So what you are saying is that Red Hat bundled 2.4 before Linus decided it was ready?
No, that's not what I said. I said Red Hat 7 has support for kernel 2.4. There is a 2.4-pre kernel on one of the discs, but it's never installed by default. What is used in Red Hat 7 is a 2.2 kernel. In fact, if you read my link, you will notice that it clearly says that Red hat 7 includes kernel 2.2.16. But you didn't read, did you?
Why does that not surprise me? *cough* GCC *cough*
I know precisely what communisim is. But i am telling you that the perception in the business world is that people who support a completely free software world is communist. Right or wrong, thats the perception.
I agree with you here. I believe this perception is utterly lame and wrong, but I have to agree that it's common. At least in the US, it seems.
About MS, what is there market they have a monopoly in? Tell me that. OS's you say, thats typical. Thats what everyone thinks But tell me, do they have a monopoly?
Yes. Also in the web browser business nowadays.
YOu say they are because they dont open docs, api's or whatnot. That is not a legal requirement.
Well, it could have helped their case if they had done that. But now, this wasn't the situation, and hence they are accused of using monopoly tactics (not opening specs to the competition can be such a tactic, yes).
Yes, they should, but no, they dont have to. Ever.
I think it clearly shows that you don't understand laws against monopoly tactics. Yes, if you are in a monopoly position, you might very well be required to not further crush the competition by not providing for example specs or equal access in other ways.
Besides that, you and I know that there are other OS's out there. Why dont they package ever Compaq with linux? Why not? Because no one wants linux.
People think linux sucks, or is hard to use. Right or wrong, thats the perception. Compaq, and its customers, want the features that MS offers - compatibility, software support, ease of use, etc.
The compability Microsoft provides is with other Microsoft products. Other types of compability has never been much of an interest to Microsoft.
Like it or not, thats true. Compaq could package Linux or FreeBSD with its computers, but doesnt
They do. Perhaps you should check your "facts"?
because sales would take a hit.
Gee, so providing more alternatives than just Windows preinstalled would make sales drop? What kind of "logic" is that? If you provide more alternatives, your number of sales either stays the same (in the worst case, because noone buys the new alternatives) or grows (because the new alternatives attract new customers). I think that the fact most famous computer vendors now happily ship computers with Linux proves that this segment is indeed also lucrative.
In fact, everyone who sells computers with Windows does so because it guarantees a certain level of compatibility.
Compatibility with other Microsoft products.
You cite Apple, Netscape, etc as being proof they are monopolies. The fact is, Netscape's product was made useless by MS, by bundling. Bundling is mostly and should be legal.
Gee, in this case maybe it isn't, because that might be considered an illegal monopoly tactic, given the situation. Once again, I think you show that you fail to understand anti-trust laws.
This move helped consumers
Reducing choice by surpressing the competition benefits consumers? How?
Browsers are now standard issue for every major OS. All of them. Before, you had to pay for them. MS helped us all there.
Yes, browsers are now standard issue. That does not mean that they have to be considered part of the OS (office software is standard issue, but also not a part of the OS), nor that we wouldn't have this situation anyway. The reason web browsers are now standard issue and popular apps is because the growth and popularity of the web. The reduction of web browser choice for the consumer on the major client platform is pretty irrelevant to the web explosion, I think.
The fact is, that MS is not, by any reasonable defintion a monopoly in OS's. They cannot control the market place for OS's, they cannot force a standard down everyone's throats (despite best efforts).
I think many people who have witnessed the home computer industry the last 15 years will disagree with you here. Microsoft has on numerous occasions tried to control their platform, and in many cases to reduce competition.
Just because they dont use 'industry standard' protocols or api's doesnt mean that they are breaking the law. Thats their choice.
Yes, it can indeed mean that they are breaking a law. That's what anti-trust laws are about. Some action that would otherwise not be illegal can be illegal if you are in a monopoly situation in your particular market. These regulations are to benefit the consumer by helping further healthy competition not to shrink below the low level that it already is in most monopoly situations.
Futhermore, I am willing to bet that you are mostly or completely MS free. How could that be if they really were a monopoly? And even if you are not MS, they are dozens, hundreds, probably thousands of Slashdotters who are. If MS were a monopoly, then this would be impossbile in such large numbers.
I think there is some misconception of the term "monopoly" here. You don't have to be the sole company in the market to be a monopoly. Being the by far biggest company is enough. Under certain circumstances this is a crime, but not always. Such circumstances can include using your existing monopoly position in one market to gain a monopoly position in another market. That's what the Microsoft trial is about.
Your second comment makes no sense. What are thousands of Slashdot users against billions of software users world-wide? And don't forget, a majority of Slashdot users use Windows. This fact has proven itself on numerous occasions in polls, and CmdrTaco has mentioned it at least once.
That sounds like a strange way to solve things - I doubt that egcs 1.1.2 will compile a lot of C++ successfully, for example. A better solution might be to just
update gcc in RH7 to gcc-2.96-69.
They don't have to think if they don't wan't to.
Both Red Hat and Mandrake provide pointy-clicky-fancy updaters for fetching and installing latest updates. Perfect for those who don't want to think, they just have to remember to run it once in a while.
IE wasn't built into Windows until Windows95 OSR2. Windows95 OSR2 was never sold to the public, but only installed on OEM systems.
So I think your conclusion that Netscape's web browsing revenues increased that year is based on false grounds - because many customers did not have access yet to a Windows version which should make them uninterested in another web browser and the rest of the web browser competition merely by it already including a web browser already.
My problem with the GPL is that it uses force to ensure that everything is as Stallman sees fit.
I think this is a very simplistic view. When you say this you basically say that everyone that uses GPL is unaware of the contents of the GPL and basically just a mindless drone that can't think for himself. I would have to disagree with that... In my experience, people that choose the GPL choose it because of the copyleft clause and because they like that model, not because they are unaware of it.
Notably: the whole Qt thing - the wording of the GPL was massaged to condemn a project (KDE) which was created in the spirit of Free Software.
Sadly, "in the spirit of" is often not enough when it comes to licenses. Licenses are not designed to be misinterpreted and sadly you can break a license even when you have noble goals. In my opinion, that was what happened. I believe that was RMS' opinion too. He never said that the KDE project was not intended to be free software or that it was evil in some way - he just pointed out a licensing error that should probably get fixed.
The GNU-approved GNOME thing uses a license (the LGPL) does not encourage Free Software. That makes no sense to me.
Of course it encourages free software. The majority of software in the GNOME project is GPL - it is basically only important libraries that are LGPL. The reason is that people that have a serious problem with the GPL and don't want to use the GPL (for example they want to use the artistic, Apache or BSD license) should still be able to use and link to GNOME code.
The KDE project uses the GPL for everything, which is perfectly fine, also according to RMS. The reason there was a fuzz about KDE was that GPL programs in KDE linked to non-GPL code that was not part of the system (QT), something that is not allowed in the GPL, unless you make an explicit clause for that yourself. That was the only problem. Sadly, people insist on misunderstanding that all the time.
I don't trust Stallman - I think he's a nutcase with his own private adjenda.
He has an agenda, that part is true, but that agenda has been public for 16 years or something like that.
I use the Artistic license because it doesn't have the nasty undertones of the GPL, or the hassles of the LGPL. The LGPL is just lame - it doesn't ensure or encourage Free software, it just forces you to use dynamic libraries, which is a problem when there are so many unstable libraries out there.
Use the stable libraries then... Your argument does not make much sense, previously you claimed that you disliked not being able to reuse code, and here you suddenly claim that you don't like using libraries.
I wish we had face-to-face bandwidth to discuss this - it's just too difficult to explain the problems with the GPL and LGPL in little snippets.
Yes. I really recommend you to go hear RMS speak some time, and ask him questions. That might make things clearer. He is a very good speaker, and I know people who still do not like the concept of the GPL but now at least know the reasons and motives behind it.
Take the NVIDIA driver GPL infringement, for example. About 10 lines of GPL'ed code was used, and for that, everyone was screaming that NVIDIA had to GPL their entire codebase.
No. In a license conflict, there are always two immediate solutions to solve it - either remove the offending code or make the conflict go away in some other way. NVIDIA chose the former solution. And stating that "everybody screamed that NVIDIA should LGPL their entire code base" is a pure lie.
Yeah, right. That is force, just like Microsoft uses.
You clearly have trouble understanding the concept of "force", or at least really mis-use that word. NVIDIA weren't forced to use those 10 lines of code, but in doing that, they had to agree to the terms for using that code. They did not, and hence they had to remove that code. What could be simpler?
The idea of Open Source is code-reuse. The GPL prohibits code reuse for any non-GPL project (proprietary or not). If you use, or link to, any GPL code, you have to use the GPL. That's force, baby, not freedom.
No, it's not force. You're free not to use the GPL code. The terms for the use of the code are clearly stated in the license before you even think about using it.
So, you bought a card with binary-only drivers?;) 3dfx didn't open their drivers until well after the V3 came out.
I bought this card when I was still a Windows user. 3dfx' announcement later on made it a really good investment though.
They never promised open specs or open drivers.
That is simply not true.
There is no obfuscated code anymore. XFree would not allow the obfuscated 2D code into the distribution.
In case this wasn't clear, we're speaking about 3D drivers here.
Umm, the drivers are free (as in beer).
Geez, when I say free, I mean in the freedom sense. I thought that would be clear.
Just like the DRI drivers, there is some lag in getting the kernel modules working when the latest devel kernel comes out (admitedly it's a bit longer than with DRI)
That is really a funny understatement.
but you can still use the NVIDIA drivers on 2.4pre kernels. Where are the extra sales going to coming from?
The only strong argument against the closed drivers is diverse platform support. BSD, Alpha Linux, PPC Linux, etc. get left out in the cold, which is unfortunate. But, from a business perspective, there's no money there. I ask again, how is it to NVIDIA's advantage to open their drivers?
All other platforms than Linux/x86 are a part of this, yes. But more important is that even many
Linux/x86 platforms are unsupported - for example SMP systems. SMP systems are not uncommon today, on the contrary. NVIDIA is simple not an option on these platforms.
No, it's not a silly argument. The difference between what you can release in-house and what goes to the outside world is huge. You need a legal team to go over everything with a fine tooth comb, you need to have professional tech writers to make everything clear - everything has to be perfect - the docs are a *product* and have to be treated as such. If they're in-house, the engineers just walk down the hall and ask someone a question. There are legal liabilities when you release specs.
No, it is still a silly argument. Standard legal disclaimers still apply. It's not like this has not been done before! The docs are not a product that you sell - the docs are specifications, provided as-is, with a standard legal disclaimer with reservations for errors and omissions.
Besides, if the development on NVIDIA is not chaos with people running across everywhere, they already have maintained, in-house docs, with hardware functions and interfaces clearly documented.
Not if you can't understand it. How can you fix function A if you don't know what function A is supposed to do? Without the specs, it's pretty darned hard to write a good driver. Case and point: there are unobfuscated open source 3D drivers for NVIDIA cards in the Utah GLX source tree. They have limited functionality, but the code is there. Without specs, they aren't going to get any better, though.
You're seriously twisting words. We spoke about theoretical fully open drivers here, with full functionality, since the author had full access to hardware info under NDA. This is not at all the case with the existing free NVIDIA drivers - they have limited functionality as you say, and there's noone that can improve their funtionality because noone outside NVIDIA has access to full specs, not even under an NDA.
That is clearly different to my example where some free software developer has access to all specs under NDA and can thus write a fully functional free driver.
My argument boils down to this: if drivers are stable, full featured, and freely available, I don't need source code. The NVIDIA drivers are stable, full featured, and freely available. Simple as that.
No, it is not as simple as that. They are not fully featured, people claim them not to be as stable as you say they are, and they are not free (as in freedom). This means that they do not have the quality they could have had, they will not work on a lot of systems that don't exactly meet the criterias of NVIDIA, either with hardware or software, and future support for these cards are unclear at its best. All of these problems would be solved if the fully functional drivers where open.
Simple as that.
Code would be less non-free if there were exactly zero restrictions on its use.
I have nothing against BSD (please no BSD flames), but I guess that the popularities of BSD vs. Linux today does prove you wrong. We have a free software boom today that is caused in large parts by the very GPL we're discussing (again, I'm aware of the wonders BSD-style software have caused too). My point is that we have more free software today because of the GPL, among other licenses (but the GPL is still the most popular).
The GPL preserves freedom. We wouldn't have more free software without it, on the contrary.
But in a nutshell the thing that bothers me about the GPL is that none of that code can be used in a commercial setting. Most programmers need to work for a commercial company of some type in order to feed themselves. If I can't use GNU code in commercial code (even Open Source commercial code), I need to duplicate it, which goes directly against the Open Source concept. It's the licence and the leader that I have a problem with, not the concept. But I digress...
Of course it can be used in commercial development. Don't you think, for example, that developers in Linux distribution companies get paid? If you mean that it can't be used in proprietary software, then you're correct. But that is the very intent of the license - it avoids the situation where "improvement feedback" is made impossible due to improvements becoming non-free. That aspect of the license is a feature, not a bug.
If you want to use free software in proprietary software, you can go two ways: Make sure that the free software you use is either BSD-style licensed software, or LGPL.
If it is LGPL you can use it in your proprietary software, as long as you ship the source code for the LGPL part somewhere. But the rest of your application can remain proprietary. Why do you think that many software libraries in the Linux world are LGPL?
So what it boils down to is that different licenses have different uses. GPL is simply not intended to be used in proprietary software, so that's not a bug, and I think it is very strange to think that that would be the case. Also, complaining that you have to "duplicate" the GPL code in your proprietary application is also very strange. The original author can choose whatever license he wants, and you should respect that decision. After all, he has given you the right to use his code to begin with, and since it's his code, he can set any rules he want.
If you think a 3D card is a good investment, I have some nice ocean front property in Wyoming for you;) I can't think of any other piece of hardware that depreciates in value, or becomes obsolete, faster than a 3D card.
No, please misunderstand me correctly. What I said was not that 3D cards in general are good investments. What I said was that one particular 3D card can be a better investment than others, given certain criteria, like for example that the lifetime of the investment is restricted to the point where the hardware becomes obsolete and not only to the point where a single company does not bother providing drivers anymore.
I still have use for my old Voodoo2, for example. That card was a good investment.
Also, remember that 3dfx supplied binary only drivers for years before they opened up. I'd like to believe that a similar thing might happen with NVIDIA some day.
Yes, I also like to believe that! But insofar there have been nothing but empty promises from NVIDIA. You're right, 3dfx were equally bad until they opened up, but in the very moment they opened up and provided free, good drivers, I knew that the hardware I would by would have life-time driver support. If NVIDIA were to fully open their drivers (read: no obfuscated crippleware) one day, no one would be happier than me to recommend NVIDIA to everyone.
Perhaps if you could make a point where it would be to NVIDIA's advantage to open their drivers? What's in it for them?
Appearantly you didn't read what I wrote in my previous post. I'll quote: Free drivers makes more people able to use the hardware, hence more hardware sales. Very simple.
Publishing open specs means they have to dedicate resources to creating the documentation, proof reading it, maintaining it, etc. This is actually a *lot* more work than it sounds like.
Oh geez, it's not like they don't have any in-house documentation already that they have to proof read and maintain. Providing documentation to outside developers is not fundamentally different than providing docs for in-house developers. That's just a very silly argument.
What I was saying is that you should support vendors who have open drivers , if it's a big deal for you. Be a positive force, not a negative one. Don't bash NVIDIA, praise the companies that do have an open policy.
Oh, I do support vendors with open drivers, and thus I have not, and don't plan to, buy any NVIDIA stuff in the forseeable future. I also advocate and praise the companies that have open policies. But you seem to forget that this whole thread started out because NVIDIA will buy 3dfx and people are worried that NVIDIAs existing closed driver policy will prevail, instead of 3dfx' open one. And you can't discuss the aspects of that without explaining why NVIDIAs way is a bad one.
Actually, I think people should be putting pressure on ATI to release specs for the Radeon. There is an Open driver in the works, but the specs aren't public, they're only available under NDA. This means that you'll have the source code to the drivers, but you won't be able to fix them.
Excuse me? If you have the source, you can improve the driver. Yes, NDAs are bad, but in this case they are only there to protect details of the hardware design, they are not there to obfuscate driver development. The drivers will not be obfuscated. This driver development method is common in the free software world, if the hardware company is worried about sensitive details of their hardware becoming public. A potential free software driver writer can then sign an NDA to get all the gory details that in-house developers get, and write an optimized, non-obfuscated, open driver that can be used in free operating systems.
ATI gets the good press of having open source drivers, but the users don't get the info they need to fix the drivers if they have problems.
I think you are confused. If the driver is open, it can be optimized, problems fixed, and features added. If you want to have access to all the details to improve it even more, you too can ask to sign an NDA if the company requires that.
This is still very differnet from NVIDIAs policy. NVIDIA does not even allow outside developers to sign an NDA to get all the details they need for driver development.
However I will always feel that it would be great if the GNU people would realize that they are not writing "free" code
How is it not free?
You're free to use the code as you wish
You're free to modify the code as you wish
You're also free to redistribute the code. The only conditions are with the redistribution, and the only "restriction" is that you make the source of your changes availiable. You don't have this restriction if you don't want to redistribute it.
In fact, it is not a restriction but a grant of rights. Basic copyright does not allow you to redistribute other people's work, but the GPL does just that, as long as you meet the very simple requirement of keeping your source freely availiable too.
...and they are in fact writing code that will never be of as much use as it could be because software will never be "free". There will always be commercial software. Always.
I think you're confusing stuff here. Commercial software can be free software on the same time. In fact, most Linux distros are.:-)
And yes, maybe there will always be non-free software. But what has this to do with that?
And it would be nice if the companies could benefit from the technology in the GNU stuff without having to worry about releasing the source and other licensing crap.
They don't need to worry about it unless they want to redistribute it. They can use all GNU tools in the world inside the company for their production. But if they redistribute it, for example as part of their product, they also have to make their changes availiable. They then have to follow those very rules that the software was given to them to begin with. What's wrong with that?
Besides, how hard is it to set up an ftp or http server? I think there are bigger hurdles than that in software development.
I've always thought of the GPL and pretty much everything RMS as the biggest hypocracy in the universe. "We want to make the code free and support open-source by putting lots of restrictions on what you can do with our code and FORCING you to make your code under the exact same license we use." Hah. What a crock. I despise the "Open-Source" movement as it currently exists.
Heh. Clearly you have no idea what RMS supports and what he does not. RMS supports free software (as in freedom), not open-source. Go listen to RMS someday, then you will know that, and you will also know that he is not a hypocrite. Then go back to this discussion.
The GPL does not force anything. It grants you the right to redistribute other people's work. The only condition is that you have to keep the changes free, as the code was given to you to begin with. I clearly cannot see what's wrong with that.
This may be a new concept to you, but when you copy software the original copy does not get destroyed. Imagine the possibilities: I can write software and give it away, but STILL keep it myself!!
This may be a new concept to you, but when some people write software on their spare time and let people all over the world use it, they might do it of the reason that it remaining free will help them improve the software in the future. They want the software to be able to cycle in a circle of development, improving over time to the benefit of everybody. They want contributions to remain open so that they can incorporate all improvements done by others back in their original tree. This makes the software evolve, merely by people using other's people work, hack it according to their needs, and putting their changes back online.
If the license does not enforce freedom, this chain is broken. The original author writes a piece of software, a company finds this software and adds lots of interesting features and turns it into a closed-source product. Will the original software ever be able to use those additions? No. Will the freedom that the software initially had, and the closed-source company used, be of any benefit to all others, ever? No.
The GPL is about helping software improve by keeping it free, as the freedom was what made the code happen in the first place.
Imagine the possibilities: I can write software and give it away, but STILL keep it myself!!
Maybe I didn't write it just for me and just for a particular moment in time, but I actually use the stuff and want to improve it, and want the rest of the world to also benefit from all improvements. Now if the rest of the world does not allow me to use their improvements of my software, how can I possibly add all improvements? How can anyone be able to use any improvements at all except their own?
Your argument just doesn't make any sense. If MS were to use some of the Linux kernel in Windows, the Linux kernel would still be free to use how ever you wanted? What would anyone lose by MS using Linux w/o releaseing their code changes?
Believe it or not, but as much as I dislike Microsoft, I do believe they can also do good stuff. Now if they fixed bugs in my code I had spent months trying to fix, or added features, but they didn't release their changes, yes, I would lose. I would lose the possibility to improve my software to everyone's benefit because of some company's refusal to let the improvements benefit anybody but themselves. So in other words, they used the freedom I provided them with to use my stuff and do improvements, but they won't give me any freedom back in return to use those changes, and let everyone benefit.
The GPL exists exactly of this purpose - to help code remain free, to everyone's benefit.
I've been using Linux almost exclusively for the last 5 years. With each day, I become increasingly frustrated with the GNU mentality. I belive strongly in Open Source, but I also strongly believe that the GPL will be the downfall of Linux. You can't have an industry without profit, and Linux will never be mainstream without industry support.
Why frustration over GNU? It was GNU that made everything with Linux at all possible. And what this has to do with profit I simply don't understand; we are talking hardware and drivers here, and the drivers shouldn't be the money-making thing. The hardware should. Free drivers makes more people able to use the hardware, hence more hardware sales.
One argument I keep hearing is support for old hardware. I would argue that in the case of 3D gfx cards, this is not nearly as important as having drivers available when the card is released.
To me the matter of it being a good investment is more important. I can wait with the buy until I know a good (free) driver is out there, but I can't rely on a single company telling me after a year that driver support will be discontinued, making my hardware crap.
The former makes me able to do a good investment, the latter makes me lose money in that I have to buy new hardware, no matter if I want it or not.
NVIDIA is giving me the support that *I* need as a graphics developer. That's not to say that their support will satify everyone - for those people who don't want binary only drivers - don't use them.
Well, it's not so much about binary drivers as it is about lack of open specs that allows a competitive free driver. If NVIDIA only provided binary drivers but also released open specs, it would be as easy as you say "not to use [the binary drivers]".
The NVIDIA drivers are enabling Linux to penetrate new markets and allow people who *don't* have a problem with binary only software as long as it is high quality and stable to have great OpenGL support. A lot of people bash Microsoft for forcing vendors to do things their way. So why is the Linux community trying to force NVIDIA to do things their way?
Those new markets are the markets of the managers that have no clue what Linux is or why it is like it is. They don't understand that the freedom enforcing is what made GNU and Linux happen in the very first place, and hence they don't understand the interest in keeping it that way.
And the analogy of forcing I do not understand - Microsoft is a single company, the Linux community is a community of users. Microsoft forcing vendors is thus indeed forcing, but the Linux communitys criticism of NVIDIAs acts is market pressure. If you think NVIDIA is somehow "forced" by bad, mean Linux users, I feel sorry for you. It is in their best interest to listen to their customers, but they are not forced. If they had been forced, we would have free NVIDIA drivers long since.
Until the open source drivers are at least as good or better than the NVIDIA drivers, I really don't think there is a very convincing argument against NVIDIA. I will always choose the best technical solution. If the open and closed solutions are equivalent technically, I would be inclined to choose the open solution. But moral superiority does not outweigh technical superiority in my opinion.
Your argument is skewed in that NVIDIA does not allow free drivers as good as their binary ones. They do not allow this because they don't release the specs, and this is the very fundamental issue.
Perhaps some day, people will understand that the GPL does make code free, not just for now, but for ever.
Freedom is something which can be enforced, and if it is enforced, it will always be freedom.
Yes, there's a difference between "force" and "enforce". You are never forced to follow the GPL, only if you accept GPL software to begin with. The original developer granted you freedom to begin with under the promise that you will also give that freedom away to others as it was given to you. This is what the GPL enforces, it does not force you to do anything else. To me that is very fair. You can't blame the original author for wanting his work to remain free. If you didn't like that, you shouldn't have accepted it in the first place. You're not forced to use or modify GPL software at all.
But what community would NVIDIA be receiving "goodwill" from? Only the most extreme GNU 'not free as in beer' evangelists would appreciate NVIDIA open sourcing their drivers.
No. Every computer user (not just Linux user) will benefit by even better drivers. Why do I say even better? Because then, every person in the world with driver knowledge and incentive can help with improving them further.
I believe most linux users that own NVIDIA are *more* than satisfied, being that they can run 3d applications, and even Quake 3, and comparable speeds to Windows boxes.
I don't think so. Maybe NVIDIA users on Windows generally feel this way, but it is a different thing on Linux. On Linux, a binary driver always removes options, beacuse it imposes restrictions that weren't there in the first place. This is seldom true on Windows, because in the Windows world, Windows itself imposes a lot of restrictions, so an additional layer of restrictions seldom matters. The company that developed the driver only focuses on the least common denominator, and the Linux world is far from homogenous. Different platforms, hardware, XFree:s, kernels, you name it. Also have a look at this post for stuff notoriously lacking from NVIDIA's drivers, and that noone in this world, except for NVIDIA, can add support for in the drivers. Will they? I doubt it. Could somebody else add it? Yes, if it was free.
So it would definitely not be in NVIDIA's best interest to release their drivers. Why would you want them anyway? Would you go through them bug hunting?
This is a common, but really stupid argument when it comes to free software. "Why would you want the source? Are you a developer?". No, I'm not. But other people around the world are. If you believe that the only people in the world that will understand the driver, and would know how to improve it, are currently employees of NVIDIA, you're seriously confused. I want the source because I know that a lot of other people can improve it then. And who knows, maybe I can improve it too someday.
I could care less about my drivers as long as they work well... that's why I buy NVIDIA cards, proven drivers and top of the line hardware.
But for how long? How long do you think NVIDIA will support cards that are not their latest, fanciest generation? And what if you someday need an SMP system because of some reason? What if you need any type of hardware or software that NVIDIA doesn't bother to add support for (or add lousy support)? Good luck.
As far as software engineering not being an engineering discipline, it most definitely is. I am currently learning the trade as an undergrad... However, since argueing with you appears to be as fruitful as attempting to talk to my dog:-)
I think this was not at all fair to the original poster. I think it is you that don't understand that fps today is not everything.
go visit http://www.extremeprogramming.org/ Good stuff about software engineering and building software in general. The linux dev community could really use some material of this nature...it's a shame so many talented programmers don't discipline their work...so much more could be done...
I think you are somewhat confused here. A lot of Linux and GNU developers have university backgrounds... not only are they skilled developers but they know disciplined programming.
My motto is "don't think you know everything about everything, or necessarily know anything better". This helps me from getting surprised or disappointed when the hard truth shows itself. What is to say that the person you replied to wasn't a CS major?
You don't *know* if open drivers will be improved either.
Well, they won't get worse...;-)
Also, see below.
They only get improved if someone has the knowlegde, time, and motivation to work on them.
Very true. But this is the very point where free software shines, and this is a big reason behind that. Drivers being free expands the possibility for potential developers that fulfill this criterion to the world at large, not just a company with a limited (maybe large, but still limited) number of developers. Also, developers on that company may also have explicit orders not to develop drivers for that hardware, because their developing resources are needed elsewhere. The entire world do not have this restriction. Now tell me in which scenario the chances are far greater that hardware support will be improved over time?
Heck, people are still tweaking and improving Linux support on Motorola 68k... what makes you think that a whole world of computer fanatics with some particular hardware will suddenly all be disinterested in that hardware?
It is *not* a given that open software will be better than closed. There really aren't that many people with the low level knowledge of the kernel, X, security, OpenGL, etc. required to work on 3D drivers.
You don't need all of those skills yourself to develop a driver, as long as someone has them. And believe me, the world is huge.
I make my choices on hardware and software based on technical superiority, not moral superiority. I use Linux because I believe it to be technically superior for the kinds of things I do, not because it's GNU. I use NVIDIA hardware and drivers because they are a technically superior solution.
I also started with Linux because I believed in a technical superiority. After I had used it for some time I was convinced of that, and started to wonder why it was superior. Well, very skilled people from all around people developing and contributing to it, without any restrictions other than imposed by themselves, is obviously a big part of that. They can contribute on the parts that exactly matches their skills, and if they want to. Then comes the question of
how they can contribute the way they do. Obvious answer: it's FREE (as in freedom).
After realizing that and this simple chain I was convinced of the mighty powers of free software (and GNU), and fell in love with it. A strange experience if one comes from the Windows world. I believe you will be convinced too someday.:-)
But also worth noticing is that as long as something critical (for example video drivers) are not free, the chain above falls apart. Everything is then up to a single company that can screw you tomorrow by saying "naaaah, we won't do drivers for that anymore, we don't care if that card is less than a year old and still works perfectly fine, we have changed strategies, you're screwed, buy our newest stuff, give us more $$$, and don't you complain". Or "no, we don't do Linux any more, go buy Windows or something, and don't you expect any Linux drivers from us until hell freezes over, because our software has sensitive IP and we will enforce those patents with lawsuits until pigs can fly. Now go away".
You might not think it is a big fuzz with some binary drivers, but one of the reason why GNU and the concept of free software has survived to this day is simple: No compromises. When you compromise, you loose in the end. Compromises is only the mother of more compromises, and in the end you're totally lost in a labyrinth of unfree licenses, commercial closed-source software, and what's still left of free software. Of course, all licenses are then incompatible to add salt to the injury, mess and confusion.
When you realize that, you'll understand that 20 fps more today with one particular driver and card is not worth sacrificing all of this, and why people are fighting to keep the freedom and want free drivers. Because if we loose the freedom, that technical superority that you like will be gone too.
No. With closed drivers you don't know if it will get better.
With open drivers and open specs, you know that support for your precious card will only get better in the future - as time goes, someone will improve it, and they have full access.
With closed drivers, your card will work fine today. But what if NVIDIA tomorrow does an "ID Software" and says that they won't support Linux in the future, due to that niche being to small? Then your card is just a worthless piece of crap... the binary drivers won't be upgraded and you are suddenly stuck with having to use old kernels and old XFree86:s.
Openness is not as much as about today as it is about tomorrow. Personally, I like to decide myself when I want to scrap my hardware.
We had the server open for a brief period after recompiling sendmail and having a misconfiguration, the ISP catched it and informed us, so I think they do a lot to make sure that doesn't happen.
Your ISP "taking care of" open relays doesn't prove that they are opposed to spam. They might still host spammer's web sites, and won't refuse to deal with that type of businesses.
In the later case, they still belong on the RBL, and personally, I have no problem with that. They can't just say that they are fighting against spam and on the same time be hosting the very people they say they're fighting against. To me that would be the definition of hypocrisy.
As far as the subscription model, most everything that they charge for is free on the Debian website, and is actually accessible as well (I've never been able to upgrade my system on the free side of RH).
Everything can be gotten for free in the RH case as well. RHN is so you can just have all your system's details entered (if you want to) and upgrades automatically installed (if you choose to), all using a high-priority server, and on all your systems.
This can of course be done manually as well -- just use your favorite RH mirror (everything is acessible for free on the mirrors) or use the up2date program, that will check your system's configuration and get the relevant updates and install them for you. So RHN is basically just a way to do this automatically and conviniently on a bunch of computers; you are of course free to write a script or something like that to do this for several systems if you don't want to use RHN.
I know of no tiered model of RHN as I described.
There are two tiers - either you get your updates yourself from a mirror, or you pay a little and get the best access to RH:s own server that RH can possibly provide you with, and the possibility to do this automatically with no trouble. I think this is exactly what you described.
RH 7.0 was a nightmare to me. The kernel wouldn't compile, PIs DRM/XFree 4.0 wouldn't compile, netscape seemed to go bonkers (even worse than usual) We tried oracle on it at work, and had to revert to 6.2...the queries would just hang the GUI.
Question: Did you even use the kgcc that is provided explicitly for kernel compiles (and all other compiles you want to use it for?). Also, I've had no problems with the XF4.01 that is provided with RH7, and no problems with Netscape either (at least none that can be blamed on any distribution).
Oracle, however, is compiled for glibc 2.1 and seems to have problem with glibc 2.2. I would wait on an upgrade from Oracle.
I like being able to use newer stuff, but I want it stable. I switched over to Debian, and it just seems very rock solid. XFree 4.0, helix-gnome, accerated openGL and quake3 -- all work great on my Debian boxes. I've never had this consistently on RH.
Red Hat has everything you mentioned. It has been rock solid for me.
Debian has a major learning curve, but I believe it is a better, more stable distro.
If you think a release every other year is good enough. Personally I enjoy getting all the new good stuff tested and packaged in a new release every half a year.
My only point is...if RH is so good at big projects and doing the GUI interfaces...and in many cases this is true...why not leave the distro integration and testing to Debian?
Because a release every other year when the current stable version has gone terribly obsolete is not good enough?
When I read the Debian bug database and compare it to RedHats...some of the people at Debian really, really shine. There are great people at RH, too....obviously. Maybe they should pool their talents...that's all I'm trying to say.
I think they do that quite well. Kernel hackers, glibc hackers, the glibc maintainer, gcc hackers, they are all working for Red Hat. To me, those people really shine.
Final good-to-read stuff would be this article with Michael Tiemann, core GNU person and Red Hat CTO.
I never truely believed that it was possible to make money from Open Source software
Maybe that is why they make money from support. Companies still want support, and they pay for it. A major support contract is big money.
And of course the occasional home user that wants to not have to download all of it, and wants printed manuals. And support, of course.
2) Have a "silver/gold/platinum" tiered subscription model, in addition to the free one, w/ guaranteed response time/login (higher level == better performance/response). Be willing to sell "one time" tickets as well as annual subscriptions.
5) Consider dropping your distro and adopting Debian. I know you are proud of RH, but the realeases appear to have significant flaws...7.0, 6.0, 5.0 were all disasters.
Eh.. pardon me, but why on earth should Red Hat become Debian? Personally, I see no advantages in Debian except maybe apt (which is basically just aconvinient frontend to a package system and can be adapted to rpm).
I also think that RH7 is the best Red Hat I have ever used. I don't see what the "disaster" should be about.
Slashdot Mirror
No, that's not what I said. I said Red Hat 7 has support for kernel 2.4. There is a 2.4-pre kernel on one of the discs, but it's never installed by default. What is used in Red Hat 7 is a 2.2 kernel. In fact, if you read my link, you will notice that it clearly says that Red hat 7 includes kernel 2.2.16. But you didn't read, did you?
Why does that not surprise me? *cough* GCC *cough*
I don't see your point here at all?
Red Hat 7 has support for kernel 2.4.
I agree with you here. I believe this perception is utterly lame and wrong, but I have to agree that it's common. At least in the US, it seems.
About MS, what is there market they have a monopoly in? Tell me that. OS's you say, thats typical. Thats what everyone thinks But tell me, do they have a monopoly?
Yes. Also in the web browser business nowadays.
YOu say they are because they dont open docs, api's or whatnot. That is not a legal requirement.
Well, it could have helped their case if they had done that. But now, this wasn't the situation, and hence they are accused of using monopoly tactics (not opening specs to the competition can be such a tactic, yes).
Yes, they should, but no, they dont have to. Ever.
I think it clearly shows that you don't understand laws against monopoly tactics. Yes, if you are in a monopoly position, you might very well be required to not further crush the competition by not providing for example specs or equal access in other ways.
Besides that, you and I know that there are other OS's out there. Why dont they package ever Compaq with linux? Why not? Because no one wants linux.
I guess the fact that you can indeed buy Compaqs with Linux on them directly from Compaq shows how insanely wrong you are in your arguments.
People think linux sucks, or is hard to use. Right or wrong, thats the perception. Compaq, and its customers, want the features that MS offers - compatibility, software support, ease of use, etc.
The compability Microsoft provides is with other Microsoft products. Other types of compability has never been much of an interest to Microsoft.
Like it or not, thats true. Compaq could package Linux or FreeBSD with its computers, but doesnt
They do. Perhaps you should check your "facts"?
because sales would take a hit.
Gee, so providing more alternatives than just Windows preinstalled would make sales drop? What kind of "logic" is that? If you provide more alternatives, your number of sales either stays the same (in the worst case, because noone buys the new alternatives) or grows (because the new alternatives attract new customers). I think that the fact most famous computer vendors now happily ship computers with Linux proves that this segment is indeed also lucrative.
In fact, everyone who sells computers with Windows does so because it guarantees a certain level of compatibility.
Compatibility with other Microsoft products.
You cite Apple, Netscape, etc as being proof they are monopolies. The fact is, Netscape's product was made useless by MS, by bundling. Bundling is mostly and should be legal.
Gee, in this case maybe it isn't, because that might be considered an illegal monopoly tactic, given the situation. Once again, I think you show that you fail to understand anti-trust laws.
This move helped consumers
Reducing choice by surpressing the competition benefits consumers? How?
Browsers are now standard issue for every major OS. All of them. Before, you had to pay for them. MS helped us all there.
Yes, browsers are now standard issue. That does not mean that they have to be considered part of the OS (office software is standard issue, but also not a part of the OS), nor that we wouldn't have this situation anyway. The reason web browsers are now standard issue and popular apps is because the growth and popularity of the web. The reduction of web browser choice for the consumer on the major client platform is pretty irrelevant to the web explosion, I think.
The fact is, that MS is not, by any reasonable defintion a monopoly in OS's. They cannot control the market place for OS's, they cannot force a standard down everyone's throats (despite best efforts).
I think many people who have witnessed the home computer industry the last 15 years will disagree with you here. Microsoft has on numerous occasions tried to control their platform, and in many cases to reduce competition.
Just because they dont use 'industry standard' protocols or api's doesnt mean that they are breaking the law. Thats their choice.
Yes, it can indeed mean that they are breaking a law. That's what anti-trust laws are about. Some action that would otherwise not be illegal can be illegal if you are in a monopoly situation in your particular market. These regulations are to benefit the consumer by helping further healthy competition not to shrink below the low level that it already is in most monopoly situations.
I think there is some misconception of the term "monopoly" here. You don't have to be the sole company in the market to be a monopoly. Being the by far biggest company is enough. Under certain circumstances this is a crime, but not always. Such circumstances can include using your existing monopoly position in one market to gain a monopoly position in another market. That's what the Microsoft trial is about.
Your second comment makes no sense. What are thousands of Slashdot users against billions of software users world-wide? And don't forget, a majority of Slashdot users use Windows. This fact has proven itself on numerous occasions in polls, and CmdrTaco has mentioned it at least once.
That sounds like a strange way to solve things - I doubt that egcs 1.1.2 will compile a lot of C++ successfully, for example. A better solution might be to just update gcc in RH7 to gcc-2.96-69.
You're right but still wrong. The version included in OSR2 was IE 3.0.
They don't have to think if they don't wan't to.
Both Red Hat and Mandrake provide pointy-clicky-fancy updaters for fetching and installing latest updates. Perfect for those who don't want to think, they just have to remember to run it once in a while.
IE wasn't built into Windows until Windows95 OSR2. Windows95 OSR2 was never sold to the public, but only installed on OEM systems.
So I think your conclusion that Netscape's web browsing revenues increased that year is based on false grounds - because many customers did not have access yet to a Windows version which should make them uninterested in another web browser and the rest of the web browser competition merely by it already including a web browser already.
The consumer benefit is better browser competition and hence more innovation. Simple.
I think this is a very simplistic view. When you say this you basically say that everyone that uses GPL is unaware of the contents of the GPL and basically just a mindless drone that can't think for himself. I would have to disagree with that... In my experience, people that choose the GPL choose it because of the copyleft clause and because they like that model, not because they are unaware of it.
Notably: the whole Qt thing - the wording of the GPL was massaged to condemn a project (KDE) which was created in the spirit of Free Software.
Sadly, "in the spirit of" is often not enough when it comes to licenses. Licenses are not designed to be misinterpreted and sadly you can break a license even when you have noble goals. In my opinion, that was what happened. I believe that was RMS' opinion too. He never said that the KDE project was not intended to be free software or that it was evil in some way - he just pointed out a licensing error that should probably get fixed.
The GNU-approved GNOME thing uses a license (the LGPL) does not encourage Free Software. That makes no sense to me.
Of course it encourages free software. The majority of software in the GNOME project is GPL - it is basically only important libraries that are LGPL. The reason is that people that have a serious problem with the GPL and don't want to use the GPL (for example they want to use the artistic, Apache or BSD license) should still be able to use and link to GNOME code.
The KDE project uses the GPL for everything, which is perfectly fine, also according to RMS. The reason there was a fuzz about KDE was that GPL programs in KDE linked to non-GPL code that was not part of the system (QT), something that is not allowed in the GPL, unless you make an explicit clause for that yourself. That was the only problem. Sadly, people insist on misunderstanding that all the time.
I don't trust Stallman - I think he's a nutcase with his own private adjenda.
He has an agenda, that part is true, but that agenda has been public for 16 years or something like that.
I use the Artistic license because it doesn't have the nasty undertones of the GPL, or the hassles of the LGPL. The LGPL is just lame - it doesn't ensure or encourage Free software, it just forces you to use dynamic libraries, which is a problem when there are so many unstable libraries out there.
Use the stable libraries then... Your argument does not make much sense, previously you claimed that you disliked not being able to reuse code, and here you suddenly claim that you don't like using libraries.
I wish we had face-to-face bandwidth to discuss this - it's just too difficult to explain the problems with the GPL and LGPL in little snippets.
Yes. I really recommend you to go hear RMS speak some time, and ask him questions. That might make things clearer. He is a very good speaker, and I know people who still do not like the concept of the GPL but now at least know the reasons and motives behind it.
Take the NVIDIA driver GPL infringement, for example. About 10 lines of GPL'ed code was used, and for that, everyone was screaming that NVIDIA had to GPL their entire codebase.
No. In a license conflict, there are always two immediate solutions to solve it - either remove the offending code or make the conflict go away in some other way. NVIDIA chose the former solution. And stating that "everybody screamed that NVIDIA should LGPL their entire code base" is a pure lie.
Yeah, right. That is force, just like Microsoft uses.
You clearly have trouble understanding the concept of "force", or at least really mis-use that word. NVIDIA weren't forced to use those 10 lines of code, but in doing that, they had to agree to the terms for using that code. They did not, and hence they had to remove that code. What could be simpler?
The idea of Open Source is code-reuse. The GPL prohibits code reuse for any non-GPL project (proprietary or not). If you use, or link to, any GPL code, you have to use the GPL. That's force, baby, not freedom.
No, it's not force. You're free not to use the GPL code. The terms for the use of the code are clearly stated in the license before you even think about using it.
So, you bought a card with binary-only drivers? ;) 3dfx didn't open their drivers until well after the V3 came out.
I bought this card when I was still a Windows user. 3dfx' announcement later on made it a really good investment though.
They never promised open specs or open drivers.
That is simply not true.
There is no obfuscated code anymore. XFree would not allow the obfuscated 2D code into the distribution.
In case this wasn't clear, we're speaking about 3D drivers here.
Umm, the drivers are free (as in beer).
Geez, when I say free, I mean in the freedom sense. I thought that would be clear.
Just like the DRI drivers, there is some lag in getting the kernel modules working when the latest devel kernel comes out (admitedly it's a bit longer than with DRI)
That is really a funny understatement.
but you can still use the NVIDIA drivers on 2.4pre kernels. Where are the extra sales going to coming from?
The only strong argument against the closed drivers is diverse platform support. BSD, Alpha Linux, PPC Linux, etc. get left out in the cold, which is unfortunate. But, from a business perspective, there's no money there. I ask again, how is it to NVIDIA's advantage to open their drivers?
All other platforms than Linux/x86 are a part of this, yes. But more important is that even many Linux/x86 platforms are unsupported - for example SMP systems. SMP systems are not uncommon today, on the contrary. NVIDIA is simple not an option on these platforms.
No, it's not a silly argument. The difference between what you can release in-house and what goes to the outside world is huge. You need a legal team to go over everything with a fine tooth comb, you need to have professional tech writers to make everything clear - everything has to be perfect - the docs are a *product* and have to be treated as such. If they're in-house, the engineers just walk down the hall and ask someone a question. There are legal liabilities when you release specs.
No, it is still a silly argument. Standard legal disclaimers still apply. It's not like this has not been done before! The docs are not a product that you sell - the docs are specifications, provided as-is, with a standard legal disclaimer with reservations for errors and omissions.
Besides, if the development on NVIDIA is not chaos with people running across everywhere, they already have maintained, in-house docs, with hardware functions and interfaces clearly documented.
Not if you can't understand it. How can you fix function A if you don't know what function A is supposed to do? Without the specs, it's pretty darned hard to write a good driver. Case and point: there are unobfuscated open source 3D drivers for NVIDIA cards in the Utah GLX source tree. They have limited functionality, but the code is there. Without specs, they aren't going to get any better, though.
You're seriously twisting words. We spoke about theoretical fully open drivers here, with full functionality, since the author had full access to hardware info under NDA. This is not at all the case with the existing free NVIDIA drivers - they have limited functionality as you say, and there's noone that can improve their funtionality because noone outside NVIDIA has access to full specs, not even under an NDA.
That is clearly different to my example where some free software developer has access to all specs under NDA and can thus write a fully functional free driver.
My argument boils down to this: if drivers are stable, full featured, and freely available, I don't need source code. The NVIDIA drivers are stable, full featured, and freely available. Simple as that.
No, it is not as simple as that. They are not fully featured, people claim them not to be as stable as you say they are, and they are not free (as in freedom). This means that they do not have the quality they could have had, they will not work on a lot of systems that don't exactly meet the criterias of NVIDIA, either with hardware or software, and future support for these cards are unclear at its best. All of these problems would be solved if the fully functional drivers where open.
Simple as that.
I have nothing against BSD (please no BSD flames), but I guess that the popularities of BSD vs. Linux today does prove you wrong. We have a free software boom today that is caused in large parts by the very GPL we're discussing (again, I'm aware of the wonders BSD-style software have caused too). My point is that we have more free software today because of the GPL, among other licenses (but the GPL is still the most popular).
The GPL preserves freedom. We wouldn't have more free software without it, on the contrary.
Of course it can be used in commercial development. Don't you think, for example, that developers in Linux distribution companies get paid? If you mean that it can't be used in proprietary software, then you're correct. But that is the very intent of the license - it avoids the situation where "improvement feedback" is made impossible due to improvements becoming non-free. That aspect of the license is a feature, not a bug.
If you want to use free software in proprietary software, you can go two ways: Make sure that the free software you use is either BSD-style licensed software, or LGPL.
If it is LGPL you can use it in your proprietary software, as long as you ship the source code for the LGPL part somewhere. But the rest of your application can remain proprietary. Why do you think that many software libraries in the Linux world are LGPL?
So what it boils down to is that different licenses have different uses. GPL is simply not intended to be used in proprietary software, so that's not a bug, and I think it is very strange to think that that would be the case. Also, complaining that you have to "duplicate" the GPL code in your proprietary application is also very strange. The original author can choose whatever license he wants, and you should respect that decision. After all, he has given you the right to use his code to begin with, and since it's his code, he can set any rules he want.
If you think a 3D card is a good investment, I have some nice ocean front property in Wyoming for you ;) I can't think of any other piece of hardware that depreciates in value, or becomes obsolete, faster than a 3D card.
No, please misunderstand me correctly. What I said was not that 3D cards in general are good investments. What I said was that one particular 3D card can be a better investment than others, given certain criteria, like for example that the lifetime of the investment is restricted to the point where the hardware becomes obsolete and not only to the point where a single company does not bother providing drivers anymore.
I still have use for my old Voodoo2, for example. That card was a good investment.
Also, remember that 3dfx supplied binary only drivers for years before they opened up. I'd like to believe that a similar thing might happen with NVIDIA some day.
Yes, I also like to believe that! But insofar there have been nothing but empty promises from NVIDIA. You're right, 3dfx were equally bad until they opened up, but in the very moment they opened up and provided free, good drivers, I knew that the hardware I would by would have life-time driver support. If NVIDIA were to fully open their drivers (read: no obfuscated crippleware) one day, no one would be happier than me to recommend NVIDIA to everyone.
Perhaps if you could make a point where it would be to NVIDIA's advantage to open their drivers? What's in it for them?
Appearantly you didn't read what I wrote in my previous post. I'll quote: Free drivers makes more people able to use the hardware, hence more hardware sales. Very simple.
Publishing open specs means they have to dedicate resources to creating the documentation, proof reading it, maintaining it, etc. This is actually a *lot* more work than it sounds like.
Oh geez, it's not like they don't have any in-house documentation already that they have to proof read and maintain. Providing documentation to outside developers is not fundamentally different than providing docs for in-house developers. That's just a very silly argument.
What I was saying is that you should support vendors who have open drivers , if it's a big deal for you. Be a positive force, not a negative one. Don't bash NVIDIA, praise the companies that do have an open policy.
Oh, I do support vendors with open drivers, and thus I have not, and don't plan to, buy any NVIDIA stuff in the forseeable future. I also advocate and praise the companies that have open policies. But you seem to forget that this whole thread started out because NVIDIA will buy 3dfx and people are worried that NVIDIAs existing closed driver policy will prevail, instead of 3dfx' open one. And you can't discuss the aspects of that without explaining why NVIDIAs way is a bad one.
Actually, I think people should be putting pressure on ATI to release specs for the Radeon. There is an Open driver in the works, but the specs aren't public, they're only available under NDA. This means that you'll have the source code to the drivers, but you won't be able to fix them.
Excuse me? If you have the source, you can improve the driver. Yes, NDAs are bad, but in this case they are only there to protect details of the hardware design, they are not there to obfuscate driver development. The drivers will not be obfuscated. This driver development method is common in the free software world, if the hardware company is worried about sensitive details of their hardware becoming public. A potential free software driver writer can then sign an NDA to get all the gory details that in-house developers get, and write an optimized, non-obfuscated, open driver that can be used in free operating systems.
ATI gets the good press of having open source drivers, but the users don't get the info they need to fix the drivers if they have problems.
I think you are confused. If the driver is open, it can be optimized, problems fixed, and features added. If you want to have access to all the details to improve it even more, you too can ask to sign an NDA if the company requires that.
This is still very differnet from NVIDIAs policy. NVIDIA does not even allow outside developers to sign an NDA to get all the details they need for driver development.
How is it not free?
- You're free to use the code as you wish
- You're free to modify the code as you wish
You're also free to redistribute the code. The only conditions are with the redistribution, and the only "restriction" is that you make the source of your changes availiable. You don't have this restriction if you don't want to redistribute it.In fact, it is not a restriction but a grant of rights. Basic copyright does not allow you to redistribute other people's work, but the GPL does just that, as long as you meet the very simple requirement of keeping your source freely availiable too.
I think you're confusing stuff here. Commercial software can be free software on the same time. In fact, most Linux distros are. :-)
And yes, maybe there will always be non-free software. But what has this to do with that?
And it would be nice if the companies could benefit from the technology in the GNU stuff without having to worry about releasing the source and other licensing crap.
They don't need to worry about it unless they want to redistribute it. They can use all GNU tools in the world inside the company for their production. But if they redistribute it, for example as part of their product, they also have to make their changes availiable. They then have to follow those very rules that the software was given to them to begin with. What's wrong with that?
Besides, how hard is it to set up an ftp or http server? I think there are bigger hurdles than that in software development.
I've always thought of the GPL and pretty much everything RMS as the biggest hypocracy in the universe. "We want to make the code free and support open-source by putting lots of restrictions on what you can do with our code and FORCING you to make your code under the exact same license we use." Hah. What a crock. I despise the "Open-Source" movement as it currently exists.
Heh. Clearly you have no idea what RMS supports and what he does not. RMS supports free software (as in freedom), not open-source. Go listen to RMS someday, then you will know that, and you will also know that he is not a hypocrite. Then go back to this discussion.
The GPL does not force anything. It grants you the right to redistribute other people's work. The only condition is that you have to keep the changes free, as the code was given to you to begin with. I clearly cannot see what's wrong with that.
This may be a new concept to you, but when some people write software on their spare time and let people all over the world use it, they might do it of the reason that it remaining free will help them improve the software in the future. They want the software to be able to cycle in a circle of development, improving over time to the benefit of everybody. They want contributions to remain open so that they can incorporate all improvements done by others back in their original tree. This makes the software evolve, merely by people using other's people work, hack it according to their needs, and putting their changes back online.
If the license does not enforce freedom, this chain is broken. The original author writes a piece of software, a company finds this software and adds lots of interesting features and turns it into a closed-source product. Will the original software ever be able to use those additions? No. Will the freedom that the software initially had, and the closed-source company used, be of any benefit to all others, ever? No.
The GPL is about helping software improve by keeping it free, as the freedom was what made the code happen in the first place.
Imagine the possibilities: I can write software and give it away, but STILL keep it myself!!
Maybe I didn't write it just for me and just for a particular moment in time, but I actually use the stuff and want to improve it, and want the rest of the world to also benefit from all improvements. Now if the rest of the world does not allow me to use their improvements of my software, how can I possibly add all improvements? How can anyone be able to use any improvements at all except their own?
Your argument just doesn't make any sense. If MS were to use some of the Linux kernel in Windows, the Linux kernel would still be free to use how ever you wanted? What would anyone lose by MS using Linux w/o releaseing their code changes?
Believe it or not, but as much as I dislike Microsoft, I do believe they can also do good stuff. Now if they fixed bugs in my code I had spent months trying to fix, or added features, but they didn't release their changes, yes, I would lose. I would lose the possibility to improve my software to everyone's benefit because of some company's refusal to let the improvements benefit anybody but themselves. So in other words, they used the freedom I provided them with to use my stuff and do improvements, but they won't give me any freedom back in return to use those changes, and let everyone benefit.
The GPL exists exactly of this purpose - to help code remain free, to everyone's benefit.
Why frustration over GNU? It was GNU that made everything with Linux at all possible. And what this has to do with profit I simply don't understand; we are talking hardware and drivers here, and the drivers shouldn't be the money-making thing. The hardware should. Free drivers makes more people able to use the hardware, hence more hardware sales.
One argument I keep hearing is support for old hardware. I would argue that in the case of 3D gfx cards, this is not nearly as important as having drivers available when the card is released.
To me the matter of it being a good investment is more important. I can wait with the buy until I know a good (free) driver is out there, but I can't rely on a single company telling me after a year that driver support will be discontinued, making my hardware crap.
The former makes me able to do a good investment, the latter makes me lose money in that I have to buy new hardware, no matter if I want it or not.
NVIDIA is giving me the support that *I* need as a graphics developer. That's not to say that their support will satify everyone - for those people who don't want binary only drivers - don't use them.
Well, it's not so much about binary drivers as it is about lack of open specs that allows a competitive free driver. If NVIDIA only provided binary drivers but also released open specs, it would be as easy as you say "not to use [the binary drivers]".
The NVIDIA drivers are enabling Linux to penetrate new markets and allow people who *don't* have a problem with binary only software as long as it is high quality and stable to have great OpenGL support. A lot of people bash Microsoft for forcing vendors to do things their way. So why is the Linux community trying to force NVIDIA to do things their way?
Those new markets are the markets of the managers that have no clue what Linux is or why it is like it is. They don't understand that the freedom enforcing is what made GNU and Linux happen in the very first place, and hence they don't understand the interest in keeping it that way.
And the analogy of forcing I do not understand - Microsoft is a single company, the Linux community is a community of users. Microsoft forcing vendors is thus indeed forcing, but the Linux communitys criticism of NVIDIAs acts is market pressure. If you think NVIDIA is somehow "forced" by bad, mean Linux users, I feel sorry for you. It is in their best interest to listen to their customers, but they are not forced. If they had been forced, we would have free NVIDIA drivers long since.
Until the open source drivers are at least as good or better than the NVIDIA drivers, I really don't think there is a very convincing argument against NVIDIA. I will always choose the best technical solution. If the open and closed solutions are equivalent technically, I would be inclined to choose the open solution. But moral superiority does not outweigh technical superiority in my opinion.
Your argument is skewed in that NVIDIA does not allow free drivers as good as their binary ones. They do not allow this because they don't release the specs, and this is the very fundamental issue.
Freedom is something which can be enforced, and if it is enforced, it will always be freedom.
Yes, there's a difference between "force" and "enforce". You are never forced to follow the GPL, only if you accept GPL software to begin with. The original developer granted you freedom to begin with under the promise that you will also give that freedom away to others as it was given to you. This is what the GPL enforces, it does not force you to do anything else. To me that is very fair. You can't blame the original author for wanting his work to remain free. If you didn't like that, you shouldn't have accepted it in the first place. You're not forced to use or modify GPL software at all.
No. Every computer user (not just Linux user) will benefit by even better drivers. Why do I say even better? Because then, every person in the world with driver knowledge and incentive can help with improving them further.
I believe most linux users that own NVIDIA are *more* than satisfied, being that they can run 3d applications, and even Quake 3, and comparable speeds to Windows boxes.
I don't think so. Maybe NVIDIA users on Windows generally feel this way, but it is a different thing on Linux. On Linux, a binary driver always removes options, beacuse it imposes restrictions that weren't there in the first place. This is seldom true on Windows, because in the Windows world, Windows itself imposes a lot of restrictions, so an additional layer of restrictions seldom matters.
The company that developed the driver only focuses on the least common denominator, and the Linux world is far from homogenous. Different platforms, hardware, XFree:s, kernels, you name it. Also have a look at this post for stuff notoriously lacking from NVIDIA's drivers, and that noone in this world, except for NVIDIA, can add support for in the drivers. Will they? I doubt it. Could somebody else add it? Yes, if it was free.
So it would definitely not be in NVIDIA's best interest to release their drivers. Why would you want them anyway? Would you go through them bug hunting?
This is a common, but really stupid argument when it comes to free software. "Why would you want the source? Are you a developer?". No, I'm not. But other people around the world are. If you believe that the only people in the world that will understand the driver, and would know how to improve it, are currently employees of NVIDIA, you're seriously confused. I want the source because I know that a lot of other people can improve it then. And who knows, maybe I can improve it too someday.
I could care less about my drivers as long as they work well... that's why I buy NVIDIA cards, proven drivers and top of the line hardware.
But for how long? How long do you think NVIDIA will support cards that are not their latest, fanciest generation? And what if you someday need an SMP system because of some reason? What if you need any type of hardware or software that NVIDIA doesn't bother to add support for (or add lousy support)? Good luck.
As far as software engineering not being an engineering discipline, it most definitely is. I am currently learning the trade as an undergrad... However, since argueing with you appears to be as fruitful as attempting to talk to my dog :-)
I think this was not at all fair to the original poster. I think it is you that don't understand that fps today is not everything.
go visit http://www.extremeprogramming.org/
Good stuff about software engineering and building software in general. The linux dev community could really use some material of this nature...it's a shame so many talented programmers don't discipline their work...so much more could be done...
I think you are somewhat confused here. A lot of Linux and GNU developers have university backgrounds... not only are they skilled developers but they know disciplined programming.
My motto is "don't think you know everything about everything, or necessarily know anything better". This helps me from getting surprised or disappointed when the hard truth shows itself. What is to say that the person you replied to wasn't a CS major?
Well, they won't get worse... ;-)
Also, see below.
They only get improved if someone has the knowlegde, time, and motivation to work on them.
Very true. But this is the very point where free software shines, and this is a big reason behind that. Drivers being free expands the possibility for potential developers that fulfill this criterion to the world at large, not just a company with a limited (maybe large, but still limited) number of developers. Also, developers on that company may also have explicit orders not to develop drivers for that hardware, because their developing resources are needed elsewhere. The entire world do not have this restriction. Now tell me in which scenario the chances are far greater that hardware support will be improved over time?
Heck, people are still tweaking and improving Linux support on Motorola 68k... what makes you think that a whole world of computer fanatics with some particular hardware will suddenly all be disinterested in that hardware?
It is *not* a given that open software will be better than closed. There really aren't that many people with the low level knowledge of the kernel, X, security, OpenGL, etc. required to work on 3D drivers.
You don't need all of those skills yourself to develop a driver, as long as someone has them. And believe me, the world is huge.
I make my choices on hardware and software based on technical superiority, not moral superiority. I use Linux because I believe it to be technically superior for the kinds of things I do, not because it's GNU. I use NVIDIA hardware and drivers because they are a technically superior solution.
I also started with Linux because I believed in a technical superiority. After I had used it for some time I was convinced of that, and started to wonder why it was superior. Well, very skilled people from all around people developing and contributing to it, without any restrictions other than imposed by themselves, is obviously a big part of that. They can contribute on the parts that exactly matches their skills, and if they want to. Then comes the question of how they can contribute the way they do. Obvious answer: it's FREE (as in freedom).
After realizing that and this simple chain I was convinced of the mighty powers of free software (and GNU), and fell in love with it. A strange experience if one comes from the Windows world. I believe you will be convinced too someday. :-)
But also worth noticing is that as long as something critical (for example video drivers) are not free, the chain above falls apart. Everything is then up to a single company that can screw you tomorrow by saying "naaaah, we won't do drivers for that anymore, we don't care if that card is less than a year old and still works perfectly fine, we have changed strategies, you're screwed, buy our newest stuff, give us more $$$, and don't you complain". Or "no, we don't do Linux any more, go buy Windows or something, and don't you expect any Linux drivers from us until hell freezes over, because our software has sensitive IP and we will enforce those patents with lawsuits until pigs can fly. Now go away".
You might not think it is a big fuzz with some binary drivers, but one of the reason why GNU and the concept of free software has survived to this day is simple: No compromises. When you compromise, you loose in the end. Compromises is only the mother of more compromises, and in the end you're totally lost in a labyrinth of unfree licenses, commercial closed-source software, and what's still left of free software. Of course, all licenses are then incompatible to add salt to the injury, mess and confusion.
When you realize that, you'll understand that 20 fps more today with one particular driver and card is not worth sacrificing all of this, and why people are fighting to keep the freedom and want free drivers. Because if we loose the freedom, that technical superority that you like will be gone too.
Yes, that might be the case. My only intent with mentioning Id was pointing to a company that changed their Linux plans. Maybe it was a bad example.
With open drivers and open specs, you know that support for your precious card will only get better in the future - as time goes, someone will improve it, and they have full access.
With closed drivers, your card will work fine today. But what if NVIDIA tomorrow does an "ID Software" and says that they won't support Linux in the future, due to that niche being to small? Then your card is just a worthless piece of crap... the binary drivers won't be upgraded and you are suddenly stuck with having to use old kernels and old XFree86:s.
Openness is not as much as about today as it is about tomorrow. Personally, I like to decide myself when I want to scrap my hardware.
Your ISP "taking care of" open relays doesn't prove that they are opposed to spam. They might still host spammer's web sites, and won't refuse to deal with that type of businesses.
In the later case, they still belong on the RBL, and personally, I have no problem with that. They can't just say that they are fighting against spam and on the same time be hosting the very people they say they're fighting against. To me that would be the definition of hypocrisy.
Everything can be gotten for free in the RH case as well. RHN is so you can just have all your system's details entered (if you want to) and upgrades automatically installed (if you choose to), all using a high-priority server, and on all your systems.
This can of course be done manually as well -- just use your favorite RH mirror (everything is acessible for free on the mirrors) or use the up2date program, that will check your system's configuration and get the relevant updates and install them for you. So RHN is basically just a way to do this automatically and conviniently on a bunch of computers; you are of course free to write a script or something like that to do this for several systems if you don't want to use RHN.
I know of no tiered model of RHN as I described.
There are two tiers - either you get your updates yourself from a mirror, or you pay a little and get the best access to RH:s own server that RH can possibly provide you with, and the possibility to do this automatically with no trouble. I think this is exactly what you described.
RH 7.0 was a nightmare to me. The kernel wouldn't compile, PIs DRM/XFree 4.0 wouldn't compile, netscape seemed to go bonkers (even worse than usual) We tried oracle on it at work, and had to revert to 6.2...the queries would just hang the GUI.
Question: Did you even use the kgcc that is provided explicitly for kernel compiles (and all other compiles you want to use it for?). Also, I've had no problems with the XF4.01 that is provided with RH7, and no problems with Netscape either (at least none that can be blamed on any distribution).
Oracle, however, is compiled for glibc 2.1 and seems to have problem with glibc 2.2. I would wait on an upgrade from Oracle.
I like being able to use newer stuff, but I want it stable. I switched over to Debian, and it just seems very rock solid. XFree 4.0, helix-gnome, accerated openGL and quake3 -- all work great on my Debian boxes. I've never had this consistently on RH.
Red Hat has everything you mentioned. It has been rock solid for me.
Debian has a major learning curve, but I believe it is a better, more stable distro.
If you think a release every other year is good enough. Personally I enjoy getting all the new good stuff tested and packaged in a new release every half a year.
My only point is...if RH is so good at big projects and doing the GUI interfaces...and in many cases this is true...why not leave the distro integration and testing to Debian?
Because a release every other year when the current stable version has gone terribly obsolete is not good enough?
When I read the Debian bug database and compare it to RedHats...some of the people at Debian really, really shine. There are great people at RH, too....obviously. Maybe they should pool their talents...that's all I'm trying to say.
I think they do that quite well. Kernel hackers, glibc hackers, the glibc maintainer, gcc hackers, they are all working for Red Hat. To me, those people really shine.
Final good-to-read stuff would be this article with Michael Tiemann, core GNU person and Red Hat CTO.
Maybe that is why they make money from support. Companies still want support, and they pay for it. A major support contract is big money.
And of course the occasional home user that wants to not have to download all of it, and wants printed manuals. And support, of course.
I believe that is what Red Hat Network tryes to accomplish.
5) Consider dropping your distro and adopting Debian. I know you are proud of RH, but the realeases appear to have significant flaws...7.0, 6.0, 5.0 were all disasters.
Eh.. pardon me, but why on earth should Red Hat become Debian? Personally, I see no advantages in Debian except maybe apt (which is basically just aconvinient frontend to a package system and can be adapted to rpm).
I also think that RH7 is the best Red Hat I have ever used. I don't see what the "disaster" should be about.
And I didn't claim neither of these. PLEASE read the comment you're replying to before posting.