Kernel developers' position on GPLv3

Posted Sep 23, 2006 15:31 UTC (Sat) by BrucePerens (guest, #2510)
In reply to: Kernel developers' position on GPLv3 by Jel
Parent article: Kernel developers' position on GPLv3

Well, it's too bad, but ultimately I think the kernel developers are shooting themselves in the foot. First, on the patents issue. You can make a plausable argument that running Linux on anything is illegal in a large part of the world, including where most kernel developers reside, due to the embedded patent infringements. There are enough software patents granted that you could say the same for essentially any software, but the kernel developers have more to lose and I believe they overestimate the force that would be brought to bear in their defense. A few suits and settlements might leave them just as encumbered as any Microsoft software. But they're going to ignore that because it's political and rely on OSDL's ineffective half-measures to improve patent "quality", which ultimately just makes the patents that will be used against them better.

And then the DRM thing. You really want your contributions to be locked down? It seems to be entirely against the spirit of Open Source and I doubt that in the face of widespread locked-down use of Linux that you could sustain contributions outside of the businesses that do the locking down.

Ultimately, we need to recognize that Linux is a 15-year-old kernel and that there will be another technical development to superscede it eventually. I can't say what that will be, but I think the best chance of mobilizing individual contribution to it would be to use GPL 3.

Bruce

to post comments

Kernel developers' position on GPLv3

Posted Sep 23, 2006 20:27 UTC (Sat) by steelpillow (guest, #40703) [Link] (11 responses)

Bruce Perens wrote, "Ultimately, we need to recognize that Linux is a 15-year-old kernel and that there will be another technical development to superscede it eventually. I can't say what that will be, but I think the best chance of mobilizing individual contribution to it would be to use GPL 3."

That made me check my crystal ball. I saw a massive fork of just about everything into -2 and -3 licensed streams. Most developers, especially the paid staffers, needed to see their stuff happen on Linux, and followed the -2 fork. The -3 forks one by one joined the HURD on the back burner. Then a new OS came along, and in order to leverage all that Linux-friendly stuff it was released under GPL v2.4.

Things got a bit blurry after that, where my tears had fallen. I wiped them away and caught someone saying, "the place to fight unwelcome licensing laws is the legislative process, not the licenses themselves." I lost contact.

Kernel developers' position on GPLv3

Posted Sep 24, 2006 17:41 UTC (Sun) by BrucePerens (guest, #2510) [Link] (10 responses)

Well, RMS has the best crystal ball of anyone I know. He sat down in the early eighties and envisioned a lot of what would go wrong today. So, I take him very seriously. I think his failure this time is mainly a PR failure. He's not involved the kernel developers in solving the problem.

Bruce

Kernel developers' position on GPLv3

Posted Sep 25, 2006 3:45 UTC (Mon) by jstAusr (guest, #27224) [Link] (1 responses)

> I think his failure this time is mainly a PR failure.
> He's not involved the kernel developers in solving the problem.

I really don't understand what else he could do. The kernel developers aren't interested in any changes. What do you think he could have done differently? It is rather difficult to involve those that don't want to be involved.

Kernel developers' position on GPLv3

Posted Sep 30, 2006 12:44 UTC (Sat) by Blaisorblade (guest, #25465) [Link]

> I really don't understand what else he could do. The kernel developers
> aren't interested in any changes.
> What do you think he could have done differently? It is rather
> difficult to involve those that don't want to be involved.

For Linus Torvalds, this is true - he absolutely said "not" to GPLv3.

However, a set of kernel developers, in their whitepaper about this
(http://lwn.net/Articles/200422/) are pointing out significant issues in
GPLv3, and he should consider those issues and try to solve them.

For the DRM thing, that clause would stop a Linux kernel signed by any
distro. When you run a binary kernel from a distro, requiring a signature
from the kernel builder (think to Debian) would effectively stop
insertion of non-authorized modules (think to Adore or such rootkits);
but GPLv3 would disallow this (I think). Also, code for this has been
written IIRC by Fedora, so this is not "theoretical".

Developers also think the distinction between such cases and DRM abuses
is impossible to draw in a license, because it is connected to political
reasons. RMS should prove them wrong on this point.

Kernel developers' position on GPLv3

Posted Sep 25, 2006 5:01 UTC (Mon) by zorgan (guest, #4016) [Link] (2 responses)

Well, RMS has the best crystal ball of anyone I know. He sat down in the early eighties and envisioned a lot of what would go wrong today.

True, its fantastic how right he was, and especially his creation of the GPL was amazingly innovative and successful. But that doesn't mean that he is always right, and in fact IMNSHO he has been wrong often enough. (Think: cathedral...)

Kernel developers' position on GPLv3

Posted Sep 25, 2006 13:01 UTC (Mon) by pinky0x51 (guest, #40742) [Link] (1 responses)

>he has been wrong often enough. (Think: cathedral...)

Can you explain this? What exactly do you mean?
"cathedral" reminds me only on "The Cathedral and the Bazaar" but that has nothing to do with RMS, that's Eric Raymond's thing.

Kernel developers' position on GPLv3

Posted Sep 28, 2006 17:25 UTC (Thu) by smoogen (subscriber, #97) [Link]

One of the arguments behind Cathederal and Bazaar that has been lost was how development around FSF itmes seemed very insular around Boston. The Emacs/Xemacs and the gcc/ecgs forks were due to a very Cathederal take that the FSF had over its own software. The HURD/Linux "fork" was also an example. HURD was built as a cathederal and had many incarnations that booted but didn't cover everything that the people wanted back then to show. Linux booted but barely had a user environment and it went out the door.

Talking about Linux to RMS was a "forbidden" topic where one would get a nice letter chastising people over using something that was taking away effort from FSF goals. He never answered my question when I could get a bootable HURD to test this stuff with. After a while the stern lecture about hurting FSF got replaced by saying that Linux had to be called GNU/Linux.

So when ESR originally wrote C&B he was talking about how the Linux development model was better than the closed door FSF model. FSF has changed a lot since then.. so you get better access to their snapshots and stuff

Kernel developers' position on GPLv3

Posted Sep 25, 2006 5:41 UTC (Mon) by charris (guest, #13263) [Link] (2 responses)

I have the impression that RMS has actually antagonized the kernel developers, among others, over the years. Just because the kernel uses the GPL doesn't mean the developers owe RMS eternal gratitude and must hang on his every word. Indeed, perhaps RMS should express some gratitude to Linus. The kernel is probably the highest profile project out there using the GPL, and without the kernel there would be no GPL operating system to run the toolset on.

Kernel developers' position on GPLv3

Posted Sep 26, 2006 4:20 UTC (Tue) by drag (guest, #31333) [Link] (1 responses)

There are FreeBSD and NetBSD kernels, which can be forked to GPL. Debian has GNU systems using a BSD kernel. Also Hurd was also functional as a kernel until they decided go L4 on it.

Not that I'd actually want to run one of those kernels.. Linux is by far the most sophisticated and best performing kernel aviable. It's just the best.

RMS is fine. You can't seem to trust him.. but as long as your goals are the same as his then everything is cherry.

What is going on here is just different points of view.

The Kernel developers like the GPL because it compels third parties to contribute code back into the Linux kernel. They don't seem care about the freedom so much or helping to ensure the freedom of end users of products using their software.

They are dedicated to making the kernel as usefull and most effective kernel it can be.

The political nature of RMS's message of his brand of 'Freedom' is counteractive to that goal, going with the current GPLv3 draft would sacrifice some commercial interests of Linux which would benifit Linux developers with code, testing, support, etc etc.

Personally I am a end user and obviously for my own self interest I would like the kernel to go GPLv3 because it would help me avoid devices that claim they 'run linux', but are not hackable. It would help to ensure that companies won't try to sneak restrictions in on me. Anything to make DRM less attractive for hardware developers is a good thing to me.

However I doubt that that realy is the highest priority for the kernel devs. I don't blame them.

Kernel developers' position on GPLv3

Posted Sep 26, 2006 17:49 UTC (Tue) by viro (subscriber, #7872) [Link]

Right. I just can see the crowds of BSD developers jumping on the
chance to work on GPLv3 forks of *BSD kernels and leaving the
original branches stranded and obsolete. Well, actually I can't,
but I guess it's just another example of kernel folks not having
sufficient proficiency with crystal balls and other fraud paraphrenalia...

Kernel developers' position on GPLv3

Posted Sep 25, 2006 20:56 UTC (Mon) by steelpillow (guest, #40703) [Link] (1 responses)

It seems to me that RMS has not succeeded in involving the kernel developers even in believing that there is a problem to solve - at least not a licensing problem - never mind in how to solve it.

In the end it is content (present or future) which sells the OS, and if GPL3 blocks out the content providers then GPL3-ed stuff will just not get used.

There are content-rich restrictive OS out there. The content providers can afford to just ignore us. To bring them in to line will, I suspect, need more than just a rewritten license. There are many growing movements for restriction-free content: musicians posting free downloads, ogg vorbis, and so on. The black hats are evolving a coordinated strategy against all this. We need to do the same. Maybe GPL3 will have its day - I sure hope so. But is now really the right moment? I think the gut instinct of the kernel developers is "not yet."

Someone commented that GPL3 is in essence political. Maybe. At any rate, one key ingredient of successful politics is persuasion. Another is timing.

Guy

Kernel developers' position on GPLv3

Posted Sep 29, 2006 18:51 UTC (Fri) by bronson (guest, #4806) [Link]

To bring them in to line will, I suspect, need more than just a rewritten license...

We're talking about freedom, aren't we? Is there a need to to bring anybody into line? If you want to impose your own morals on other people then, I agree, it will take more than a license to do it!

Kernel developers' position on GPLv3

Posted Sep 24, 2006 16:21 UTC (Sun) by jejb (subscriber, #6654) [Link] (4 responses)

To respond to the patents statement: it was definitely the part that garnered the least agreement, so as the document goes, it's probably least representative of the global kernel community view. That said, it wasn't phrased to make any judgements about software patents. What it said was that the current patent clauses in GPLv3 could be seen by corporations as a deterrent to distribution of GPLv3 software. It's this deterrent effect that is called out as the bad thing. If it comes to a patent war over Linux, we won't entirely be relying on OSDL ... we'll also be relying on OIN and the patent portfolios of several other large Linux contributors.

On DRM ... and really, I'm getting a little tired of these muddled issues here, so let me try to separate them.

1. The Tivo case: What Tivo did (making the boot loader only accept signed kernels) was at the behest of its content providers. If GPLv3 had been in force when this came along, the stark choice would have been go out of business now; or go out of business while trying to replace the entire OS.

Firstly, the proposed changes would have affected and probably killed Tivo, but done nothing whatever to impact the people who forced the change: the content providers. It's like treating a symptom, not curing the disease

Secondly, to consider the idea that companies in this position don't give anything back lets use the example of a Tivo competitor here: Moxi (also subject to the same lock down rules by the same content providers). It's produced by a company called Digeo who, by my possibly inaccurate count provided us with several driver enhancements, some nice filesystem work and a kernel maintainer, all under the old GPLv2. Please explain the inequity here.

Finally, embedded Linux in systems as firmware is an end use by the manufacturer. It's also getting embedded further and further away from the user. The GPLv3 introduces a new requirement that wherever these uses of Linux are, you have to be able to find them and modify them ... this represents dictating to the manufacturers how they build systems. It doesn't just affect deliberate things like keyed boot loaders; it also affects people who, because they didn't think about it, didn't provide such a modification channel. This is why it alters the equation: hardware manufacturers must feed this in now as a design requirement. That's what I don't think should be in the licence.

2. All of this is a proxy battle for the real war which is the content providers who're trying to dictate the rules in the first place---and they're doing it purely because they're too stupid to find a new business model, but have the cash to try to buy one. The technological challenges in delivering a copy protected stream all the way from the media to the screen are daunting; and expensive, which is why the manufacturers keep telling the content providers to bog off purely on business grounds. The content providers won't take no for an answer and are now trying to force this by leveraging both their assets (the content) and their money (by lobbying for legal measures). Attacking the manufacturers, who are natural allies in the fight against this type of DRM on business grounds, isn't going to budge the content providers one whit, and might end up diluting the opposition to them by fragmenting it.

Linux in fifteen years? Well ... as long as we maintain the innovation stream, I don't see why not, which is why GPLv3 gets considered on the grounds of its potential impact to that innovation stream.

James

Kernel developers' position on GPLv3

Posted Sep 24, 2006 17:45 UTC (Sun) by BrucePerens (guest, #2510) [Link] (2 responses)

I think Richard is trying to use the weapons we have to solve the problems we have. I think he understands that this problem is a much more difficult one to solve with just political discourse. The other site had a lot of money power, while our power is in our software and we refuse to use it. It doesn't sound like we're going to be effective on the DRM problem this way.

Bruce

Kernel developers' position on GPLv3

Posted Sep 24, 2006 19:44 UTC (Sun) by jejb (subscriber, #6654) [Link] (1 responses)

OK, and I'm saying I don't necessarily have a problem with the ends, just the means.

Just because you don't think you can reach the true culprit (the content providers) doesn't mean you should settle for shooting the messenger (or at least the people caught in the middle).

Firstly, this is wrong on purely moral grounds.

Secondly, the collateral damage from doing this is too great. Even if I accept that Tivo has made no useful contributions (which I don't) you'll end up shooting Digeo and a whole host of embedded device vendors too; entities who, I believe, have made (and continue to make) valuable contributions.

Finally, it antagonises a whole crowd of people who're unhappy about being forced to go along with the content providers and who might have more pressure to bring to bear in the cash and power arena than we do. I really see this as a vast strategic mistake.

I fully respect yours and Richard's right as private individuals to use the weapons you have available. However, I take issue when you expect to use my gun. I'm also not ready to throw in the towel on coming up with a strategy to bring this home to the content providers.

James

Kernel developers' position on GPLv3

Posted Sep 26, 2006 18:36 UTC (Tue) by ajaypal (guest, #40762) [Link]

Did i see a flash back, history repeating itself, when the original GPL was opposed and considered a crazy idea.

Only time will tell.

Kernel developers' position on GPLv3

Posted Sep 28, 2006 5:13 UTC (Thu) by AJWM (guest, #15888) [Link]

Just as an aside, a quick grep of the source turns up more TiVo than Digeo copyrights in the kernel. Admittedly that's a pretty crappy measure of relative worth, though.

One rather off-topic question -- if it's the TiVo bootloader that vetos loading an unsigned kernel, what stops someone from changing the bootloader? Maybe it needs a hardware hack (I have no idea), but that surely hasn't stopped the community before. (Xbox, as I recall, required a hardware mod to load a different OS.)

Kernel developers' position on GPLv3

Posted Sep 28, 2006 3:22 UTC (Thu) by etrusco (guest, #4227) [Link]

"I can't say what that will be, but I think the best chance of mobilizing individual contribution to it would be to use GPL 3."

LOL. Before reading the full post I thought Bruce Perens meant "use GPL v3 in Linux" and it will be the best chance of a competitor :-P

Kernel developers' position on GPLv3

Posted Sep 28, 2006 9:23 UTC (Thu) by pkern (subscriber, #32883) [Link]

Ultimately, we need to recognize that Linux is a 15-year-old kernel and that there will be another technical development to superscede it eventually.

Wait, 15 years old... only 3 years until it becomes mature. In my legislation, though. SCNR.

Kernel developers' position on GPLv3

Posted Sep 28, 2006 18:31 UTC (Thu) by smoogen (subscriber, #97) [Link] (1 responses)

I just don't understand how changing everything to the GPLv3 will fix it when humans will come up with ways to circumvent that pretty quick to... and they will continue to do so as long as their is a strong motivation to do so that fulfills their needs. The way that you stop it is by convincing enough people that it is in their best interest to play nicely with each other and share their toys when asked.

How to accomplish this we have differing views... but I do not currently see where the middle ground that people will meet on. And without a middle ground the majority of people who aren't developers, aren't coders in any form, don't want to futz with their computer anymore than they want to change the oil on their cars, are going to congregate. And until we get them convinced, it doesnt matter if we write the worlds most perfect License that makes sure that the Four Freedoms are never put aside.. it would just be declared invalid by the paid for courts etc etc.

My view is that work on getting the public knowing why Libre software is important, what it means, and what they get out of it. AND why it makes the Four Needs (Reproduction, Sleep, Food, Safety) easier to accomplish than anything else. Then you can get the problems that stop the Four Freedoms dealt with by consensus in a political/social manner.

Kernel developers' position on GPLv3

Posted Oct 5, 2006 19:35 UTC (Thu) by linuxrocks123 (subscriber, #34648) [Link]

Offtopic but you asked for it...

> AND why it makes the Four Needs (Reproduction, Sleep, Food, Safety)
> easier to accomplish than anything else.

You can live without reproducing. Many people do. Therefore, it is not a
need. In any case, bringing in made-up crap like this during a discussion
about software licenses is asinine.