The Torvalds Transcript (InformationWeek)

[Posted March 21, 2007 by corbet]

InformationWeek interviews Linus Torvalds about GPLv3. "Finally, the real basic issue is that I think the Free Software Foundation simply doesn't have goals that I can personally sign up to. For example, the FSF considers proprietary software to be something evil and immoral. Me, I just don't care about proprietary software. It's not "evil" or "immoral," it just doesn't matter. I think that Open Source can do better, and I'm willing to put my money where my mouth is by working on Open Source, but it's not a crusade -- it's just a superior way of working together and generating code."

to post comments

The Torvalds Transcript (InformationWeek)

Posted Mar 21, 2007 14:56 UTC (Wed) by zooko (guest, #2589) [Link] (41 responses)

There's a middle path here. You don't need to sign up on a proclamation that proprietary software is immoral, but you can still believe that, all other things being equal, Free Software is better for the society that depends upon it than proprietary software is. There's no need to go to an extreme stance in order to value Free Software as such -- for its freedom in addition to its usefulness.

Likewise, the other extreme is the belief that any constraint imposed by a software licence is immoral and that all software should be useful enough to survive under a permissive (BSD-style) licence. This is an unnecessarily extreme position -- imposing GPL-style constraints can help produce more and better software, and this is one of the reasons why Linux is so successful.

If you read the entire interview with Linus Torvalds it would seem that he too follows a middle path, but if you read just the out-of-context sentence quoted in this LWN article it would seem that he is being extreme.

Regards,

Zooko

The Torvalds Transcript (InformationWeek)

Posted Mar 21, 2007 15:36 UTC (Wed) by TRauMa (guest, #16483) [Link] (40 responses)

It's interesting that that's the first time I read something pragmatic from Linus about the GPLv3. The points he has about license splintering are valid, IMHO, but then again what you get by combining multiple GPL offerings with different added restrictions is the GPLv3 with all restrictions applied. So when releasing something under the GPLv3 you should perhaps be comfortable with all the possible restrictions it contains, as code flowing back to you may have them. Seen this way there is no splintering at all, or at least not more than with GPLv2, where you can add any permission you like, too, but code coming back may not have those permissions anymore.

The other point he has about market forces sorting out the closed vs. free development models may be true, but I don't really believe him when he says he doesn't care about it politically. If tomorrow closed source would just crush free software and he would be forced to live with an IT infrastructure that wouldn't cater to his needs, but those of the vendor (or to live the digital equivalent of a hermits life), he'd mind it _very_ much. I see ignoring DRM issues as one way to get there.

The analogy between CS vs. FS and Alchemy vs. Chemistry is a nice one, it alone made it worth reading the article. Of course, with closed source, the alchemy actually works to some degree, and can be harmful to the user.

The Torvalds Transcript (InformationWeek)

Posted Mar 21, 2007 16:44 UTC (Wed) by anonymous1 (guest, #41963) [Link]

Closed Source in todays world is not just alchemy, it alchemy along with the coercion of copyright.
Public domain binaries without the corresponding source code is analogous to alchemy. But copyrighted binaries are not a good analogy to alchemy.

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 9:35 UTC (Thu) by malor (guest, #2973) [Link] (38 responses)

I remain convinced that the bad blood between Stallman and Linus over 'GNU/Linux' is at the core of Linus' objections. I don't think he would be attacking the GPLv3 so fiercely if he weren't personally invested somehow.

First, you see him talk about 'motives'; that's a code phrase for 'what Stallman is trying to do'. Most of what he's saying here isn't even arguing the technical merits of the license, he's saying, obliquely, that it's unwise to trust Richard Stallman. (and considering the GNU/Linux imbroglio, he's not entirely wrong to do so.) Most of his explicative text covers variations of this theme.

When examining his specific objections, it's obvious they come from the standpoint of the single most successful Open Source project; at this point, the kernel devs get more personal fame and fortune from broader Linux use, and to get that, they seem willing to trade away future freedoms. It's okay, they argue, for manufacturers to use their code to give us devices we can't change or use in ways of which the manufacturers don't approve.

It appears the devs don't want this restriction added because it will interfere with their own success; they are thinking of their own pocketbooks and fame. Closed-device manufacturers, after all, often hire people to work on the kernel code, increasing the 'jobs for Linux kernel hackers' market. The devs are not, unfortunately, holding end-user rights paramount, one of which is, and always has been, the right to use devices freely that are dependent on Free Software.

If you can't remove and replace the code on a device, but the manufacturer can, particularly without your permission, you don't truly own it. The code has been hijacked. You are enslaved by that code and dependent on it. (see: Microsoft Vista.) Stallman is trying to prevent this from ever happening with anything under the GPL, and he is right to do so.

Linus et al are clearly in the wrong on this issue. They appear willing to trade away our freedoms for their own success, and I'm saddened by this.

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 12:43 UTC (Thu) by jospoortvliet (guest, #33164) [Link] (4 responses)

I'm totally with you on this. I think the freedom to RUN the software is as important as the 'you have to share the source' part of it. The world could fully consist of Open Source Software (GPLv2) running on paladium/TCP hardware, so nobody could do with it what they want, except the big hardware and software vendors. I'm pretty sure Linus wouldn't like that, but apparently he's not worried, doesn't care, or doesn't think it's possible.

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 15:45 UTC (Thu) by zlynx (guest, #2285) [Link] (3 responses)

I don't think that people are really worried that *all* computing devices will be locked down with TCPA or whatever. It always sounds to me that they're worried cheap and easy computers will be locked down.

Does anyone really believe FPGAs, 8 bit micro-controllers, embedded ARM and MIPS chips will *all* be running trusted computing?

Heck, someone with time on their hands could breadboard a 6502 or 4004 chip. It'd be big, ugly, expensive and slow, but it's a computer.

It would never be "nobody could do with it what they want". It'd be "nobody could do with it what they want for less than $5,000".

I guess I'm in the "not worried" and "don't think its possible" camps on this one.

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 18:29 UTC (Thu) by mmarq (guest, #2332) [Link]

"" Does anyone really believe FPGAs, 8 bit micro-controllers, embedded ARM and MIPS chips will *all* be running trusted computing? ""

Absolutely. Its a good way to prevent tampering with some OEM product. And those microcontrollers will have the TCPA in the VLSI design, because at 32nm node they would be too small to be only 8 bit FPGAs.

TPA everywhere?

Posted Mar 25, 2007 2:49 UTC (Sun) by kevinbsmith (guest, #4778) [Link] (1 responses)

With the proper political climate, it could become illegal (in some/many countries) to sell a non-TPA computer. Or to build one. Or even to own one that was built pre-TPA. Hopefully that won't happen, but I can't see any reason right now that it absolutely positively could not happen in the next ten years.

If we have to live in that world, I would still want a thriving FLOSS ecosystem.

TPA everywhere?

Posted Mar 25, 2007 4:44 UTC (Sun) by dlang (guest, #313) [Link]

sure, and it could become illegal to own a car, illegal to drink, illegal to smoke, illegal to be unemployeed, illegal to have a job if you are a Jew, etc

stupid things happen. if they happen in one country, that country will suffer and fix itself or go down the drain.

and the probability of non TPA computers being illegal is infintesimal (just like everything else I listed above, there are people who want it)

if you think about the cost of implementing a full TPA environment, and consider adding that to everything that's run by computer (down to your toaster, coffie maker, vending machines, cars, etc) the cost of forcing all new products to implemtn TPA would scuttle it, and if you look at how it's basicly been impossible for the HD-TV to replace all the TV receivers around the world in anything resembling a reasonable timeframe, what makes you thing that there would be any more sucess in replacing all these other devices?

easily programmable computers are too much a part of the worlds economy for any country to sucessfully outlaw them, even if you ignore people's home PC's

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 13:31 UTC (Thu) by paulpach (guest, #20903) [Link] (17 responses)

> If you can't remove and replace the code on a device, but the
manufacturer can, particularly without your permission, you don't truly
own it.

What about when you dont own it?. A company I worked for LEASED appliances
that run linux and gave the source with it in full compliance with GPLv2.
To ensure the safety of our customers, we used DRM to
restrict what the appliance could run. This was a SELLING POINT because
the customers did not have to trust anyone else but us. We where clearly
distributing software and that would violate GPLv3 if it looks anything
like the drafts.

BTW, if you drink the kool-aid, GPLv4 should restrict some other uses that
restrict user freedom:
* Code shall never be used in ROM because you can never run your
modifications in it's intended device (looks a lot like DRM doesn't it?)
* Code shall never be used to run jails
* Code shall never be used in weapons ( if you are dead, arguably you are
no longer free to run the software)
* Code shall never be used in security system because people are no longer
free to enter the building
etc...

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 18:08 UTC (Thu) by malor (guest, #2973) [Link]

I'm not sure how it would apply to a lease, but with a normal sale, you'd be required to give them a master key; they don't have to exercise their right to modify the device running Free Software, but you have to make sure they can do so. You can't reserve rights to yourself that your downstream clients don't have.

No, you as a business can't do everything you want in that situation. That's your cost for using Free Software. If you want to give your customers devices where you have more rights than they do, find code that's under a more permissive license. It's really that simple.

As far GPLv4 not being usable for jails and wars and crap... that's obviously a strawman. This modification of the GPL is merely to make sure that downstream users get all the same rights as everyone else... that no matter where you are in the chain of code transfer, you can do everything that every other person in the chain can do.

That is the spirit of the GPLv2; version 3 simply extends it to provide additional freedom, so that everyone is empowered. You can't hold people hostage with hardware if it's running GPL3 code. Period. If you want to, use code that's licensed more liberally.

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 18:40 UTC (Thu) by mmarq (guest, #2332) [Link] (15 responses)

"" We where clearly distributing software and that would violate GPLv3 if it looks anything like the drafts. ""

Well... not necessarily

Private contracts are private, and if a costumer trust a vendor with a DRM implementation, that is ok... what the vendor shouldnt be able to do with GPL, is to push a DRM scheme down the trough of every possible user...

The Torvalds Transcript (InformationWeek)

Posted Mar 23, 2007 4:27 UTC (Fri) by drag (guest, #31333) [Link]

Well with this sort of DRM there is absolutely no need to give anybody the keys besides the people who own the machine.

That is why things like 'can't use GPLv3 with voting machines' is such a red herring.

(assuming the draft makes it through)
Yes you can use GPLv3 in voting machines and YES you can use DRM to protect those voting machines from tampering.

Same way you can sell machines to people and use DRM to protect those machines from being tampered with.

The deal is is if your customer wants the ability to change the software in there you have to give them, nobody else, a way to modify it. If they don't want to modify it and if they don't want to disable the DRM then they don't have to.

It's the end user that gets put into control of the hardware and the software not the manufacturer.

So with voting machines, since they are owned by the government, nobody but the government needs to be allowed access to it. And they can regulate it how they please since the machines belong to them. Which is exactly how it would be any other way GPLv3 or not.

So it's a pretty spacious arguement that your going to use DRM for end user's own good. If they want trusted computing or other such things to help make their systems more secure then there is absolutely nothing in the GPLv3 that prevents them from taking advantage of TPM or whatever.

Remember that the goal is here is to allow the end user control over the software.

Not the hardware.
Not non-GPL code.
Not DRM'd media.
Or anything like that.

All that is perfectly legal and if you invent a way for DRM to work while having end users being able to understand how the programs work and be able to modify the code, then it would be perfectly fine to use GPLv3 licensed code.

It's not designed to prevent that or limit that stuff in any way.

GPL and private contracts

Posted Mar 23, 2007 22:48 UTC (Fri) by giraffedata (guest, #1954) [Link] (13 responses)

Private contracts are private, and if a customer trusts a vendor with a DRM implementation, that is ok...

No, it's a violation of the license conditions under which the vendor acquired the code. The original author's copyright is violated.

If you have an enforceable promise from your licensee not to avail himself of the "license," then you haven't actually given a license. A condition of GPL is that you give an actual GPL license when you redistribute the code.

That's the weird and rather unique thing about GPL: it interferes with private transactions that don't even involve the copyright owner. It's meant to be a social engineering device, not a commerce tool.

GPL and private contracts

Posted Mar 23, 2007 22:52 UTC (Fri) by malor (guest, #2973) [Link] (12 responses)

... "it interferes with private transactions that don't even involve the copyright owner." ...

If you're using my code, it involves me. If you want to give my code to someone by loading into a device, that's fine, but the price for doing so is that they get the same rights that you have. You can't lock them out of their hardware, because the hardware is a necessary component to running the software that you licensed from me.

If you don't like that price, you can pay for other code with money.

GPL and private contracts

Posted Mar 23, 2007 23:49 UTC (Fri) by giraffedata (guest, #1954) [Link] (11 responses)

My statement is (for the sake of simplicity) over-broad. One could write pages on what it means for a transaction to "involve" someone else. John Donne said (The Tolling Bell) even the death of a stranger damages you, because you're both part of Mankind.

But the fact remains that the level of involvement you have in a transaction between two people using your code somewhere down the chain is fundamentally different from the kind of involvement most private legal acts (licenses, contracts, deeds, etc.) care about.

How many copyright licenses other than GPL seek to give value to some total stranger down the chain who will give the copyright owner nothing in return for it except the warm feeling of knowing Right has prevailed?

Missing this special quality of GPL is what leads people like the poster above to believe that GPL couldn't possibly stop a recipient from voluntarily giving up his rights (in return for something of value from his distributor).

GPL and private contracts

Posted Mar 24, 2007 1:12 UTC (Sat) by malor (guest, #2973) [Link] (10 responses)

The purpose of the GPL is to keep the code free and open for everyone. No matter where you are in the chain of transfer, you have the same rights and responsibilities that everyone else does. GPLv3 just shuts down some methods of removing rights from downstream recipients.

Anytime you're transferring a copyrighted work, the original creator is 'involved'. You have the right to transfer your only copy of something. You do NOT have the right to copy it and keep a copy for yourself.

With ordinary copyrighted code, you have to pay money for copies; Microsoft is just as 'involved' as a GPL author in that case. Or, you can opt to pay with freedom; you give up some of yours, and you increase the freedom of the transferee. But it's the same fundamental transaction; you are buying the right to use the code.

You do NOT have the right to give someone a copy of Windows without paying for it. By your assertion, Microsoft is 'involving itself' in a 'private transaction' between you and a customer. You can't legally steal Microsoft's code. Why should GPL code be any different?

GPL and private contracts

Posted Mar 24, 2007 2:51 UTC (Sat) by giraffedata (guest, #1954) [Link] (9 responses)

I wouldn't say Microsoft is involved in the transaction per se, but rather that it is involved in the copying that is part of it. And in contrast to the GPL publisher, Microsoft wouldn't care about the details of the transaction -- who paid whom for what. Unlike a sale of Windows from Microsoft to a Microsoft customer, where Microsoft is involved in a whole different way and is deeply interested in the terms of the deal.

I really don't know what battle you're fighting; it sounds like you're defending the use of GPL, but nobody has attacked it.

We agree the copyright owner is always involved. We agree the purpose of offering code under GPL isn't to selfishly protect the author's wealth, but rather to bring about a social goal. We agree that it's perfectly legitimate for a person to offer, and accept, code with GPL restrictions.

The only thing I'm saying, which I don't see how you can disagree with, is that GPL is special (I also used the term "fairly unique") in its effect on downstream copiers.

Maybe I could say, "unlike conventional copyright licenses, GPL interferes in transactions in ways in which the copyright owner doesn't have a material interest."

GPL and private contracts

Posted Mar 24, 2007 13:59 UTC (Sat) by vmole (guest, #111) [Link] (2 responses)

The GPL is special in its effect on downstream copiers because it's very rare for licenses to *allow* downstream copiers.

GPL and private contracts

Posted Mar 24, 2007 17:29 UTC (Sat) by giraffedata (guest, #1954) [Link] (1 responses)

The GPL is special in its effect on downstream copiers because it's very rare for licenses to *allow* downstream copiers.

The license doesn't allow downstream copiers, but the copyright owner does, by issuing additional licenses to them. Same for GPL and non-GPL.

But look at what the "stream" is: The stream is the modification of the software. A writes code; B adds stuff to it and passes it on to C, C does the same to D. With non-GPL, A insists on royalties for the C-D copy, but otherwise A usually doesn't care what the deal between C and D is.

GPL and private contracts

Posted Mar 24, 2007 18:09 UTC (Sat) by malor (guest, #2973) [Link]

> With non-GPL, A insists on royalties for the C-D copy, but otherwise A usually doesn't care what the deal between C and D is.

With commercial code, I've never seen that kind of chain relationship, where B had full rights to become a competitor to A. If it's a value-add chain, where each step adds more stuff, A's not giving up any rights. They insist on being paid at each step of the way, and remain involved and active throughout the chain. If I buy a custom database written on Oracle, Oracle gets paid (a lot!) for the database instance I'm running, and that's above and beyond whatever I pay for the custom database code.

With GPL, A insists on being paid by making sure that all downstream users have the same rights to all code that was added... including, likely, him or herself. The payment terms are "freedom and code" instead of money. B through D voluntarily give up rights in either GPL or non-GPL scenarios. With non-GPL code, they cannot freely make copies and must, generally, pay per running instance. With GPL code, they don't have to pay, but have to give away any improvements they make along with their improved binaries. In general, B through D give up fewer rights under the GPL than they do with regular copyright, but they always give some up.

Commercial A, as recompense, gets money. GPL A probably gets access to downstream code improvements. That's not a GPL requirement, but in practice, that's usually what happens.

Ultimately, there's little difference. Both As are still getting paid. As B through D, you're buying either codebase, you're just paying differently.

As GPL B through D, you're getting a great deal; you can set yourself up as a full competitor to anyone else in the chain. You're not hostage to anyone's code, and in exchange, you can't hold anyone else hostage. It's a pretty sweet deal, overall. It's why the GPL prospers.

TANSTAAFL. If you want good code, you usually have to pay for it. Fortunately, the GPL is not particularly onerous.... unless you intend to enslave your customers instead of serving them.

GPL and private contracts

Posted Mar 24, 2007 16:10 UTC (Sat) by malor (guest, #2973) [Link] (5 responses)

> "Maybe I could say, "unlike conventional copyright licenses, GPL interferes in transactions in ways in which the copyright owner doesn't have a material interest."

No, you can't say that either. As the other poster points out, the copyright owner most emphatically DOES have a material interest in all reproductions of the copyrighted work.

You get the unique ability to reproduce *and modify* the work freely, in exchange for ensuring that everyone downstream from you has the exact same rights that you do.

Again: yes, the GPL changes the nature of the relationship you can have with a customer. If you don't like that, you can pay for code with money, or you can use BSD licensed programs instead. You don't have to accept the GPL if the unique privileges and powerful utilities it offers aren't worth the price. v3 makes the price a little steeper, but I think only irrational people can argue that v2 wouldn't have said exactly the same thing if DRM had existed twenty years ago.

GPL and private contracts

Posted Mar 24, 2007 18:16 UTC (Sat) by giraffedata (guest, #1954) [Link] (4 responses)

Maybe I could say, "unlike conventional copyright licenses, GPL interferes in transactions in ways in which the copyright owner doesn't have a material interest.
No, you can't say that either. As the other poster points out, the copyright owner most emphatically DOES have a material interest in all reproductions of the copyrighted work.

Irrelevant; the quote above doesn't say anything about material interest in the reproduction. It talks about material interest in ways of interfering with a transaction.

You don't have a material interest in whether some stranger gets the source code for some code written by someone else. Through the magic of GPL, you may control it anyway.

And don't try to define "material interest" so that all interests are material. There's a separate term for a reason.

GPL and private contracts

Posted Mar 25, 2007 0:38 UTC (Sun) by malor (guest, #2973) [Link] (3 responses)

> You don't have a material interest in whether some stranger gets the source code for some code written by someone else. Through the magic of GPL, you may control it anyway.

If you are commercial, you have an interest in making sure you get paid for all copies of your code. If you are GPL, you have an interest in making sure anyone that adds to your program honors the license.

Both types of code have a price. One is money, but allows the freedom to keep your code closed. The other is usually low- or no-cost, but denies you freedom to hide your code.

Pay one or the other, your choice.

(or go BSD code, if you prefer.)

GPL and private contracts

Posted Mar 25, 2007 0:53 UTC (Sun) by vonbrand (subscriber, #4458) [Link] (2 responses)

Both types of code have a price. One is money, but allows the freedom to keep your code closed. The other is usually low- or no-cost, but denies you freedom to hide your code.

This is wrong. You can certainly keep your modifications to GPL code secret. If you distribute the modified code, you have to share the source to the changes.

Note that in the case of closed source you don't have the right to distribute anything, not the original version nor your modified one.

GPL and private contracts

Posted Mar 25, 2007 0:57 UTC (Sun) by malor (guest, #2973) [Link] (1 responses)

Yes, this is true, but this particular thread is talking about a chain of people, from A to B to C to D. giraffe seems to be asserting that the GPL is somehow unethical or something in that it 'interferes with' other business relationships, and I'm trying to point that this is the PRICE of using the code, and that it's entirely optional. If you don't like the price, buy commercial code instead.

You give up rights in either case, it's just a matter of which rights fit your particular situation the best.

GPL and private contracts

Posted Mar 25, 2007 1:15 UTC (Sun) by giraffedata (guest, #1954) [Link]

giraffe seems to be asserting that the GPL is somehow unethical or something in that it 'interferes with'

It's "or something," and in particular I'm asserting that GPL is unusual among copyright licensing schemes. Nothing more. I jumped into this thread when someone made an incorrect statement about what one can do with GPL code, based on reasoning that would be sound for any conventional licensing scheme. I went beyond correcting the particular misstatement to explain you have to use a whole different mindset when looking at what's allowed with GPL code.

The misstatement, IIRC, was that a downstream distributor could make a private deal with his distribuee in which the latter doesn't avail himself of GPL freedoms, and the copyright owner wouldn't have anything to say about it.

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 23:06 UTC (Thu) by barbara (guest, #3014) [Link] (14 responses)

>I remain convinced that the bad blood between Stallman and Linus
over 'GNU/Linux' is at the core of Linus' objections. I don't think he
would be attacking the GPLv3 so fiercely if he weren't personally invested
somehow.

You're probably right. I also think the reason that Linus can't stand
Richard Stallman is that he reminds Linus of his parents who were involved
in left-wing politics (his Dad was a communist and spent one year studying
in Moscow -- horrors!!). Politics was a big part of Linus' family, he's
rebelling against that in my view, and he now sees *any* kind of politics
in Linux as a Bad Thing (tm).

With all the threats to our freedom to use our computers as we wish, to
use an unfettered Internet, and to simply have the freedom Linux brings
us, it's important to keep our eye on and fight these threats (be it the
DMCA, the RIAA, the MPAA, and a whole host of other control freaks).
Politics is indeed important despite what Linus says and thinks.

The Torvalds Transcript (InformationWeek)

Posted Mar 22, 2007 23:48 UTC (Thu) by malor (guest, #2973) [Link] (7 responses)

Well, I can't speak about that directly, as I don't know him personally; I saw him do a Q&A session once, but that's about as close as I've gotten. :) I am quite sure that there's something more to his GPLv3 hatred than he claims, but what it is, exactly, I don't know.

Stallman hatred seems plausible to me, but your idea of it being rebellion against his home politics could also be an explanation. Regardless, his apparent disdain for supporting the idea of Free Software has even bitten him, very badly, in the BitKeeper fiasco. Apparently, he didn't learn that it's important never to make yourself hostage to code.

In my view, the GPLv3 is the best way to make sure that doesn't happen, and I find it curious and disturbing that he'd fight so hard for the ability to be chained up again.

GPL3 = BitKeeper2?

The Torvalds Transcript (InformationWeek)

Posted Mar 23, 2007 7:03 UTC (Fri) by dlang (guest, #313) [Link] (6 responses)

even stallman used closed source tools when there weren't opensource tools that would do the job.

at the time the bitkeeper use started there wasn't anything available that would do the job nearly as well (for that matter, when the bitkeeper use stopped there wasn't a good option available, that's why he started the git project)

stallman could have done all his programming on a system with no closed source tools. it would have inconvienced him, but it could have been done (not all computers HAVE a BIOS on them), he choose not to do this and instead use existing closed source components until he considered the opensource componenets 'good enough'

Linus does the same thing, it's just that what each of the two is willing to define as 'good enough' is different.

beyond that there is a fundamental difference in approach in winning users

Stallman wants you to switch to opensource (free) software becouse it's 'The Right Thing To Do'

Linus wants you to switch to opensource software becouse it's the best choice available to you.

Personally, I think that Linus' approach is going to be far more sucessful. you always do better when you involve people's self interest then when you tell people that they should suffer for the cause.

David Lang

The Torvalds Transcript (InformationWeek)

Posted Mar 23, 2007 12:12 UTC (Fri) by malor (guest, #2973) [Link] (5 responses)

He was widely and strongly criticized for choosing BitKeeper, and as it turns out, the critics were correct for doing so. Code that can hold you hostage is bad.

Closed source, in and of itself, isn't automatically bad: what's important is whether it's used to assert control over how you use it and what you do with it. BitKeeper did that, and it bit Linus very badly.

The anti-DRM provisions of the GPL3 are there to prevent people from being BitKeepered via hardware restriction. It amazes me that Linus, of all people, would come down against that.

The Torvalds Transcript (InformationWeek)

Posted Mar 25, 2007 0:17 UTC (Sun) by vonbrand (subscriber, #4458) [Link] (4 responses)

He was widely and strongly criticized for choosing BitKeeper, and as it turns out, the critics were correct for doing so. Code that can hold you hostage is bad.

Yes, Linus was criticized for using BitKeeper. No, he was absolutely right in doing so: The then imminent "Linus meltdown" was avoided, the kernel development picked up speed, and in the end all have got better SCM tools.

Was it painless? No. Was it worth is? Yes!

The Torvalds Transcript (InformationWeek)

Posted Mar 25, 2007 0:41 UTC (Sun) by malor (guest, #2973) [Link] (3 responses)

It was unnecessary pain. If he had simply refused to be chained in the first place, the whole fiasco wouldn't have happened. Kernel development essentially came to a dead stop for ages while they redid their tools; had they simply done the tools in the first place, they'd never have had the problem.

Code that can hold you hostage is bad. It's always bad. Anytime someone wants to tell you how or when you can use code, you're at a profound disadvantage.

I can't believe you're painting BitKeeper as a net positive. You sound like a Republican spinning the Plame outing.

The Torvalds Transcript (InformationWeek)

Posted Mar 25, 2007 2:34 UTC (Sun) by bronson (guest, #4806) [Link] (1 responses)

First, it really wasn't as painful as you describe. Kernel development definitely did not come to a standstill. While Linus ran off and wrote Git for a bit, everyone else continued using whatever tool they liked. No big deal (unless you happen to think that only Linus can develop the kernel...?)

Because existing distributed SCMs were crap (they were), Linus had to write his own tool no matter what. It isn't like one path involved great pain and the other path involved zero pain. What scenario are you picturing where pain is somehow avoidable and there's no need to write Git?

Now, would you rather:

- Linus write his tool cold, having very little first-hand experience of distributed development.

- Linus write his tool after thoroughly using the most advanced distributed development tool available at that time.

Seems to me like either way works pretty well.

For the record, I agree that the Republican spin on Plame (and now Gonzales) are just stupefying. But they have pretty much zero in common with Linus choosing an SCM.

BK and git: Lessons learned (or not)

Posted Mar 25, 2007 3:05 UTC (Sun) by kevinbsmith (guest, #4778) [Link]

Linus (and the world) got lucky with git, and many folks have learned the wrong lesson from the whole affair. I don't mean that Linus isn't a great programmer or that git isn't a great tool. What I mean is that the BK folks *chose* to give the kernel folks several weeks to stop using BK, rather than terminating all kernel dev licenses immediately (which they could have done). And git turned out to be relatively quick and painless for Linus to write.

If either of those (or other factors) had come out differently, the kernel could have been hurt very badly by the whole BK experience. Was BK net positive? With hindsight, I would say yes. Could it have been a disasterous net strong negative? Absolutely. The risk was high (and was known at the time BK was selected), the bet was made, and Linus got lucky and won.

Wrong lesson: Depending on closed source for a key tool is fine, because git easily replaced BK.

Right lesson: Depending on closed source for a key tool is very risky, and you really don't know how things might turn out.

The Torvalds Transcript (InformationWeek)

Posted Mar 25, 2007 4:33 UTC (Sun) by dlang (guest, #313) [Link]

if he had refused to use bitkeeper at all then kernel development would have ground to a halt as Linus overloaded, and people would not have had the experiance with distributed SCM tools to be able to create git or it's equivalent.

if the opensource SCM systems weren't up to the job of being able to handle the kernel after several years of bitkeeper being an example, what makes you think that any of them would have been suitable for it back then?

as far as I can tell, other then git, there is still no opensource SCM that can handle the scale and distribted workflow of the kernel development.

yes bitkeeper was a gamble, but the options at the time were gambles as well (what decision isn't a gamble for that matter). Yes, if Larry McVoy had been an evil person out to scramble the kernel development process things could hve turned out worse then they did

remember that even stallman says that you are justified in useing a closed source tool if there are no opensource options.

The Torvalds Transcript (InformationWeek)

Posted Mar 23, 2007 21:07 UTC (Fri) by h2 (guest, #27965) [Link] (1 responses)

barbara and malor, thanks. It's been, especially since the Groklaw rantings by Linus, very easy to see that the foundation of his opposition was clearly not rational, even though he's now of course trying to paint some rational cover for this. But what was a mystery to me was the actual source of that irrational opposition. Now it finally makes sense, even though I'm usually not a big fan of psychological explanations, this one just fits a bit too well. I think there are some other core reasons too, equally irrational, but there's not a lot of point getting into details.

What continues to amaze me is that Linus and the majority of the kernel team day in and day out enjoy the freedoms a license like the GPL explicitly envisioned when it was written, but they seem somehow unable to admit that their actual day to existence is the direct result of the freedom the GPL made a space for. That's the core freedoms, not the fiction of Open Source.

They live good, free, creative lives, almost dreamlike in fact. But they don't seem able to connect the dots. Odd. The fact that they have come to mistake some corporation for the actual real person end user, that's kind of sad, but predictable too, that's the world they live in.

The GPL always restricted the freedoms it imparts enough to make it able to sustain itself against a world that is generally quite hostile to that concept. And BSD type licenses have always been there for users who don't want that restriction. And the world in the past 15 years has grown more aggressive, you can without any problem see patent attacks on the horizon, so failing to take something like that into account would be a serious error on the part of the FSF. And it's not an error they have any plans on making. The entire Tivo thing collapses as soon as you remember that the GPL is all about the rights of the end users, not some corporations. People that is. That also has not changed, nor I hope will it ever.

The Torvalds Transcript (InformationWeek)

Posted Mar 28, 2007 19:10 UTC (Wed) by wilck (guest, #29844) [Link]

Linus understands the GPLv2 perfectly. His statements in the Newsweek article demonstrate that clearly. He acknowledges that the GPLv2 fits his needs _although_ the intentions of the FSF when they published the GPLv2 may be different from his own.

Whatever Linus' true motivations may be, it's pathetic to dismiss his statements because of psychological speculations. You can disagree with him (I do, too, at least partially), but please respect the man and his arguments.

The Torvalds Transcript (InformationWeek)

Posted Mar 25, 2007 23:55 UTC (Sun) by jschrod (subscriber, #1646) [Link]

> I also think the reason that Linus can't stand
Richard Stallman is that he reminds Linus of his parents who were involved
in left-wing politics (his Dad was a communist and spent one year studying
in Moscow -- horrors!!)

You are either a troll, or have quite obviously never been to Finland, or to Europe for that matter. Postings like yours and those of marlor who accuses Linus Torvalds and the kernel developers to "behave out of their pocketbooks" make me want to ask Jon for a KILL file feature in the LWN comment section.

Joachim

The Torvalds Transcript (InformationWeek)

Posted Mar 28, 2007 18:57 UTC (Wed) by wilck (guest, #29844) [Link] (2 responses)

I also think the reason that Linus can't stand Richard Stallman is that he reminds Linus of his parents

That's a dangerous line of arguing. Why not go one step further and ask what what Stallman is rebelling against (man, that guy's parents must be weird!)? And what are your motivations? Mine? Why the heck are we fighting against industrial giants, and trying to defend rights that 99.9% of the population don't care about? Dr. Freud, please explain!

Even if you were right, that would have nothing to say about the validity of Linus' arguments.

The Torvalds Transcript (InformationWeek)

Posted Mar 28, 2007 20:30 UTC (Wed) by malor (guest, #2973) [Link] (1 responses)

Well, I can't talk about barbara's idea, as I don't know anything about his past in that regard.

I CAN, however, see that the arguments he presents are largely based in 'don't trust Richard Stallman', and that's why I made the observations I did. I don't he likes Stallman, and I can't blame him: Stallman's GNU/Linux name grab made ME furious, and I'm not even one of the people in the project.

Regardless, I think it's a bad reason to try to sabotage GPLv3. I don't KNOW that that's the reason, but it sure looks likely to me, given their past and the way he's framing his arguments.

I mean, he couldn't use the GPLv3 even if he wanted to, so why try to mess it up for everyone else?

The Torvalds Transcript (InformationWeek)

Posted Mar 29, 2007 9:13 UTC (Thu) by wilck (guest, #29844) [Link]

I can't see that Linus is trying to "sabotage" GPLv3. He says that he doesn't like it, and that he won't adopt it if key parts of its current draft for remain unchanged. Why shouldn't he?

The Torvalds Transcript (InformationWeek)

Posted Mar 21, 2007 17:42 UTC (Wed) by atai (subscriber, #10977) [Link] (8 responses)

Linus does not like GPL v3. Fine. The Linux kernel stays with GPL v2. But Linus seems to be on a path to try to derail GPL v3 in general. It is not clear why he wants to do this.

The Torvalds Transcript (InformationWeek)

Posted Mar 21, 2007 18:40 UTC (Wed) by emkey (guest, #144) [Link]

Because he thinks it is a bad idea?

There are real problems with the GPL v2. I don't think many people debate that. The real debate is over what some people view as the over politicization of the GPL v3. Personally I'd like to see a GPL v2.5 that corrects the issues with the v2 version without trying to save the world in one fell swoop.

The Torvalds Transcript (InformationWeek)

Posted Mar 21, 2007 20:00 UTC (Wed) by jengelh (subscriber, #33263) [Link] (6 responses)

The Linux kernel (binaries) stay GPL2 because combining all the source files that have "GPL2 only" with all those source files "GPL2 and later" only have one common denominator: GPL2. And it would be troublesome to contact every GPL2-only holder and ask them to relicense, not to mention that they might not agree.

The Torvalds Transcript (InformationWeek)

Posted Mar 21, 2007 21:04 UTC (Wed) by drag (guest, #31333) [Link] (5 responses)

Yep.

Which is kinda ironic since that is exactly how:

Now, totally independently of that, I'm doubly happy that I long, long since made that decision because at least the drafts of the GPLv3 have been much worse than the GPLv2 is. They've had glaring technical problems (license proliferation with not just one single GPLv3, but "GPLv3 with various additional rights and various additional restrictions"), and while I certainly hope that the final GPLv3 won't have those obvious problems, I've been singularly unimpressed with the drafts.

Stuff is going to work.
The lowest common denominator is always going to be GPLv3. It's never going to realy end up being any more restrictive then what that license allows.
This stuff doesn't realy affect the kernel any since the kernel is a isolated, lowest-level, sort of thing whose license is completely and utterly irrelevent to any other software that runs on it.

Currently one of the major problems with the GPLv2, which does not affect kernel developers in any real way, is the fact that right now people want to have open source licenses that are more permissive, but still retain some of the 'copyleft' effect of the GPLv2. They like copyleft, but what it weaker in this or that paticular way.

With the GPLv2 these people have no choice but to realy design their own licenses. So nowadays you have a number of fairly popular projects with their own paticular license that nobody else uses. They are usually GPL-like, but aren't realy compatable. Invariably these licenses are not paticularly well drafted or well thought out and tend to have lots of little bugs associated with them. That is the spirit of the license is great, but the legal effects of it do not nessicarially match the intention. It may be bad wording or what makes sense in California law may not make sense in India or Germany.

With the GPLv3 they are attempting to make it so that if you want to do GPLv3 + less copyleft, you just take the GPLv3 and then add your own little whatever. That way you have a strong legal standing for your license and you don't have to worry about incompatabilities with other Linux software. It's cheaper, easier, safer. Win-win.

I think that the problem is that with Linux kernel a lot of the issues regarding GPLv2 simply do not apply to them at all. It's major selling point, the licensing flexibility, is completely irrelevent to them. Also Linus isn't a lawyer, he has shown complete disregard for the licensing proccess to the point were in the past he has shown complete ignorance on how everything works. Saying he didn't have time to fly around the world for this or that meeting, were in reality discussion is all over mailing lists, irc, and such things.

So it's probably not the best idea to take legal advice from Linus. He is a very good programmer, but probably a pretty bad lawyer. (And I am neither, so take it for what it's worth)

the FSF wants the GPLv3 to be able to be LESS free then v2???

Posted Mar 21, 2007 23:43 UTC (Wed) by dlang (guest, #313) [Link] (4 responses)

this is the first time I've heard anyone argue that the FSF wants to make the GPLv3 have the ability to be _less_ free then the GPLv2.

I don't believe this at all. I think that what the FSF wants is for people to think that they are compatable, convert code from the other licenses to GPL+options, and then combine it with other code to convert it to GPL without options

if people only look at things one step at a time this is going to be legal, but if people look at the history of the code it won't be.

for example, if GPLv3 vanilla is not compatable with the apache license, but GPLv3+option is compatable, then code released under apache licenses can still never be combined with GPLv3 vanilla code, even though GPLv3+option code could be combined with GPLv3 vanilla code (resulting in GPLv3 vanilla code)

the FSF wants the GPLv3 to be able to be LESS free then v2???

Posted Mar 22, 2007 0:35 UTC (Thu) by njs (subscriber, #40338) [Link] (3 responses)

>if people only look at things one step at a time this is going to be legal, but if people look at the history of the code it won't be.

...this can't possibly be right.

No-one can remove restrictions from a license willy-nilly; the only time that's possible is if the copyright holder/original licensor explicitly granted that permission. Apache may be compatible with GPLv3+extra-clause, or GPLv3+extra-clause may be compatible with vanilla-GPLv3, but these cannot both be true at the same time (unless Apache is itself compatible with vanilla-GPLv3). Which of them is true depends on the wording of the extra clause -- if it says "you may remove this", then Apache ain't compatible with it; if it doesn't say that, then GPLv3 ain't compatible with it.

the FSF wants the GPLv3 to be able to be LESS free then v2???

Posted Mar 22, 2007 3:13 UTC (Thu) by drag (guest, #31333) [Link] (2 responses)

I was mistaken a bit. Mostly right, but somewhat wrong...

Well if you look at the license the amount of restrictions that you can add are _very_ limited.

It's a bit of a legal jedi mind trick realy.

You are allowed to add these specific restrictions under the GPLv3 license.

If you add them they then your software is has additional restrictions as allowed under the GPLv3 license.

However GPLv3 is still compatable. Because these GPLv3 allows these restrictions.

So if you take software from project "A" that has no additional allowed restrictions and combine it with project "B" that has additional allowed restrictions. Then you end up with software "AB" that is GPlv3 with additional allowed restrictions.

It's all still allowed and compatable.

Also keep in mind that the additional allowed restrictions is pretty limited to what you can add. It's mostly pretty non-important stuff. The only one that is scary is the additional patent one, but it's critical if your going to be compatable with Apache, Mozilla, et al.

the FSF wants the GPLv3 to be able to be LESS free then v2???

Posted Mar 22, 2007 3:17 UTC (Thu) by drag (guest, #31333) [Link]

Also another thing to keep in mind is that the GPLv3 has a restriction that if you add additional terms (either permissions or restrictions) then it has to be in one central location in the code base.

So it's not like if your worried about this that you would end up having to grep through header files or whatever to find out what your dealing with.

the FSF wants the GPLv3 to be able to be LESS free then v2???

Posted Mar 25, 2007 0:28 UTC (Sun) by vonbrand (subscriber, #4458) [Link]

So if you take software from project "A" that has no additional allowed restrictions and combine it with project "B" that has additional allowed restrictions. Then you end up with software "AB" that is GPlv3 with additional allowed restrictions.

And then nobody can take changes back from "B" to "A" without changing "A"'s license... or doing a detailed audit to find out where the changes originated, perhaps in "D" which is compatible with "A"

Somebody comes along and uses "A" stuff to create "C" with other added restrictions. "B" stuff can't be used with "C" now.

Can you spell "balkanization"?

The Torvalds Transcript (InformationWeek)

Posted Mar 21, 2007 20:18 UTC (Wed) by mmarq (guest, #2332) [Link]

"" Me, I just don't care about proprietary software. It's not "evil" or "immoral," it just doesn't matter. I think that Open Source can do better,... "

Agreed!... so the FSF success through the years rests on the *open and easy transfer* and *scale economy* of the FOSS model, and not on its *moral* attributes. And being absolutely true, as i belive it is, than no war against FSF campaigns is needed or even logic.

"" I absolutely love the GPLv2 -- because it embodies that "develop in the open" model. ""

Why cant v3 embody the same exact model, plus some extensions to address patents and DRM in a compatible way,.... extensions needed to plug legal holes capables of preventing the code of being used in a legal way, even by their own developers.

"" I don't want to limit what people can do with my code. I just want their improvements back. But if they do something stupid with it, that's their choice. ""

hmmm... even in a scenario where every computational machine from smartphone to supercomputer adopts draconian DRM protections, build with v2 code, and though all improvements are available, they are useless, because they can only run where and when the *keymaster* says so...

"" Now, I certainly hope that the final GPLv3 will simply not have those kinds of restrictions, so I'm not totally against it ""

Sure! but being so, logically, it must have some kind of *provisions* ... restrictions not by some political whim of the moment or against any kind of use, but *PROVISIONS AGAINST* those that want to *RESTRICT* the use of the software, stupid or not, including riping and backing up a DVD movie (which in my country isnt ilegall as long as i have an original copy).[why not ?]

"" "I will get more effort out of other people working on it too, than I have to give back."... "positive feedback!" ""

yeah... but how much more difficult it would be with same wild beasts roaring out lout, that FOSS is an IP thieves lair!... hmmm... cant a license simply state, upon recognition of the participant partys, that patents are non-fee for a particular repository,... and everythinhg will be as before, meaning the beasts would be inside cages ?...

( IANAL, but cross-licencees with full right to any patentend code used or adopted can endorse the rights to a FOSS repository if they use it or adopt it(for sale) ? )

Torvalds licensing ignorance

Posted Mar 21, 2007 20:36 UTC (Wed) by bug1 (guest, #7097) [Link] (8 responses)

Why does Linus still act as if he has a choice as to how the Linux kernel is licensed ?

Legal it is very difficult to move a collaboratively developed project to a new license unless they have assigned copyright to a common body, they would need to track down all the copyright holders (like Mozilla tried to do many years ago) and if they couldnt contact everyone, or if some people didnt want the license change there code could not be in the relicensed project. Not very practical.

For all his claims of pragmatism over the years, why cant he see the pragmatism in the "or later" part for projects who dont assign copyright to the project, why cant he accept that legal its not his choice anymore...

Torvalds licensing ignorance

Posted Mar 22, 2007 0:41 UTC (Thu) by njs (subscriber, #40338) [Link]

He does point that out too sometimes. Fortunately, he gets the same result either way, so he doesn't have to choose. And even if he can't change the licensing, his thoughts on why he wouldn't even if he could may still be valuable and interesting.

OTOH, the Mozilla project *succeeded* in relicensing everything[1]. It was painful and took a while, but this sort of thing is possible if necessary. (Note that there is a fallback if someone cannot be contacted -- you can remove/rewrite their code. Not pleasant, but possible.)

[1] http://weblogs.mozillazine.org/gerv/archives/2006/03/reli...

Torvalds licensing ignorance

Posted Mar 22, 2007 10:51 UTC (Thu) by Wol (subscriber, #4433) [Link] (6 responses)

It's actually quite easy to relicence a project ... particularly when it has a high code churn like Linux.

All it takes is Linus to say "I'm relicencing my code as 'v2 or v3'", and to demand that all new code is licenced the same.

Then you contact all the other "v2 only" people and ask them to relicence their old code.

Finally, you wait (probably not that long) for all the "v2 only" authors that couldn't or wouldn't relicence to be written out of the system. And if you publish their names I'm sure a janitor project would rapidly spring up to do that ...

Bingo. The kernel is now "v2 or v3" right through, and can be changed to "v3 only".

Cheers,
Wol

Torvalds licensing ignorance

Posted Mar 22, 2007 12:47 UTC (Thu) by jospoortvliet (guest, #33164) [Link]

You make it sound easy ;-)

You're right, this is the way to do it, but it would take years and years...

Torvalds licensing ignorance

Posted Mar 22, 2007 16:57 UTC (Thu) by viro (subscriber, #7872) [Link] (4 responses)

Sure, I'll even help you to start that list: Al Viro <viro@[anything]>
There's your first entry, now you have something to rewrite. Quite
a bit, actually, so by the time you are through with that you'll
probably have more list entries submitted. Have fun.

Torvalds licensing ignorance

Posted Mar 22, 2007 17:22 UTC (Thu) by bronson (guest, #4806) [Link] (1 responses)

Hang on, Al. I agree that the GPLv3 as of the last draft is really overreaching. But what if someone writes the license that should have been the GPLv3? A respin, not a ground-up rewrite. Call it GPLv2.1. It fixes up a few ambiguities in GPLv2, clarifies some language for international courts, but is otherwise is substantially exactly the the same as the GPLv2.

Would you be against relicensing your code to that?

Torvalds licensing ignorance

Posted Mar 22, 2007 17:41 UTC (Thu) by viro (subscriber, #7872) [Link]

I don't believe that FSF will go for that, at least not until they
*really* give up all hope for v3 being feasible. And probably not
even then, due to perceived loss of face. IOW, it's even less likely
than relicensing of the kernel. Sure, if we run into real trouble
with v2 (i.e. serious failure in court somewhere) we'll have to do
something that would fit your description. But that's (a) unlikely,
as far as we know and (b) it would be v2 + explicit modifications,
not FSF-blessed v2.1.

Torvalds licensing ignorance

Posted Mar 23, 2007 4:48 UTC (Fri) by ldo (guest, #40946) [Link] (1 responses)

Why do I get the feeling somebody thinks he's indispensable?

Torvalds licensing ignorance

Posted Mar 23, 2007 6:08 UTC (Fri) by viro (subscriber, #7872) [Link]

Not at all. So, are you volunteering for rewrite? Of course it's
doable, just... will take some work.

GPLv2 enforcement

Posted Mar 21, 2007 21:00 UTC (Wed) by orospakr (guest, #40684) [Link] (12 responses)

I just wish Linus would enforce the existing GPL *V2* license.

Non-free kernel modules (in particular, non GPLv2 compatible ones) are copyright infringement, and should be greeted with lawsuits!

This bullshit about "not originally developed against the kernel API" noise is nonsense.

The very act of porting makes it a derivative work.

If I took my proprietary MFC application, ported it to the GPL version of Qt, and then sold binaries without source code, Trolltech would haul my ass into court. And rightly so.

Why should Linux be any different?

GPLv2 enforcement

Posted Mar 21, 2007 23:42 UTC (Wed) by ekj (guest, #1524) [Link] (11 responses)

It's not quite that simple. The *LAW* (and judges interpreting same) defines what is, and what isn't a "derived work". If the law says something or other is not a derived work, there's nothing you can do to change this.

A driver written for Windows, with no knowledge of, interface to, or interest in Linux is clearly not a derived work of Linux.

If someone writes a wrapper (and releases this under GPL) that nevertheless allows the former to be loaded by Linux, then what ?

The wrapper *is* a derived work -- but that's ok, since it's released under GPL anyway. The loaded driver very clearly is not, so whatever the GPL says about it has no influence on it.

GPLv2 enforcement

Posted Mar 22, 2007 8:25 UTC (Thu) by bug1 (guest, #7097) [Link] (7 responses)

The wrapper is probably a derivate work of both Linux and the windows driver.

If it is, then to distribute the wrapper in compliance with GPLv2 you have to have the source to the windows driver as well as Linux.

GPLv2 enforcement

Posted Mar 22, 2007 9:09 UTC (Thu) by malor (guest, #2973) [Link] (3 responses)

You're clearly and unambiguously wrong about this.

Say, oh, Broadcom releases Windows drivers for their wireless chips. They work nicely in Windows.

Then the LinuxAnt project comes along and writes a wrapper that allows you to load the the Broadcom driver under Linux. That driver knows nothing about Linux; it thinks it's being loaded under Windows.

LinuxAnt is required to release their source code, because their module is obviously a derived work of the kernel. They are not, however, required to do so with the Broadcom source code, and neither is Broadcom. It wasn't developed for Linux and knows nothing about it, so it cannot be a derived work. It is really that simple.

By your argument, if we write a special wrapper to let Windows itself run under Linux, suddenly Windows becomes a derived work and Microsoft loses the ability to sell it without source code.

You're really trying to go after NVidia's binary blob approach here. I think they are immune; it's my understanding that their binary is just the Windows driver, and they wrap it in a glue layer. That doesn't make the binary blob a derived work.

GPLv2 enforcement

Posted Mar 22, 2007 11:00 UTC (Thu) by Wol (subscriber, #4433) [Link]

Plus it's only a copyright infringement by the DISTRIBUTOR.

If the distribution downloads the blob from NVidia on installation, who's breaking copyright? Not the distro, they never mixed GPL and proprietary. Not NVidia, they own it. And not the user, they had NVidia's permission.

It is PERFECTLY LEGAL to mix GPL and proprietary code. You just can't distribute the result. If I wanted to ship a program that was a mix of proprietary and GPL, I would simply ship a build environment. All the GPL as source, the proprietary as object modules. Get the user to run a make script to build the program. ALL LEGAL AND ABOVE BOARD.

It just wouldn't look good to a user, and I wouldn't do it because I don't think it's in the spirit of the GPL. But it's perfectly legal ...

Cheers,
Wol

GPLv2 enforcement

Posted Mar 22, 2007 11:00 UTC (Thu) by thebluesgnr (guest, #37963) [Link]

If LinuxAnt distributes only the GPL loader and nothing more it is obviously not violating the license.

The problem is distributing the loader and proprietary bits together, as nVidia does.

Also, take for example a distribution that ships such driver with the kernel. They distribute a version of Linux that "can accelerate nVidia cards", which requires modifications to Linux which they don't make fully available. IMO they're clearly violating the GPL.

GPLv2 enforcement

Posted Mar 22, 2007 11:07 UTC (Thu) by bug1 (guest, #7097) [Link]

According to the US gov, a derivative work is, "a work that is based on (or derived from) one or more already existing works" (1)

Could the LinuxAnt wrapper (compatibility layer) be written without the windows broadcom driver, without the broadcom driver does the compatibility layer have a purpose ?

I think most real people can see there has to be some sort of dependency there... But maybe the compatibility layers dependence is through a standard interfaces, which are allowed exceptions in most countries i believe. (but im not sure how that effects the GPLv2).

You misunderstand me in your windows example, what im claiming is that the compatibility layer is the derivative work of both layers, the GPL requires athe source of dependencies be distributable. There are no requirements placed on either layer unless they want to be legally compatible with the compatibility layer.

If Microsoft doesnt distribute this hypothetical microsoft/linux compatibility layer they dont have to do anything.

But its all depends on the interfaces, if it uses standard interfaces to windows then the compatibility layer is probably ok, but as you suggest in your first post, only a judge can say for sure.

Another example would be shell script, it serves no purpose without a shell, so it is dependent on _a_ shell, but the shell is a standard interface (POSIX SUS), there are any number of shells that can interpret it, so the script isnt a dependency on any specific shell.

Can the linuxAnt compatibility layer claim the same sort of independence from the windows broadcom driver, does it access the driver directly, or _only_ through standard windows calls ?

1) http://www.copyright.gov/circs/circ14.html

GPLv2 enforcement

Posted Mar 22, 2007 17:43 UTC (Thu) by drag (guest, #31333) [Link] (2 responses)

I was pretty convinced of this also for a while.

But I was corrected.

The FSF and friends stating that linking code against GPL code would require that all the code is GPL is simply _wrong_ and _misleading_.

The reality of the matter is that it's up to a Judge to decide what is and what is not derived work. The GPL is a copyright license and by it's nature is limited by the scope of copyright.

The issue of 'derived work' is _intentially_ left up the open by law and is completely a matter of jurisdiction, interpretation and such things. It is not something that the GPL or FSF or anybody else has any control over.

So the thing is is that you _can_ violate the GPL by having a propriatory kernel module, but not nessicarially all kernel modules are in violation.

The Nvidia Linux driver, at it's core, is the same code as what is used in the Windows driver. Not that I know this for a fact, but it's just that that's what I am told.

Also the OpenAFS driver is based on code that existed long time before Linux even existed.

Both of these drivers are using code that existed completely independant from any sort of code that is in the Linux kernel.

Sure they know about kernel internals and such things, but that is required to make the software compatable and modifing code for compatability doesn't mean that it's derived, right? Of Course not.

Now with Linux you have all this modularity. You have the VFS and all sorts of functionality that people remove from individual drivers and then lump it all together in one system.

Like sound drivers and the sound core.
Or Sata drivers and libata and the scsi subsystem.

Now if I was to try to make a propriatory sound card driver, but used a crapload of alsa functionality to do that, then I would more then likely be in violation of the license and any kernel developer in any parts of Alsa would be be free to sue the pants off me.

GPLv2 enforcement

Posted Mar 23, 2007 6:39 UTC (Fri) by gmaxwell (guest, #30048) [Link] (1 responses)

>The FSF and friends stating that linking code against GPL code would require >that all the code is GPL is simply _wrong_ and _misleading_.

>The reality of the matter is that it's up to a Judge to decide what is and >what is not derived work. The GPL is a copyright license and by it's nature >is limited by the scope of copyright.

You were corrected incorrectly, or at least incompletely.

When someone other than the copyright holder distributes GPLed covered software, he does so only by the good graces of the license. The license could, for example, demand that the software only be distributed inside fortune cookies.

A court might decide to invalidate an instance of the license, but without a license you are not able to distribute the software, at least not without performing an actionable copyright infringement.

Thus, in cases where both GPL and non-free parts are distributed by the same party the license can effectively enforce whatever byzantine interpretation of derivative is desired.

The GPLv3 specifically avoids standard legal terms in order to avoid creating confusion like this. Instead the license adopts new terms and defines them clearly in the license.

Obviously the case of separate distribution is another matter entirely, but I'm not aware of anyone sane trying to argue against such cases on legal grounds.

GPLv2 enforcement

Posted Mar 23, 2007 7:22 UTC (Fri) by drag (guest, #31333) [Link]

Well distribution is covered elsewere the license also. Don't forget that.

Also, as you are aware from Microsoft EULA goes.. just because somebody has something in a license it doesn't mean that it's enforcable, even if you agree to it. I don't know how far that goes.

But there is a nasty little clause in the GPLv2 that goes as follows:
(in section 2)

""In addition, mere aggregation of another work not based on the Program
with the Program (or with a work based on the Program) on a volume of
a storage or distribution medium does not bring the other work under
the scope of this License.""

So you _can_ distribute non-derivative (see the portion immediately before the part I quoted.) work along with GPL'd software then the GPLv2 license expressly states that it is not covered under the terms of the GPLv2 license.

In other words,
A: If code is non-derivative then it's not covered under the terms of the GPLv2.
And
B: The license expressly allows it's distribution along side non-GPL works.
So
C: Therefore if a kernel module is non-deravitive then it's legal to distribute it along with the kernel.

IF its non-derivative. It's gray.. so some can and will violate while others won't.

("Deravitive" IS a well established legal term, btw. It's defined as part of the US copyright code. Of course in other countries this may vary, but I am operating under US law.)

http://www4.law.cornell.edu/uscode/html/uscode17/usc_sec_...

GPLv2 enforcement

Posted Mar 22, 2007 8:29 UTC (Thu) by MKesper (subscriber, #38539) [Link] (2 responses)

This does only apply to things like ndiswrapper, but clearly not to such things like ATI or nvidia (or whatever...) binary drivers. They _have to_ know about kernel details.

GPLv2 enforcement

Posted Mar 22, 2007 14:02 UTC (Thu) by i3839 (guest, #31386) [Link] (1 responses)

Not really. The underlying hardware is the same, so they make certain assumption in their code. How to get to access to the hardware might depend on what kernel is run, but not much more. For that the Nvidia driver has a BSD wrapper.

The GPL doesn't disallow that wrapper, nor does it disallow loading binary only modules at runtime. That wrapper doesn't disallow being linked to the Nvidia binary either.

So if the driver is written independently from Linux (which seems to be the case, as it's a modified Windows driver), Nvidia really doesn't break the GPL.

It only becomes interesting when the driver is distributed together with the kernel, as then the question is whether it's mere aggregation, or a combined work. But even then it isn't Nvidia who's breaking the GPL, but the distributors.

GPLv2 enforcement

Posted Mar 29, 2007 20:15 UTC (Thu) by TRauMa (guest, #16483) [Link]

The BSD wrapper is a derivative work of the kernel. NVidia doesn't have the copyright to the kernel, so it's no BSD wrapper at all, it's a GPL wrapper. Now it's sole purpose is to link itself to a non-GPL-compatible blob... I think a linux copyright holder could make a case out of that, if so inclined?