Talk:Linux/Archive 25

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

Archive 20

←

Archive 23

→

Flawed lead paragraph needs replacement

The lead paragraph calls "Linux" as an "OS", but then misleadingly mentions it as "an example of FOSS development". Since nearly all distributions of this "OS" contain non-free software (e.g., gNewSense being a rare exception) and since more generally there is such an enormous diversity of groups covered (everything from OpenOffice to companies making non-free add-ons), the "example" phrase is false and devoid of meaning. "Example" would be appropriate here if applied just to the free software subset of the "Linux" kernel (this kernel is always distributed with non-free files).

Similarly, the phrase "typically all underlying source code can be freely..." is also nonsense, since only very few instances (e.g., gNewSense) of the subject of this article have all source code that is freely modifiable, etc. Thus, it's untypical, not typical.

Perhaps a "Linux" enthusiast can replace the lead paragraph with something correct. Freed42 (talk) 04:44, 29 July 2008 (UTC)

Well, let's look at that sentence: "Linux is one of the most prominent examples of free software and open source development". I think the problem with this sentence is the word "example"; the sentence is being used to indicate that the OS has a prominent mindshare in thinking about FOSS, but "example" rather supposes that it embodies such values. Rather than nitpick over whether it does embody those principles, we could change this to reflect that the public perception is that it does. What about "Linux is one of the most commonly-cited examples of free software and open source development"?

The second one is even easier. Replace "all" with "almost all". Chris Cunningham (not at work) - talk 08:14, 29 July 2008 (UTC)

The second suggestion is indeed an easy fix. The first suggestion needs more attention. First, it is not nitpicking to note the nonsense of what is, in effect, "A massively diverse conglomeration of free and non-free development projects is an example of free software development." Not only do I question the appropriateness of the unqualified emphasis of how a falsehood is widely cited, but I also question the repeated implication in this paragraph that the results represent some kind of unified effort. It not only is false, but never was true. Freed42 (talk) 14:22, 29 July 2008 (UTC)

I don't know about that. Torvalds's aim was to create an operating system. GNU's aim was to create an operating system. Outwith a relatively tiny group of hypergeeks who would wage decade-long wars over the semantics therein, the OS is just as monolithic as Windows is (note that I'm picking Windows precisely because free software people typically treat it as monolothic when Microsoft geeks would be horrified at the suggestion). The intro to this article is a 10,000 feet look at the subject of "the free OS with the Linux kernel", which is all that most people know it as. We can elaborate in the core of the article. To nitpick too much in the intro is the downfall of many a scitech article on Wikipedia. Chris Cunningham (not at work) - talk 14:40, 29 July 2008 (UTC)

We can leave those semantics behind. Could you explain why "Linux" is as monolithic as MS Windows? How many distinct distributions from how many wholesalers of Windows can I order off the net or mangle and redistribute as I wish? Where are the charges of "monopoly" in "Linux"? I highly doubt that "most people" know what "Linux", "OS", or a "kernel" is. They might think of "Linux" as a company, some "computer system", "some kind of computer software that is incompatible with Windows", or "the competitor to Microsoft that is cheap or free of charge." I am having trouble seeing why a 10,000 feet look should be obscured by perpetuating misperceptions. Does such a look require baseless hype? Freed42 (talk) 17:31, 29 July 2008 (UTC)

If it were a "misconception" then we wouldn't still be arguing over what exactly the makeup of the OS is and trying to decide who the key vendor is, would we? The point is that to some degree there simply isn't a definite answer here. By trying to wedge one into the lead we do two things: we confuse readers with trivia they probably don't need at that stage, and we incite geeks (that's us) to bicker about things even more. Basically, the only compromise we have here is to provide a mushy half-answer which isn't likely to please a lot of partisans - but it's proven to be accessible enough for inexpert readers. Chris Cunningham (not at work) - talk 19:43, 29 July 2008 (UTC)

The misperception in this case is that there is some single thing--that's one of the things I keep trying to point out, i.e., your "no definite answer". Whatever I might have suggested was not intended to be a single answer to wedge in--that's why I suggested that a "Linux" supporter do something about it. Now how about something like "commonly perceived..."? It is on much safer ground since indeed most people have no clue that it would in fact be a wild and chaotic mix of free and non-free software development; they probably have zero clue about "free" and "non-free" anyway. It also subsumes "commonly cited" because if I perceive "Linux" as representative of FOSS, then I will cite it as such when the occasion arises. Freed42 (talk) 22:20, 29 July 2008 (UTC)

For the simple reason that we can't source "perceived", but we can source "cited". Not much more to it. Chris Cunningham (not at work) - talk 22:33, 29 July 2008 (UTC)

OK, thanks. I could work with "cited" but still see the need to briefly note the enormous diversity behind the "Linux" label. I'll try to come up with something. Freed42 (talk) 22:59, 29 July 2008 (UTC)

I don't see any "enormous diversity behind the Linux", Linux is the kernel and distributions that use it. What else is "Linux"? Did I miss anything? -- man with one red shoe (talk) 23:18, 29 July 2008 (UTC)

I was sloppy. "Linux" referring to an "OS" could be the kernel or the kernel plus anything installed or installable with the kernel, including any distro containing the kernel (at least hundreds of distinct "kinds" with any manner of different desktop environment, window manager, and myriad other defaults, etc.). Consider people referring to "Windows". Would these people also be thinking of the Oracle DB, Quicken, MS Office, etc. as part of the OS? I highly doubt that; they probably recognize that different *prices* indicate distinct products. They probably end up being far more precise than the average person speaking of "Linux". Saying that Windows XP is an example of, say, non-free software development is irrefutable. After all, if software is not free, it is non-free. The typical manifestation of "Linux"--even the kernel itself--is also non-free since it contains many non-free files (e.g., non-free firmware, etc.). There are rare exceptions (e.g., gNewSense), but as I have been repeatedly told and try to accept, they barely matter on Wikipedia. This non-free software is not some kind of triviality either: it exists in nearly every distro at high and low levels; NDISwrapper (for non-free drivers) still matters a great deal; more and more distinctions are being made between free (more of a goal than a reality) and non-free distros; and then there is the notorious example of Tivo's non-free "Linux" installations, in addition to a very likely growing list of embedded installations (also at higher layers such as non-free Maemo on some Nokia products) with non-freeness mattering a great deal. One might try to argue that when people refer to "Linux" they refer to something analogous to, say, "Windows XP". "Linux", however, is commonly indentified with "distro", for which, say, Windows XP, has no remotely comparable analog, so I reject this argument. Thus, I think that repeating a blanket, unqualified claim that the nebulous "Linux" is commonly cited as an example of FOSS development only worsens (and exploits) already muddy waters. Freed42 (talk) 05:30, 30 July 2008 (UTC)

(outdent) Hm. Are you arguing that a typical distro has non-free components in the default install, or are you arguing that a typical distro has non-free components readily available? The distinction is not trivial; a typical Debian-based distro such as Ubuntu usually has all the core as "free", with installers having to enable special repositories to get at the "non-free" components. As for the idea that there are non-free components in the kernel... ^{[citation needed]}. --Alvestrand (talk) 06:24, 30 July 2008 (UTC)

Regarding the non-free components in the Linux kernel, there are any number of inventories that list these. One of the more cited ones is here[1]. The gNewSense distribution has at least two files that together list most if not all of the non-free components of Linux kernel 2.6.24. Some non-free files (of a vanilla Linux kernel) are firmware images, some have non-free distribution terms.

"Readily available" is my criterion. We could easily argue that "readily available" is weaselly or a slippery slope. My counterargument is simply that it is an issue of norms: what characterizes this so-called FOSS development example that the label "Linux" is supposed to convey? What is the de facto policy about a committment to FOSS? "Readily available" non-free software conflicts with the "example of FOSS". Ubuntu is instructive. Is the typical Ubuntu user going to care about multiverse and restricted? In any case, the user will end up identifying it with Ubuntu, i.e., as a part of the Ubuntu OS, which, as so many are quick to remind us, here, is an instance of the "Linux OS". In trying to grab all the "good" for "Linux", it is unrealistic to ignore the "bad" out of convenience. - Freed42 (talk) 12:57, 30 July 2008 (UTC)

Well, no, it isn't "unrealistic". Opinions differ amongst those who actually work in the kernel as to what exactly qualifies as non-free once you get down to firmware, and as pointed out the current stable Ubuntu doesn't load proprietary drivers without specific post-install approval by the end user. In essence, you're now arguing that the potential to use proprietary software is what makes a system tainted, which is preposterous. gNewSense is exactly one shell script less tainted than Ubuntu in that sense. Chris Cunningham (not at work) - talk 13:54, 30 July 2008 (UTC)

I was careful to give "Linux" the benefit of the doubt, Ubuntu being relatively benign. I did not bother mentioning PCLinuxOS, Suse Linux, Linspire, or whatever. Second, "potential" is a direct consequence of policy here--that's why I focused on policy. If the degree of non-free stuff were not significant, the distros could just drop it all right now with no major problems, but we know those days are long, long over if they were ever here at all. Which distros (which, again, are identified with "Linux" which is identified as an "OS" ...) are committed to FOSS? It's not as if the non-free stuff floats by on its own and lands in a repo; this requires maintenance and policy. Not only are you misleading in this way, but your facts are wrong about gNewSense. Hundreds of hours of license auditing (results publicly recorded on wiki.gnewsense.org) have resulted in removal of non-free files beyond a mere shell script. People do not make that kind of tedious, thankless effort on a whim. The files are removed immediately and affected equipment stops working immediately (due to daily updates). Significantly different experience than Ubuntu, and gNewSense is just starting--they aim to audit every file of every package. Moreover, if "potential" were so trivial, we would not have gNewSense, Fedora, BLAG, Ututo, etc. Indeed, the "enterprise" versus "community" thing seems to be a growing trend.

Do you think the average user would identify non-free elements (whether "default" or "readily available") any less with "Linux" than free elements? What else would they see it as other than things that are more or less convenient to install? Adobe Flash is an important, ubiquitous example that I have should have mentioned.

Finally, consider business contexts. Those users also refer to "Linux", of course. Would they typically be any less dependent on non-free software on "Linux" than "consumers"? - Freed42 (talk) 15:09, 30 July 2008 (UTC)

This seems to be part of a disagreement about how carefully one should verify claims that stuff is allowed to be distributable under the GPL rather than a consensus agreement in the Linux community. See [2] for Debian's take on the current status. --Alvestrand (talk) 22:54, 30 July 2008 (UTC)

This old chestnut again. Enterprise Linux users typically pay people to work on the mainline kernel because it is vastly more economical in the long run than keeping their own proprietary bits out-of-tree as "value added" code. This has been the case for years. People who repeat the nonsense theory that the term "Linux" is promoted by companies who have something to gain by minimising the free software aspect of the OS typically don't know what they're talking about, and the FSF is sensible enough not to touch this argument with a bargepole as far as I can see.

As for your other points: firstly, of the three major distros at this point (Fedora and OpenSUSE: Xandros etc. certainly exist, but their share of the market is simply nonexistent next to the big three) two (Fedora and Ubuntu) use the "no unfree software by default" model now and the third is likely to go that way. Distro proliferation has little to be with ideology and a lot to do with how easy Canonical and Red Hat have made it to fork Ubuntu and Fedora. Chris Cunningham (not at work) - talk 09:16, 31 July 2008 (UTC)

I have no particular dispute with the observations that you make here. I should emphasize, however, that unless I specifically refer to the Linux kernel, the points I make are in the context of the subject of this talk page, the OS (ill-defined) called "Linux". E.g., it would probably be hard for me to direct most of my criticism at the activity around the kernel specifically, and if I did, it would probably involve the embedded market. The ambiguity around "Linux" and the ensuing problems are so predictable that I think I will just let this rest for now. Freed42 (talk) 16:01, 31 July 2008 (UTC)

Move page to GNU/Linux

"Linux" is the kernel. The proper name for the operating system is "GNU/Linux". Please rename this page to "GNU/Linux". See [3]. Nate879 (talk) 18:35, 17 August 2008 (UTC)

Please read the article itself and the 24 talk page archives listed at the top if this page to see why this request is problematic. - Ahunt (talk) 18:40, 17 August 2008 (UTC)

Please read the link I provided to see why this request should be done. Nate879 (talk) 23:51, 17 August 2008 (UTC)

It has been discussed here in every page of the archive, if you bothered to read the archive you'd have seen it. If you continue to ignore the archived talk pages you'll just give a bad name to GNU/Linux proponents, which I have to say wouldn't surprise me too much. -- man with one red shoe (talk) 23:56, 17 August 2008 (UTC)

Nate879: What you are quoting there is just one person's opinion. There are many thousands of other opinions on this subject. Quoting one person's opinion is not sufficient cause to rename this article. As well as reading the archives above you will want to acquaint yourself with the article GNU/Linux naming controversy, which is really where the bulk of this debate has taken place. - Ahunt (talk) 00:32, 18 August 2008 (UTC)

Might I just point out that while you are correct that it is one person's opinion, it happens to be the opinion of the person whose vision made the existence of a free operating system possible eight years before the name "Linux" was first used. Linux was a latecomer to the system; while this does not downplay its importance, naming the entire operating system after the kernel ignores all that the GNU Project did for the system before Torvalds entered the scene. Without GNU, Linux would probably never have amounted to more than a hobby project; without Linux, GNU/Hurd would still have existed, albeit a few years later on. I can understand the practical advantage of saying "Linux" colloquially, but for an encyclopedia "GNU/Linux" is a much more descriptive (and correct) name. Sjmcd (talk) 03:11, 19 October 2008 (UTC)

Nate879, please don't turn this into a protracted argument. The issue has been discussed exhaustively, and I can guarantee that if the best you can bring to the discussion is a link to GNU's web site, you aren't going to provide any perspective we haven't collectively spent years considering. Have some trust that this has been thought through. Warren -talk- 01:49, 18 August 2008 (UTC)

Good introduction

I sense that there are immense wars over minute details of GNU and linux and <insert long list here>; however, this is an encyclopedia article with a general readership - so mangling reason, perspective and the English language is an unacceptable compromise.

The introduction could benefit from a little tender loving the care; the sort that an English professor would give while grading papers. ...and perhaps less food fighting by everyone else. Blablablob (talk) 19:31, 29 August 2008 (UTC)

The introduction could benefit from a conceptual expansion - e.g. linux can and is installed on just about anything is more accurate and more helpful than a partial list. (why such emphasis on video games in the intro?). The support that companies give is varied: Oracle gives the client more for their money by not wasting it in the inefficient Microsoft economy. Red Hat and Novell maintain distributions, Nokia installs it on cell phones, and so on, and so forth... Blablablob (talk) 19:55, 29 August 2008 (UTC)

It is notable that many/most linux users are experts/programmers; and so linux has had an effect on software well beyond it's user base - e.g. php (and etcetera) ported to windows. —Preceding unsigned comment added by Blablablob (talk • contribs) 23:29, 29 August 2008 (UTC)

I've removed this paragraph for now because it is unsourced. We need this article to be based predominantly on arguments directly advanced from reliable sources, rather than extrapolated by experienced editors. I don't believe that the statement "many Linux users are experts" is as true as it once was, nor that it is important enough for the lede, and portability etc are issues with all operating systems and are not especially notable in the context of Linux. Chris Cunningham (not at work) - talk 08:39, 4 September 2008 (UTC)

the key term is many as opposed to most. many is still true. Your issue with notability is ironic, given that linux is itself notable in large part because of its effect outside its own user base. Be kind to the general audience in the introduction; arguments over minutia between experts are inappropriate when the general reader isn't supplied with the basics. I'll edit it in regards to sourcing. You'll have to expand on the portability issue for me to understand your point. Blablablob (talk) 09:11, 4 September 2008 (UTC)

I'm guessing that your point about portability is that linux programs are no more portable than other unix programs; but that's not the point, rather it explains that programs written in linux have an immediate scope beyond the rather small user base; this is not merely notable, it does much to explain the notability of linux to the uninitiated - it answers an internal question: why are they reading about linux and why does it get so much exposure despite a small user base. It is notable in that this is in stark contrast to Windows programs and Mac programs before OS X; for the general reader who is accustomed to Microsoft (a huge part of the population) this is notable - alternatively, now that people are becoming aware of cross platform software, linux is notable for its central role in that trend. As this article evolves to include the economic and historical significance of linux (e.g. free markets, innovation, competition), this will be relevant to that as well. Blablablob (talk) 09:39, 4 September 2008 (UTC)

This article has had a historic tendency to bite off more than it can chew and try to explain the entire free software movement to people. This isn't appropriate. The Microsoft Windows article did not, last time I checked, start with Gates's Open Letter to Hobbyists because of Microsoft's prominence in the rise of software as a product. This is the article about the operating system - that the OS has a key place in the free software movement does not mean it is the free software movement. The article already discusses free software and Linux's place in it in detail - while this could perhaps do with slightly more coverage in the intro, this does not warrant a whole extra paragraph at this stage IMO. Chris Cunningham (not at work) - talk 11:35, 4 September 2008 (UTC)

I agree on limited scope - especially within the introduction. But this sounds reactionary. Enthusiasts on all sides can and likely will run with scissors in the body of the article - but for the general reader, a quick bit about linux's relevance is appropriate. To say that `gasoline comprises a mixture of light hydrocarbons` is both correct and inadequate: `gasoline is used in automobiles` is important enough to put in an intro, arguably more important than its chemical composition.

Back to limits: the point is independent of FOSS, and limited: a specific, recognizable contribution to the general populations experience of software Blablablob (talk) 23:59, 5 September 2008 (UTC)

Mitochondria

I'm not hip enough to be on either side of the GNU/Linux versus Linux debate, and really it's not that important...

However, GNU and Linux have an intertwined history and the two have coevolved together. The last paragraph in the introduction does not reflect that - instead it looks like a compromise, and in that compromise rests an inaccuracy.

To simply say that modern eukaryotic cells and mitochondria both exist is not a good description of their relationship; they coevolved, they have an intertwined history. To cave into the demands of Mitochondria enthusiasts or Eukaryote enthusiasts would be unproductive, but so would caving in to them both. That sort of neutral falsehood is better suited to geopolitics than an encyclopedia.

So, for starters, linux is commonly compiled with gcc, gcc evolved to fit the needs of compiling linux, and the linux kernel is covered by a ... GNU license. And there's much, much more! But a sentence would suffice.

... and P.S. this really isn't part of the naming dispute; you can name the article whatever you want. This is about the relationship between the two. - Blablablob (talk) 08:07, 4 September 2008 (UTC)

I've tacked a sentence onto the end regarding this. Again, we really want it to be attributed directly to a reliable source. Chris Cunningham (not at work) - talk 08:41, 4 September 2008 (UTC)

I really don't want to twist your ankle over this - you seem unenthused. I'll rephrase the issue I see: unless you're aware of the controversy, the paragraph doesn't make a lot of sense. Therefore, as it exists because of the controversy, and a short bit about their symbiosis would likely placate both sides, and above all it happens to be true and relevant ... the bit about symbiosis would be a good idea. It would improve the article and it would make sense to the general reader. Anyone could write it up. Blablablob (talk) 09:52, 4 September 2008 (UTC)

And I don't disagree. However, it needs to be attributed directly to a reliable source. Chris Cunningham (not at work) - talk 11:40, 4 September 2008 (UTC)

In that case, would you like me to remove what you put in? Blablablob (talk) 13:32, 7 September 2008 (UTC)

Servers and Supercomputers

This section uses inaccurate data.

Reports from Netcraft change every month so the section should rather cite some statistics based on those reports. One can easily pick the month that shows the better data for his/her bias.
The citation from top500.org is wrong. As of August 2008 , out of the top 500 systems, 465 (91.6%) run Linux. 423 are those that run linux only. Other 36 run "mixed" os, which is Linux+some flavour of UNIX.[4]

The Linux adoption in supercomputing is way larger than that stated in this section. Update the text, please.--84.222.238.235 (talk) 16:00, 11 September 2008 (UTC)

Desktop PC Adoption

{{editsemiprotected}} In the introduction: ranging from embedded devices and mobile phones ... would now be common to include desktop systems as well >>>>>> ranging from personal computers, embedded devices and mobile phones ...

To me the way the sentence is worded ("Ranging from embedded devices and mobile phones to [[supercomputers]") strongly implies a range that contains personal computers. ~ mazca ^{t | c} 10:51, 14 September 2008 (UTC)

Not done per Mazca.--Aervanath lives in the Orphanage 18:46, 15 September 2008 (UTC)

In section 1.3, is it really appropriate to claim that home desktop and enterprise desktop have a 'significant' share of the desktop market? All except 1 of the 7 references claim a <= 1% share. One of them claims 1.34%. One of them (reference 14) is a link to BROWSER market share (I suppose it should be http://marketshare.hitslink.com/report.aspx?qprid=8 ).

Also, these stats are then discussed in section 4.4 where references 11-17 are duplicated in 51-57 - including the browser market share error. —Preceding unsigned comment added by 203.217.13.211 (talk) 02:52, 16 October 2008 (UTC)

Who writes the operating system

Here's an interesting presentation which includes some data about who contributes to the operating system:

http://www.kroah.com/log/linux/lpc_2008_keynote.html?seemore=y

With case examples, in order, of the kernel, GCC, X.org, and binutils:

. Gronky (talk) 10:17, 23 September 2008 (UTC)

This one's quite good too. Chris Cunningham (not at work) - talk 12:44, 23 September 2008 (UTC)

So I'm to infer from this that SUSE don't like Canonical much?--NapoliRoma (talk) 14:21, 23 September 2008 (UTC)

I think that the argument misses the point - there are lots of ways of contributing to Linux other than writing code for the kernel. That is important, but there are lots of other things worth doing too. Those who promote Linux uptake bring more people in, which results in more people working on the kernel as well as everything else. - Ahunt (talk) 14:26, 23 September 2008 (UTC)