Linux OS? Not GNU/Linux ? :)

[_wb_]

I don't quite agree: for most hardware, there are Free drivers available. Accelerated graphics and wifi chipsets are about the only category of hardware where binary blobs are still common - but it is certainly possible to have a laptop or desktop computer and fully use the device without needing any proprietary closed-source drivers, if you are a bit careful in your selection of hardware. Slowly but surely, hardware manufacturers seem to be embracing the idea that FOSS drivers are a good thing. See for example the list on http://en.wikipedia.org/wiki/Graphics_hardware_and_FOSS

 

[diligentcircle]

have to mirror the FSF's position that people viewing talking about freedom as extreme or annoying is dangerous. Freedom is important, even if it is of no use to you now. I talked to my brother about this when I refused to play League of Legends both because it is nonfree and because it is Windows-only (a huge blow in my opinion, and I need a damn good reason to want to play it in that case), and I also refused to play Quake Live because I don't see it as being much better than AssaultCube, and I consider the terms pertaining to copying to be too restrictive. He asked me why I cared that I'm not permitted to download more than 1 copy of Quake Live's software per computer, or that I can't modify the software, if I'm not going to do it anyway. My retort was: does he want the freedom to protest, even if he doesn't choose to exercise this freedom?


To clarify, for games, I'm willing to sacrifice some freedom, because it's just entertainment, not something essential or utilitarian. In fact, the only thing that will surely keep me from playing a game is spyware, DRM that prevents me from copying (for archive/backup), and DRM that controls the number of machines I can install or play the game on. But I prefer free (as in freedom) games over nonfree ones if possible. It depends on how interesting the game is, and Quake Live and LoL didn't look interesting enough to me (and I can't play them on the Pandora). For any type of software that has utility, though, I will only use nonfree software out of necessity; if available, I personally insist on using free software.

 

[moxie]

We're having midsummer celebrations here:


http://www.youtube.com/watch?v=8I5BGsK5ZAU


So I won't have time for a more eloquent reply than this at the moment: Exophase is saying exactly what I want to say. In particular, the case of GNU software vs X, Gnome/KDE... and the rather ugly use of very loaded terms from the FSF. The use of words like "Freedom" to mean not freedom, but "freedom as we define it, and all who do not agree is evil" is a particularily cultish.


Now, back to the pickled herring and schnaps.

 

[Exophase]

I don't quite agree: for most hardware, there are Free drivers available. Accelerated graphics and wifi chipsets are about the only category of hardware where binary blobs are still common - but it is certainly possible to have a laptop or desktop computer and fully use the device without needing any proprietary closed-source drivers, if you are a bit careful in your selection of hardware. Slowly but surely, hardware manufacturers seem to be embracing the idea that FOSS drivers are a good thing. See for example the list on http://en.wikipedia....rdware_and_FOSS

I can almost guarantee you that any laptop or other major computing device you choose will have microcontrollers that you are not aware of, containing flash and/or ROMs with code that you also have no idea about. I'm all but certain that applies to rms's Lemote laptop as well.


And when you're talking about software that either can't be modified, generally won't be modified, or at least will generally only be modified by experts who have a lot of special hardware and knowledge, then I don't really see the clear line between software and hardware. ASICs are generally designed in high level HDLs, and the choice to use one instead of an FPGA or a microcontroller with a conventional program is more of a cost decision than anything fundamental. And while there are people advocating for open hardware I'm hard pressed to find anyone applying their zealous, all-encompassing, moralizing, and exclusionary social mindsets to it. When rms talks about his Lemote I haven't seen him say "I am using the best I can but am campaigning for something better because they can become more open." No, I see him saying "I endorse this laptop", especially since it doesn't run Windows (a highly pragmatic viewpoint for such an idealist, but I think he picks and chooses on this), drawing a line in the sand so he can pick sides.


And of course this goes further.. I don't see the same zeal about putting all works of writing and art under Creative Commons or whatever the equivalent is. Advocated, yes. Considered an ethical requirement, no. People somehow talk about "Free games" but don't seem to levy the latter requirement; how many games really include open assets? Where open includes whatever you need to build it. A lot of game assets require expensive tools to create, is it unethical to use these tools too, because now the average person can't afford to modify the files even if they're allowed to?


You can call anything freedom and in violation thereof, it's pretty damn arbitrary. You could consider enforcing copyright infringement to be a violation of freedom to download copyrighting material (note this is not the same as advocating against utilizing copyrights). You could argue that disallowing vandalism is denying freedom of expression. You could argue that it's unethical to start a software project and make mention of it before releasing the first version - to me this seems to parallel FSF's demands. I see a logical extension of it to start insisting that a personal filesystem be open to the world at all times so everyone can be part of the full development process. It's easy to see how this begins to impinge on privacy, which rms is a large defender of.


When people use the word "freedom" in the context of social clauses I think of basically two things: civil equality and tolerance. The former amounts to applying the same standards to all groups without discrimination. The latter amounts to not punishing people for pursuing activities that don't interfere with you. I imagine most people think similarly. I'm sorry, but I don't see levying a requirement on software developers to allow third parties to modify their releases as enforcing freedom in any conventional meaning.

 

[diligentcircle]

I don't see the same zeal about putting all works of writing and art under Creative Commons or whatever the equivalent is. Advocated, yes. Considered an ethical requirement, no.

You really don't understand this? Games and art are only for entertainment or other pleasure or expression, so losing freedom with a game is far less of a loss than losing freedom with a tool.

A lot of game assets require expensive tools to create, is it unethical to use these tools too, because now the average person can't afford to modify the files even if they're allowed to?

It's a question of freedom and cost is a minor concern. But if you absolutely must use a certain expensive nonfree software tool, that probably means the format in question is patented or secret, in which case it shouldn't be used. If it's actually just a new format or a free tool to modify the format just hasn't been created due to lack of interest, you can develop a free tool for it yourself.

 

[Exophase]

You really don't understand this? Games and art are only for entertainment or other pleasure or expression, so losing freedom with a game is far less of a loss than losing freedom with a tool.

I've never seen FSF make a distinction between what the software is used for and whether or not it is acceptable for it be "free." And there isn't a clear distinction between what is and isn't used for recreation and discretion. Many tools are used exclusively or near exclusively for the production of arts and entertainment. And why draw the line there but elevate for instance the acquisition of knowledge/information, particularly in less practical fields? If you extend the distinction to luxury vs necessity you'll find very little software is written for health and survival.


Besides that, what does that have to do with groups who are critical about game code being open but not game assets? Game code isn't only for entertainment?

It's a question of freedom and cost is a minor concern. But if you absolutely must use a certain expensive nonfree software tool, that probably means the format in question is patented or secret, in which case it shouldn't be used. If it's actually just a new format or a free tool to modify the format just hasn't been created due to lack of interest, you can develop a free tool for it yourself.

Constantly repeating that it's a matter of freedom and that freedom always must be defended is getting no one anywhere. It's little more than a tautology. And I don't agree with you at all that the requirement to use an expensive tool is because it's using a secret or protected file format. Do you really think payment offers zero advantage in engineering time and quality?

 

[_wb_]

Yes, the line between hardware and software is somewhat arbitrary, and yes, the focus on software (instead of just creative works in general) is also somewhat arbitrary. The more things that are Free, the better. I don't think it is problematic that the FSF restricts its scope to software. There's other organizations like EFF and CC that have a broader scope.

 

[Exophase]

I don't think it's a problem to restrict your scope when promoting things positively. I think it is a problem when you restrict your scope while taking a zero tolerance and extremely critical stance within that scope. If your ethics are defined to include absolutely everything within a completely arbitrary dividing line then your ethics are completely arbitrary.

 

[diligentcircle]

I've never seen FSF make a distinction between what the software is used for and whether or not it is acceptable for it be "free."

FSF opposes nonfree games. I assumed you knew this, so I was talking about why it isn't a central focus of the FSF, and why some people (like me) make a sort of exception here.


With examples like Doom, where the levels, graphics, etc (the content) is the only part that is nonfree, it's still free software and useful to the community, as the Freedoom project clearly demonstrates. Plus, it's the Free Software Foundation.

 

[Exophase]

Yes, I understand that you are making that distinction, not FSF (I said as much), but either distinction (within software or within other sorts of creations) seems vague and arbitrary to me, when defining an all-encompassing ideology.


Explaining that the name indicates they're only concerned about software from an ideological point of view is just a circular argument. ie, "they're only concerned with software because they're the Free Software Foundation!" is the same as saying "they're named the Free Software Foundation because they're only concerned with software!" Both are trivial statements.


Similarly, "it's bad because it isn't free" isn't very removed from saying "it isn't free because it's bad."

 

[WizardStan]

You really don't understand this? Games and art are only for entertainment or other pleasure or expression, so losing freedom with a game is far less of a loss than losing freedom with a tool.

As the story goes, RMS came up with the concept of "free" software when faced with a printer that wouldn't work. He had identified a bug in the driver but, without the source for that driver, was in no position to fix the bug. If the driver were free to be examined and modified, he considered, then he wouldn't be in the position he was. He didn't lose the ability to print because of closed source drivers, he lost the ability to print using the printer printer that he was using.


A specific game serves a specific function, regardless of whether it is "entertainment" or not. When you lose the ability to run that game, you have lost the ability to operate that very specific function, play that specific game. A printer driver, on the other hand, serves a generic purpose. When one fails, it is well within your ability to go and find a different printer that will perform the exact same function.


This is why I do not like the FSFs philosophy: I can use a closed source device and be perfectly happy with that forever. If the driver somehow fails, as it did for RMS and his printer, I will chalk that up to the risk I took going with a well supported closed driver vs a questionably supported (and don't try and convince me the open source drivers are consistently better supported than the closed, that's a whole different debate) open driver. So long as a feature comparable option exists, I have no say in whether a company chooses to open its drivers or not.


RMS didn't want free and open printer drivers, he wanted free and open printer drivers for the printer he already owned. From that he evolved to his belief that everything should be open and on that I fundamentally disagree: so long as there is choice, you should be free to make that choice, and no one should force anyone to do anything, nor think poorly of someone because their priorities differ from your own. You believe that free drivers are important, only buy hardware with as free of drivers as possible. That is not my priority and suggesting that I am wrong for not holding that same priority is conceited. If there was any suggestion of a threat to choice, that perhaps the day would come when all devices become necessarily locked down, then I would be standing beside you to prevent that from happening, but as it stands that doesn't look to be the future. In fact, more devices have open drivers now than ever before. That doesn't mean the companies that haven't at least got some support for free drivers are absolutely wrong, it is just a different priority level, and you have the freedom of choice not to use them.


Hopefully I've articulated myself well enough that a rebuttal is not necessary.

 

[Exophase]

Well said Stan.


BTW, GNU actually does refer to the OS in contention as merely "The GNU operating system", not even "GNU/Linux." See for yourself:

GNU is a Unix-like operating system that is free software—it respects your freedom. You can install Linux-based versions of GNU which are entirely free software."

It's very difficult for me to feel sorry for these people.

 

[_wb_]

Well, the LInux kernel is not the only kernel you can use in a GNU system, and the Linux kernel is not part of the GNU project (unlike the Hurd kernel), so it makes sense to talk about the GNU OS, which can be either Linux-based or Hurd-based. If you would have read just a bit further down on that page, you would see that they refer to the "GNU/Linux operating system":

The Hurd, GNU's own kernel, is some way from being ready for daily use. Thus, GNU is typically used today with a kernel called Linux. This combination is the GNU/Linux operating system.

 

[Exophase]

If you would have read just a bit further down on that page, you would see that they refer to the "GNU/Linux operating system":

No actually, you quoted gnu.org, I quoted fsf.org (thought I put it in the quote tag but forgot to). The latter only refers to the GNU Operating System. I'm aware gnu.org doesn't.


It seems to me that the basic claim is that since GNU came up with the manifesto that "planned" the Linux kernel which then knowingly fulfilled the role. The problem is that the manifesto outlined mainly making replacements to a bunch of Unix software, which is not really as novel of an idea as they may think. It'd be like if years ago I had written something called the ENM manifesto that expressed a desire for an open source Matlab clone, then years later demanded that octave change its name to ENM (yes I know that octave is part of GNU, that isn't relevant ;p).. The idea would be kind of hard to claim ownership of..

 

[_wb_]

Well the first occurrence of the word "GNU" on fsf.org is in this quote:

We promote completely free software distributions of GNU/Linux, and advocate that users of the GNU/Linux operating system switch to a distribution which respects their freedom.

I cannot find your original quote on fsf.org, but I did find it on gnu.org, which is why I assumed you got it from there.

 

[_wb_]

It seems to me that the basic claim is that since GNU came up with the manifesto that "planned" the Linux kernel which then knowingly fulfilled the role. The problem is that the manifesto outlined mainly making replacements to a bunch of Unix software, which is not really as novel of an idea as they may think. It'd be like if years ago I had written something called the ENM manifesto that expressed a desire for an open source Matlab clone, then years later demanded that octave change its name to ENM (yes I know that octave is part of GNU, that isn't relevant ;p).. The idea would be kind of hard to claim ownership of..

I suggest you read the original GNU manifesto (http://www.gnu.org/gnu/manifesto.html). You claim this was not a novel idea - do you have historical evidence of earlier projects with the same kind of goals? And even if there is such evidence, the point is that the GNU project did not just outline these ideas, they also succesfully implemented them.


They only were a bit too ambitious in the design of their kernel (Hurd), which was to be a microkernel instead of the traditional and simpler monolithic kernels. For this reason, it took them a lot of time to get anything usable working, and meanwhile, the conceptually simpler Linux kernel was developed and gained a tremendous amount of momentum (mostly because it ran on x86 computers which by then everyone had easy access to, and because all other components of the OS were already there - thanks to the GNU project - so it immediately gave you a full OS), which meant that interest in further development of Hurd dropped almost to nothing (which is a pitty in my opinion, because technically it could be superior).

 

[Exophase]

I have read the GNU Manifesto, why would you think that I haven't? And no, I don't think that the concept of a Unix-like kernel written with the GPL is especially novel. Linus Torvalds was more inspired by Minix than GNU or particularly trying to overtake Hurd. The GPL itself is perhaps novel, but GNU is not defined as the complete collection of GPLed software.


Microkernels aren't conceptually more complex than monolithic, you can really argue the latter; the real historical disadvantage against microkernels is performance. I can't help but wonder if you think Hurd would have been superior simply because it's part of GNU.


There is no true GNU/Linux distribution. Almost no one outside of rms himself limits themselves to an operating system that is nothing more than GNU software + Linux. Calling it a full system is yet another product of circular reasoning - it's not because of any external criteria that defines what an operating system but the subjective feelings of what GNU feels qualifies to be part of its software collection. When this collection starts to include specific games (but reject every other non-GNU game) then things become especially silly. The original "Unix-compatible" intentions may have meant something, but really beyond a few pieces of software this gets pretty nebulous too, and clearly stopped being a baseline a long time ago.


There's also a historical precedent where every other operating system starts with development the kernel and grows around it. Seeing the kernel as yet another "piece" that is fine to be completed last is completely backwards from what everyone else does. Which is probably why people are comfortable calling it "Linux", because from Torvalds' point of view he wasn't making a kernel for GNU, he was making a kernel which GNU (and other) software was then ported to.


It is interesting that much of the GNU Manifesto is about the philosophies of the GPL and not the GNU project specifically. Which makes it sad that FSF has turned their efforts on lambasting GPLed software like the Linux kernel.

 

[diligentcircle]

There is no true GNU/Linux distribution. Almost no one outside of rms himself limits themselves to an operating system that is nothing more than GNU software + Linux.

I really don't get this reasoning. No, no one limits themselves to GNU software and Linux. No user of Windows limits him/herself to only Microsoft products, either. RMS doesn't limit himself to only GNU software either, strictly speaking, even if he doesn't use non-GNU software in practice. The X Window System is the most obvious example of a project which the GNU project used and embraced as part of the GNU system, and still does.


Speaking of Windows: imagine for a moment that all major programs in Windows were taken out and replaced; Notepad was replaced with gedit, Windows Explorer was replaced with Nautilus, Paint was replaced with KolourPaint, Internet Exploder was replaced with Firefox, etc. No more currently included Windows software on the surface. Would you then insist that it is silly to call the OS "Windows"?

 

[_wb_]

I have read the GNU Manifesto, why would you think that I haven't?

Well, you summarized it as "the manifesto outlined mainly making replacements to a bunch of Unix software", which kind of ignores the main points of that manifesto, which are the reasons why they want to do that. Making replacements for existing software was not a novel idea at all at that point in time, the commercial Unix systems basically did that all the time; the novel thing in the manifesto is the emphasis on freedom and being open source.

And no, I don't think that the concept of a Unix-like kernel written with the GPL is especially novel. Linus Torvalds was more inspired by Minix than GNU or particularly trying to overtake Hurd. The GPL itself is perhaps novel, but GNU is not defined as the complete collection of GPLed software.

No, but I think we agree that coming up with the GPL is an important contribution from the GNU project.

Microkernels aren't conceptually more complex than monolithic, you can really argue the latter; the real historical disadvantage against microkernels is performance. I can't help but wonder if you think Hurd would have been superior simply because it's part of GNU.

No, I don't think the GNU label by itself makes any difference to the technical quality of software. I do think that the architecture of microkernels has certain advantages over that of monolithic kernels, provided the performance issues can be solved (which I believe is the case).

There's also a historical precedent where every other operating system starts with development the kernel and grows around it. Seeing the kernel as yet another "piece" that is fine to be completed last is completely backwards from what everyone else does. Which is probably why people are comfortable calling it "Linux", because from Torvalds' point of view he wasn't making a kernel for GNU, he was making a kernel which GNU (and other) software was then ported to.

Yes, I understand that point of view and to some extent it is of course valid. However, from the point of view of the GNU project, the kernel really _was_ the final "piece". Of course not just another piece, a quite central and important one, and a very fundamental one, I would say about as fundamental as a compiler/linker. In order to honor both viewpoints, I prefer the name "GNU/Linux", not "GNU" or "Stallmanix", not "Linux", but "GNU/Linux".

It is interesting that much of the GNU Manifesto is about the philosophies of the GPL and not the GNU project specifically. Which makes it sad that FSF has turned their efforts on lambasting GPLed software like the Linux kernel.

I don't think the FSF is lambasting the Linux kernel, in fact they are promoting it. The only critique I know of by the FSF is that they don't like binary blob drivers that some distros include in their kernels, but those blobs are obviously not GPLed software, and I pretty much agree with the FSF in that we should pressure hardware vendors into giving us proper drivers including sources.

 

[gomocecuaist]

If FSF has in fact convinced you that your distribution is nothing more than a selected configuration of GNU software and the Linux kernel then they're accountable for every bit the unfair misdirection you blame on the distribution managers, kernel developers or Linus in particular for calling it "Linux." Actually probably more, since they're actively lobbying for something rather than just "implying" anything by a name. But frankly I'm more willing to just blame you for the misunderstanding.

There is no true GNU/Linux distribution. Almost no one outside of rms himself limits themselves to an operating system that is nothing more than GNU software + Linux. Calling it a full system is yet another product of circular reasoning - it's not because of any external criteria that defines what an operating system but the subjective feelings of what GNU feels qualifies to be part of its software collection. When this collection starts to include specific games (but reject every other non-GNU game) then things become especially silly. The original "Unix-compatible" intentions may have meant something, but really beyond a few pieces of software this gets pretty nebulous too, and clearly stopped being a baseline a long time ago.

The GNU project is a project to provide a complete free operating system (it was originally intended to be a complete replacement to Unix). Torvalds principally authored the Linux project. Some people had the bright idea of combining the Linux kernel together with the incomplete GNU operating system (the incomplete GNU kernel was the missing final piece) to form a complete OS. The Linux system together with the GNU system form the principle components of such a system.


GNU+Linux by itself is a full system; it is a complete replacement to Unix (Unix is considered a complete system). GNU+Linux are normally used as a basic platform for further work. As far as I know, no other software systems are as fundamental (to the system as a whole) and contribute as much work (in terms of lines of code) as the GNU system and the Linux system.


The idea is: RMS asks (dictates) to credit the GNU part of a GNU+Linux system in reference to the complete system. Yes, there are plenty of other non-GNU and non-Linux contributions to the system. These other software are not credited as belonging to the GNU system or the Linux system. I hypothetically propose the name "GNU+Linux+other software" for the name to the implied concept of the other software that exist in a GNU+Linux system. While it is correct to credit this other software, none of these other projects are as fundamental to the whole system as GNU+Linux. When a problem in the system occurs (that is not a user error) we can assign fault to where it belongs - it belongs to GNU or Linux or another project.


The Linux/Android system is an example of a Linux-based system that doesn't rely on GNU so it'd be erroneous to credit GNU in the case of Linux/Android; GNU is not fundamental to this system. A Linux/Busybox/uClib system does not imply the GNU/Linux/Busybox/uClib credit when GNU is not a principle part of the system. While it's not erroneous to credit GNU/Windows to a Windows system with the GNU system installed, Windows is already complete with or without GNU so the name GNU/Windows is not necessary. If one would modify Windows to become fundamentally dependant upon GNU, it would become a different case.