest Mandrake version isn't there too. Do you want table with every-operating-system-on-the-world-which-ever-existed? No, it belong to complete list of operating systems, not in comparison. Look at the main tables of this pages. Do you really think that there is what should be there? I think 99% of visitors never heard about most operating systems in table and you certainly know it too. Why do you complicate that tables? It loose point now. I still think that small comparison of most used and most known operating system with vulnerabilities info is acceptable, reasonable, interesting and clear. And even maybe inspirating to another tables on that page. --Ptomes 17:56, 23 Jun 2005 (UTC)
Its selling extremely well at the moment, and apparently has some pretty major market share in Germany. Its also still sold, unlike Windows ME, and its actively maintained (rather than passively), unlike Plan 9 and OS/2. Remove those three first before you start going after an actively sold, actively maintained, widely used OS.
Another thing to remember is that during the paranoid overhyped "This is the year of Linux on the desktop" period from 1998-2001, BeOS was the only non-Windows OS that x86 system vendors would touch... it was extremely important in getting Microsoft to target NT at the home desktop rather than milk a bit longer out of 9x, because companies like AST, Compaq and Hitachi were selling machines with a workstation OS to the home market... --Kiand 18:04, 23 Jun 2005 (UTC)
Thank for interetsing details, I don't know it. Maybe we should start some general discuss about further direction Comparison of operating systems, but I think, security at least minimal scope are important here. --Ptomes 18:23, 23 Jun 2005 (UTC)
The security table at the very least should go into a separated article. Lost Goblin 13:13, 2005 Jun 24 (UTC)
This has been already well discussed in http://en.wikipedia.org/wiki/Talk:Comparison_of_operating_systems_%28security%29 and related topic has been discussed in http://en.wikipedia.org/wiki/Wikipedia:Votes_for_deletion/Comparison_of_web_browsers_%28security%29 Most people agree that this comparison should be in this article. --Ptomes 13:48, 24 Jun 2005 (UTC)
Ptomes: "Most people agree that this comparison should be in this article." I see no such consensus in the discussions you link to. The discussion at Comparison of operating systems (security) has exactly one person speaking up for the table, namely you yourself. And incidentally, you're also the person who created that article/table in the first place. The other people merely stated that it should be merged. The browser discussion is about a completely different table.
It's clear you are wrong here because related discussion is at http://en.wikipedia.org/wiki/Wikipedia:Votes_for_deletion/Comparison_of_web_browsers_%28security%29 and this is related topic and so direct analogy. --Ptomes 28 June 2005 10:26 (UTC)
I wouldn't be against security related information in the tables, if it were possible to give some condensed overview in an NPOV way. Unfortunately, I don't think that's possible. Security is very hard to quantify, and just giving some "unfixed bugs" number is misleading at best. The security of a system running a certain OS depends on a lot of factors, and the OS itself is just one of them.
I agree with you that security is very hard and depends on many factors, but it is main purpose of table show number of unfixed bugs. It is reliable data based on security reports two independent and respected security agencies. In future we could add some columns there, for example integrated firewall and so on. --Ptomes 28 June 2005 10:26 (UTC)
And even if we agreed to include a table, it should list the same operating systems as the main table, in the same order (and not according to some questionable security rank). And it should provide a few well-condensed figures, not a large number of individually irrelevant information (e.g. non-critical bugs probably don't say much about the general security of an OS).
Why questionable security rank??? You have false argument. It's ranked for number of bugs and old of public that bugs. Yes, I agree that non-critical bugs probably don't say much about the general security of an OS, but there are rather for fullness. --Ptomes 28 June 2005 10:26 (UTC)
"Questionable" because the number of bugs isn't a good metric for over-all OS security. But see my full reply below, especially regarding completeness. --K. Sperling June 28, 2005 15:31 (UTC)
In your post above, you have stated yourself that your motivation for this table is showing "that Microsoft made worse quality software". That's blatant POV (and while I would agree that MS doesn't have the best approach to OS security, this article is not the place for opinions like that). Unprofessional MS bashing just makes Wikipedia look bad. Why don't you take your table over to Operating system advocacy. All things considered, I'm removing the table again. Please don't just re-add it without addressing any of the criticisms (as you did the last time).
Please be carefull about your statement. My motivation for this table isn't showing bad results of particular systems, but showing really results on well aranged form. And this is what this table is about. I have ***no responsibility*** about ability some vendors fixes vulnerabilities quickly and on time. It's their issue or problem. So I can't that Microsoft products are in the end of table.
I'm not sure where you get the idea that I'm asking you to take responsibility for any vendors' problems, I'm not. Nor am I trying to make Microsoft look better than they are. But I'm also not trying to make them look worse than they are, or make Linux/BSD look better than they are. If there is to be a security table, I want it to cover the breadth of OS security related information in a concise and fair manner. --K. Sperling June 28, 2005 15:31 (UTC)
It is logic, important and well aranged component of this topic and this article (see related disscusionabout security of web browsers). It has particulary relation with critism of Microsoft article (and can be used as a basis for it). It is based on two independent and respected sources. It provide some general idea about responsibility and ability of creator to fix security vulnerabilities and so about disposition for security risks. That table shouldn't be removed only because some little used and marginal systems haven't been monitored for vulnerabilities by that sources. So please reconsider your verdict about delete table, if you want maintain neutral point of view about operating systems security (or more precisely one important component of it, maybe most important one). See computer security for more details about the importance of unpatched known flaws and Comparison of web browsers - Security for related topic with related table as this would be. --Ptomes 28 June 2005 10:28 (UTC)
I'm not concerned about the fact that some OSes are not monitored, for those the corresponding box could simply be left blank. I just don't see why OSes that aren't in the main tables should be listed in the security tables (e.g. Win2k and the Linux distrubutions (those can go into Comparison of Linux distributions)) --K. Sperling June 28, 2005 15:31 (UTC)
And about the general direction of this article: In my opinion, it already has a bit of a bias towards popular desktop OSes, and could do with quite a bit more information about server operating systems (e.g. Z/OS, HP-UX, ...). Maybe also embedded or handheld/phone OSes... Though while adding them would be nice, it also makes it harder again to do meaningful comparisons. --K. Sperling 19:12, 2005 Jun 26 (UTC)
How about this: We add a "Security" table that lists the same OSes in the same order (i.e. alphabetically) as the other tables, and make "Number of unfixed critical bugs" (e.g. moderately critical and above) one column of it, and maybe also the date of the oldest of those bugs. Other columns could be "integrated firewall", "encrypted filesystems", "stack execution prevention", "isolation of subsystems" (needs a better name I guess, I'm thinking of things like BSD jail or POSIX chroot here (though chroot provides only partial isolation)), and probaby some others.
I agree with that, but I think others editors should help with it. It seems good. I complain about add that table, but now we can only edit it. --Ptomes 4 July 2005 07:18 (UTC)
Ordering by bug number or any other contrived metric is bad, because listing any OS as number 1 is equivalent to saying "OS XYZ is the most secure OS". However, there is no such thing as "the most secure OS", just as there is not "the fastest OS" or "the OS with the nicest GUI". It all depends on how and in what contexts you measure security/speed/...
This article provides is a high-level overview, it is neither necessary nor desirable to include every bit of detail information about every OS. What's important is breadth of coverage, not depth. Just as the "Technical Information" table (hopefully) covers the most important technical aspects of an operating system, a "Security" table has to reasonably cover the subject of OS security. A mere bug count doesn't even come close to covering the subject of operating system security.
BTW, I'd also like to know some other editors' opionions about these issues! --K. Sperling June 28, 2005 15:31 (UTC)
Seeing as you've just reinstated the table, and again without any modifications, and without even responding here, i'm posting this on Wikipedia:RfC. --K. Sperling July 1, 2005 10:50 (UTC)

From RfC

After reading through the security table discussion and perusing the artcile, I concur with Sperling's position. Specifically, I agree with his idea of focusing on security features rather than number of bugs.

In its current state, the security table isn't as informative as it could be. There's far too many cells without data and the rankings aren't very helpful. OS security is an extremely complex subject and can't realistically be quantified or ranked. I'm interested in reading more opinions, but I'd recommend removing the table or modifying it to focus on security features. Carbonite | Talk 1 July 2005 12:13 (UTC)

"BTW, I'd also like to know some other editors' opinions about these issues!"

Are you really sure that you want to know? :-)

IMNSHO, this entire article is one of "those" articles. You know the ones I mean: Nuclear power, George W. Bush, Abortion, and the like where there are far too many partisans on all the sides of the issue for the article to ever converge into something even remotely truthful and usable by a naive reader. If it were up to me, we'd have a Wiki template that we'd apply to articles of this nature that said something like:

"Caution: Nearly EVERYTHING in this article is just someone's Point-of-View, and if you re-read it two minutes from now, it will just be someone else's Point-of-View (unless the entire content of the article has been temporarily replaced by "Xyzzy is Gay!"). Why not skip this article and go read about ((random page inserted here)) instead?"

Atlant 1 July 2005 12:16 (UTC)

Hehe, yeah, I would have to agree this article is (or at least has a strong disposition to be) one of "those". But since it's here and VfD-ing it would probably fail, one might as well try to provide some basic information in a half-way truthful manner :) --K. Sperling July 4, 2005 01:42 (UTC)