I thought this segment on binary drivers was fascinating, however I think a point in this discussion was missed. It's not a good comparison to compare closed source drivers to closed source applications, because you are free to write a competitor application to a closed source application.
With drivers, it's different. You have paid for your hardware, and with proprietary drivers you lose control. The linux community should encourage open access to information, so the next generation of os will have an easier time than linux did.
So, with that in mind, in my opinion you should avoid hardware unless they do one of three things:
1: Provide documentation
2: not prohibit reverse engineering
3: open the source
If you are nvidia, you are providing a binary driver which you cant examine, they prohibit you from reverse engineering the driver (to write your own driver), and they refuse to document it. It confounds me how this would be acceptable to the linux community.
tllts 138: binary drivers
Moderators: snarkout, Patrick, dann
Well, I wouldn't run a blob on a server, but I'm personally damn glad for the nvidia driver. I'm not really "the linux community" though - I'm just a user, and a user who doesn't think closed source is inherently evil, at that.
That being said, as someone who uses primarily linux on his machines, I couldn't agree more with the boycot sentiments - I simply do not buy linux unfriendly hardware. It always pisses me off when I'm combing the ubuntu forums for something to see all of the user *DEMAND* for support for devices that are known to be anti-linux.
That being said, as someone who uses primarily linux on his machines, I couldn't agree more with the boycot sentiments - I simply do not buy linux unfriendly hardware. It always pisses me off when I'm combing the ubuntu forums for something to see all of the user *DEMAND* for support for devices that are known to be anti-linux.
Shared pain is lessened, shared joy is increased; thus do we refute entropy.
--Spider Robinson
--Spider Robinson
-
Anonymous
I guess being friendly enough to linux just isn't enough to me; I want vendors to open documentation a lot more than I want a binary driver. I don't want another windows, where linux gets drivers, but they are all proprietary, so if I want to use another os, i'm out of luck.
This, however, is not the same as saying drivers must be opensource. Quite honestly, open source drivers are not the best option; hardware documentation is the best option.
nvidia not helping free software is one thing, but it's quite another when they actively try to stop people from writing nvidia drivers, that's what im opposed to. The driver itself? I could care less if it's open source or closed source (their code, their rules). My hardware though? I have a right to know how it works, I think.
I guess I would compare it to a case if intel wouldn't tell you how to write assembly language, but would sell you a closed source compiler. They then further added license terms that made it illegal to produce your own compiler. I wouldn't find this acceptable, and it's the same sort of situation.
This, however, is not the same as saying drivers must be opensource. Quite honestly, open source drivers are not the best option; hardware documentation is the best option.
nvidia not helping free software is one thing, but it's quite another when they actively try to stop people from writing nvidia drivers, that's what im opposed to. The driver itself? I could care less if it's open source or closed source (their code, their rules). My hardware though? I have a right to know how it works, I think.
I guess I would compare it to a case if intel wouldn't tell you how to write assembly language, but would sell you a closed source compiler. They then further added license terms that made it illegal to produce your own compiler. I wouldn't find this acceptable, and it's the same sort of situation.
Snarkout wrote:Well, I wouldn't run a blob on a server, but I'm personally damn glad for the nvidia driver. I'm not really "the linux community" though - I'm just a user, and a user who doesn't think closed source is inherently evil, at that.
That being said, as someone who uses primarily linux on his machines, I couldn't agree more with the boycot sentiments - I simply do not buy linux unfriendly hardware. It always pisses me off when I'm combing the ubuntu forums for something to see all of the user *DEMAND* for support for devices that are known to be anti-linux.
In many cases, reverse engineering is against the law - the DMCA has especially made this true in the tech sector. It sucks ass, but is a fact of life. There is simply no way nvidia and/or ati are going to open source their hardware configs - these configs are the *only* thing they have of value that their competitors do not. These are companies that are in the business of making money - they don't give a rat's ass about linux. Do you know of any video card producer that has handed their hardware specs over to linux hackers?
Shared pain is lessened, shared joy is increased; thus do we refute entropy.
--Spider Robinson
--Spider Robinson
I think a point that wasn't stressed in this debate is that binary drivers aren't included in the kernel source tree because it's illegal. The kernel is GPL and can't be linked with non-free drivers without violating the license.
No one is saying you can't have or use proprietary drivers, they just can't legally be included in the KERNEL SOURCE TREE. The kernel source tree is not a Linux distro.
No one is saying you can't have or use proprietary drivers, they just can't legally be included in the KERNEL SOURCE TREE. The kernel source tree is not a Linux distro.
-
Anonymous
The DMCA's case against RE is exagerrated. The DMCA simply does not apply to drivers in basically 99.999999% of the cases, as there is no circumvention involved. Additionally, even if there was circumvention involved, if there is no other way to get the information, dmca permits circumvention for purposes of interoperability (which would probably not apply in the nvidia case, admittedly -- but in the nvidia case, there would be no circumvention needed).
You don't need access to their hardware configurations to write a driver, so that is irrelevant. Intel doesn't provide access to the internals of how their chips work, but they certainly tell you how to write software which works with their chips. That's all people are asking for.
Admittedly, things arent quite so bad as I make it out to be, particularly on the reverse engineering front. One can just reverse engineer the drivers and ignore the eula in many countries (certainly in the uk and in australia reverse engineering is permitted for interoperability, and the eula can not take that right away) but the fact that nvidia (and to a lesser extent, ati) try to PROHIBIT their customers from using the hardware they paid for is quite offensive.
In the end, support of this sort of behavior is going to doom linux. If you support this sort of behavior in nvidia, then you also must support the same sort of behavior in vendors that dont bless linux with proprietary drivers, and consequently bless their right to prohibit linux kernel hackers from reverse engineering the windows driver to figure out how it works.
Reverse engineering (in all its forms) is a perfectly legitimate way of finding out how something works. It's very rarely illegal, dispite the hype about the DMCA (which doesnt apply in most cases anyways). Without reverse engineering, we wouldnt have samba nor nearly as many drivers as we have now, so to support a vendor who prohibits it is quite insane.
That said, reverse engineering is the approach of last resort. Nobody wants to reverse engineer (likely windows) drivers. Most people just want documentation. Vendors that dont provide documentation or source and try to legally prevent you from using the hardware you paid for should not be supported imo.
You don't need access to their hardware configurations to write a driver, so that is irrelevant. Intel doesn't provide access to the internals of how their chips work, but they certainly tell you how to write software which works with their chips. That's all people are asking for.
Admittedly, things arent quite so bad as I make it out to be, particularly on the reverse engineering front. One can just reverse engineer the drivers and ignore the eula in many countries (certainly in the uk and in australia reverse engineering is permitted for interoperability, and the eula can not take that right away) but the fact that nvidia (and to a lesser extent, ati) try to PROHIBIT their customers from using the hardware they paid for is quite offensive.
In the end, support of this sort of behavior is going to doom linux. If you support this sort of behavior in nvidia, then you also must support the same sort of behavior in vendors that dont bless linux with proprietary drivers, and consequently bless their right to prohibit linux kernel hackers from reverse engineering the windows driver to figure out how it works.
Reverse engineering (in all its forms) is a perfectly legitimate way of finding out how something works. It's very rarely illegal, dispite the hype about the DMCA (which doesnt apply in most cases anyways). Without reverse engineering, we wouldnt have samba nor nearly as many drivers as we have now, so to support a vendor who prohibits it is quite insane.
That said, reverse engineering is the approach of last resort. Nobody wants to reverse engineer (likely windows) drivers. Most people just want documentation. Vendors that dont provide documentation or source and try to legally prevent you from using the hardware you paid for should not be supported imo.
Snarkout wrote:In many cases, reverse engineering is against the law - the DMCA has especially made this true in the tech sector. It sucks ass, but is a fact of life. There is simply no way nvidia and/or ati are going to open source their hardware configs - these configs are the *only* thing they have of value that their competitors do not. These are companies that are in the business of making money - they don't give a rat's ass about linux. Do you know of any video card producer that has handed their hardware specs over to linux hackers?
Wow. It's obvious that you and I look at this issue from very different angles. IMO, if anything were to doom linux, it would be the fact that it has support for very few devices compared to windows (no, I do not mean "out of the box"), and that there are a million zealots waiting to pounce on the user who just wants a way out of the windows world. I completely accept the fact that I can't expect to take place in linux discussions without politicos and fanatics making themselves heard - that's part of the GPL way - but it is a major, major, MAJOR turnoff for not-political, non-religious new users. Me, personally, I'm about 50/50 - I believe in the free software model, but I don't pretend to let it affect the way I view the world.In the end, support of this sort of behavior is going to doom linux.
Shared pain is lessened, shared joy is increased; thus do we refute entropy.
--Spider Robinson
--Spider Robinson
-
Tsuroerusu
- Posts: 2551
- Joined: Mon Sep 05, 2005 8:51 am
- Location: Silkeborg, Denmark
- Contact:
The DMCA basically says: "It is not permitted to disable and/or reverse engineer copy protection systems"tiocsti wrote:The DMCA's case against RE is exagerrated.
Even if they might be rootkits, thanks Sony.
"Hatred does not cease by hatred, but only by love. This is the eternal rule."
- Siddhattha Gotama (Buddha), founder of Buddhism.
-
Anonymous
Unless you believe linux will get to somewhere in the 20-40% market share range, it is never going to be the case that vendors are going to provide linux with all the proprietary drivers it needs. Drivers require expertise to build, and are not cheap from a development standpoint.
With that in mind, the only way to solve the driver issue is to refuse to use proprietary drivers to push vendors into providing what linux really needs, which is documentation.
If users accept proprietary drivers (and by proprietary drivers, I mean proprietary drivers combined with no documentation on the hardware and no right to reverse engineer), then they will forever be relegated to irrelevance.
Unlike rms, I don't think open source drivers are all that important, but certainly documentation on hardware is vital. The openbsd people seem to be the only ones fighting this fight (and their results benefit everyone).
With that in mind, the only way to solve the driver issue is to refuse to use proprietary drivers to push vendors into providing what linux really needs, which is documentation.
If users accept proprietary drivers (and by proprietary drivers, I mean proprietary drivers combined with no documentation on the hardware and no right to reverse engineer), then they will forever be relegated to irrelevance.
Unlike rms, I don't think open source drivers are all that important, but certainly documentation on hardware is vital. The openbsd people seem to be the only ones fighting this fight (and their results benefit everyone).
Snarkout wrote:Wow. It's obvious that you and I look at this issue from very different angles. IMO, if anything were to doom linux, it would be the fact that it has support for very few devices compared to windows (no, I do not mean "out of the box"), and that there are a million zealots waiting to pounce on the user who just wants a way out of the windows world. I completely accept the fact that I can't expect to take place in linux discussions without politicos and fanatics making themselves heard - that's part of the GPL way - but it is a major, major, MAJOR turnoff for not-political, non-religious new users. Me, personally, I'm about 50/50 - I believe in the free software model, but I don't pretend to let it affect the way I view the world.In the end, support of this sort of behavior is going to doom linux.
-
Anonymous
That's not quite true, as the dmca does have three relevant exceptions:
1: Circumvention is possible for interoperability 1201(f)
2: Circumvention is possible to identify encryption flaws 1201(g)
3: Circumvention is possible to identify security vulns 1201(j)
There's various gotchas, of course. 1201(j) is basically worthless (it requires authorization), but for reverse engineering something like a sony rootkit, 1201(g) is applicable, but it does require a good-faith effort to aquire authorization, but if one can not aquire it, you can circumvent lawfully (whether it's a dmca violation or lawful circumvention, of course, is only determinable in court).
That said, the dmca basically does not apply to most situations (particularly in the driver space), as there is simply no circumvention involved. In the driver space, eula restrictions prohibiting reverse engineering are the primary factor (iow, contract law).
1: Circumvention is possible for interoperability 1201(f)
2: Circumvention is possible to identify encryption flaws 1201(g)
3: Circumvention is possible to identify security vulns 1201(j)
There's various gotchas, of course. 1201(j) is basically worthless (it requires authorization), but for reverse engineering something like a sony rootkit, 1201(g) is applicable, but it does require a good-faith effort to aquire authorization, but if one can not aquire it, you can circumvent lawfully (whether it's a dmca violation or lawful circumvention, of course, is only determinable in court).
That said, the dmca basically does not apply to most situations (particularly in the driver space), as there is simply no circumvention involved. In the driver space, eula restrictions prohibiting reverse engineering are the primary factor (iow, contract law).
Tsuroerusu wrote:The DMCA basically says: "It is not permitted to disable and/or reverse engineer copy protection systems"tiocsti wrote:The DMCA's case against RE is exagerrated.
Even if they might be rootkits, thanks Sony.
-
CptnObvious999
- Posts: 798
- Joined: Fri Jun 03, 2005 7:54 pm
- Location: Maryland
- Contact:
Ok I am not against proprietary drivers or software if it fits your business model, what I am against is using the law to restrict making an open source option.
Also right now we are in no position to ask for companies to not only make drivers for linux but provide documentation and have the drivers open source. That has always been the job of the hackers and I am glad that nVidia has decided to support Linux at all. That being said I would like open drivers but I can see nVidia's business model almost requires it. We don't have to worry about them charging for it or giving bad or no support to them so I am perfectly happy with them. They have done an excellent job with them and have even been active working with some open source projects like XGL/AIGLX.
Also right now we are in no position to ask for companies to not only make drivers for linux but provide documentation and have the drivers open source. That has always been the job of the hackers and I am glad that nVidia has decided to support Linux at all. That being said I would like open drivers but I can see nVidia's business model almost requires it. We don't have to worry about them charging for it or giving bad or no support to them so I am perfectly happy with them. They have done an excellent job with them and have even been active working with some open source projects like XGL/AIGLX.
-
Anonymous
If they provide documentation, they don't need to provide drivers. If they provide both, the driver does not need to be open source. Having the information needed to write a driver is a far far far far FAR preferable choice than having a driver (opensource or not).
I would agree that it would be unreasonable if we were to demand that they provide drivers, make them opensource, and provide good hardware detail so other drivers can be written, but nobody is asking for all of that.
Nvidia's business model does not require closed drivers. They are selling hardware, how to control that hardware does not expose anything they would not want their competitors to know (if it does, they need to redesign -- since drivers are not very difficult to reverse engineer).
The hardware vendors in taiwan and the like realize it is in their interest to provide information, it is mostly american vendors (ati, nvidia, adaptec, etc) that are opposed to allowing their customers use the products they paid for on their choice of os (which might not even be linux, what if I want to use my nvidia card on, say, netbsd or even something more obscure like vsta?).
kerneltrap recently had an interview with Theo DeRaadt (of openbsd), where some of these issues were discussed.
http://kerneltrap.org/node/6550
I would agree that it would be unreasonable if we were to demand that they provide drivers, make them opensource, and provide good hardware detail so other drivers can be written, but nobody is asking for all of that.
Nvidia's business model does not require closed drivers. They are selling hardware, how to control that hardware does not expose anything they would not want their competitors to know (if it does, they need to redesign -- since drivers are not very difficult to reverse engineer).
The hardware vendors in taiwan and the like realize it is in their interest to provide information, it is mostly american vendors (ati, nvidia, adaptec, etc) that are opposed to allowing their customers use the products they paid for on their choice of os (which might not even be linux, what if I want to use my nvidia card on, say, netbsd or even something more obscure like vsta?).
kerneltrap recently had an interview with Theo DeRaadt (of openbsd), where some of these issues were discussed.
http://kerneltrap.org/node/6550
CptnObvious999 wrote:Ok I am not against proprietary drivers or software if it fits your business model, what I am against is using the law to restrict making an open source option.
Also right now we are in no position to ask for companies to not only make drivers for linux but provide documentation and have the drivers open source. That has always been the job of the hackers and I am glad that nVidia has decided to support Linux at all. That being said I would like open drivers but I can see nVidia's business model almost requires it. We don't have to worry about them charging for it or giving bad or no support to them so I am perfectly happy with them. They have done an excellent job with them and have even been active working with some open source projects like XGL/AIGLX.
In the case of nVidia, I believe their reason for not open sourcing the drivers is that there is code in the driver which is not their IP.
The thing that I think is a real shame is, when I bought my Amiga it came with a manual which contained the entire circuit schematic for the machine. The same when I bought a BBC micro before that.
The question I have is, for companies that have open-sourced their drivers, has this had a determental effect on their business?
The thing that I think is a real shame is, when I bought my Amiga it came with a manual which contained the entire circuit schematic for the machine. The same when I bought a BBC micro before that.
The question I have is, for companies that have open-sourced their drivers, has this had a determental effect on their business?
Thats why linux users should boycott nvidia and other 'offensive' companies. Don't give into the urge of playing fancy games or having fancy eyecandy, and don't buy their hardware. In fact, today its quite possible to run a desktop completely devoid of closed bits of software.Anonymous wrote:the eula can not take that right away) but the fact that nvidia (and to a lesser extent, ati) try to PROHIBIT their customers from using the hardware they paid for is quite offensive.
In the meanwhile the community can wait for an underdog company to see the light and release a decent, open card with excellent 2D acceleration and reasonable 3d acceleration.
That company is called Intel. They already have decent chipsets with free drivers.thetza wrote:In the meanwhile the community can wait for an underdog company to see the light and release a decent, open card with excellent 2D acceleration and reasonable 3d acceleration.
Sorry, I'm not buying a new graphics card just so I can say I'm not running a proprietary driver. If this card ever dies, I'll probably buy something with a free 3D driver.