Linux's Hit Men (Forbes) [LWN.net]

Linux's Hit Men (Forbes)

Posted Oct 14, 2003 14:20 UTC (Tue) by coriordan (guest, #7544) [Link]

At the end of the story, there's a link to a comment form. I've sent them my thoughts. Just remember to be calm when writing a comment.

Is this calm enough?

Posted Oct 14, 2003 15:18 UTC (Tue) by cr (guest, #3685) [Link]

You want they should steal it? Comrade??

This author's viewpoint is silly. He's saying that Cisco should
be allowed to take other people's code (Linux), and violate
those peoples' copyrights to take that code private. That's
theft, Comrade, no matter who you call names. Cisco is seeking
to get and keep unearned wealth in trying to renege on the
covenant-of-sharing which the people who wrote Linux set as the
price of their code.That bargain is as American as barn-raising,
though apparently some people have real problems in
understanding the ethics of it.

Theft?

Posted Oct 14, 2003 18:02 UTC (Tue) by vonbrand (subscriber, #4458) [Link]

Technically (yes, I read Groklaw ;-) it is "missapropiation", not "theft".

Is this calm enough?

Posted Oct 14, 2003 21:50 UTC (Tue) by mmarq (guest, #2332) [Link]

But they could had already done that!

Can you garantie that among the proprietary code base there isnt "A LOT" of stollen code, from GPL code ?

That is why Linux/OSS has to win all the way from servers, embeded and specialy Desktop,..., and why the SCO & Ma$ter and others, that stand to lose in stock and legal, if M$ collapse in its monopoly, are giving this war.

Instead of bad words, that only bring mor FUD, what is really needed is a more global focus on the real enemy, and his Desktop monopoly.

If Linux/OSS can came close to windows support in terms of Hardware, then M$ is history!

Humor

Posted Oct 14, 2003 15:24 UTC (Tue) by ncm (subscriber, #165) [Link]

There's no question but that the Forbes editors are hostile. They are well-known kooks, right up to the publisher, Steve Forbes, son of the founder. (His father wasn't a kook.) The magazine has an editorial slant just shy of the Weekly World News. To get an column published in it -- and it pays well -- you have to play along with the publisher's whacko biases, which this column demonstrates clearly. Nonetheless the author is a professional, which the column also demonstrates clearly by his careful handling of quotes and facts. Any rational reader can see right past the slant, and recognize the truth underneath.

It's a lot like reading the Chinese "People's Daily". The reporters know not to mess with the party line, but they don't care enough to alter the facts to align with it.

Humor

Posted Oct 14, 2003 20:25 UTC (Tue) by torsten (guest, #4137) [Link]

"Nonetheless the author is a professional, which the column also demonstrates clearly by his careful handling of quotes and facts."

Eh? Did you read it? It's riddled with factual errors. Start with the statement that the FSF created the GNU GPL License in 1991 and work your way down.

FSF != KGB

Posted Oct 14, 2003 14:54 UTC (Tue) by jensend (guest, #1385) [Link]

People associate sharing with Marx because they don't want to share and so they want to make sharing look evil. Marxism has always been a mess, even before Communist countries further botched it up. It combines an uneconomical economics (the time theory of value) with a method of philosophical reasoning which is both unphilosophical and unreasonable ("dialectical materialism"- Hegelianism was already pretty far-gone before Marx decided to "turn Hegel on his head") and a practical ethics which is unethical and impractical (recommendation for the seizure of power by the proteleriat and the expectation of a subsequent utopia). Exhortations to be charitable, religious societies of community such as monasteries of all different religions or the Christian community of the book of Acts, and socialistic government services models are all much harder to make bogeymen out of.

FSF != KGB

Posted Oct 14, 2003 22:19 UTC (Tue) by mmarq (guest, #2332) [Link]

Have you read Marx ?

Actually his central point of doctrine is not sharing, but equalizing everything to the more numberous faction!?... at 1900 that was easy:- equalize by the proletary.

But Marxism never had nothing to do with freedom of choise and sharing, by the contrary. Marxism is a social ideology of war for power, that could easely be found at the bed chambers of Hitler, Staline, Mossuline, Mao Tse Tung, and any modern War Lord from eastern to western includind Europe and the USA;... You change Proletary by Capitalists ( pretending they are numberous and oppressed!!) and Marx theory suits every flavor from right to left including the one of Steve Forbes and the like.

My letter to the Forbes author

Posted Oct 14, 2003 15:22 UTC (Tue) by dank (subscriber, #1865) [Link]

Hi Mr. Lyons,
I hope you don't get too much nasty mail from Linux advocates; I
apologize in advance for the ill-tempered among us!

Your article made two basic errors which I encourage you to learn
more about.

First, it failed to distinguish between two very different situations:
garden variety GPL violations (which are generally uncontroversial and
easy to resolve), and binary kernel modules.

The NuSphere case was a garden variety GPL violation. The company was flagrantly
violating the GPL, and had plenty of warning even before they started
that there would be vigorous enforcement. NuSphere really has no grounds
for complaint; they did something obviously wrong, and got caught.
Note that the GPL doesn't require *all* source to your product to
be released; you only have to release the source code that is tightly
"linked" in a precise technical sense to existing GPL'd software.
PeopleSoft, for instance, doesn't need to release any source code,
even though their software runs on Linux, because simply running software
on Linux doesn't "link" it to GPL'd software in the sense needed to trigger
the GPL's source release requirement. I can try to find a clearer
explanation of the distinction if you like. The upshot is,
in future articles, you should recognize these and not show too
much sympathy.

The Linksys case is quite different. It had two parts:
a garden variety GPL violation, which they have now resolved by posting the code,
and a "binary kernel module" issue, which is much less open-and-shut.
You see, Linus Torvalds granted an exception to the GPL years ago
for companies shipping binary kernel modules, so companies had a reasonable
expectation that shipping them would not trigger the GPL's source code
release requirements. Now, however, he seems to be no longer
granting that exception. This is big news, and deserves a nuanced story!
I'd be glad to help you explore it. One good source is the comments in
http://lwn.net/Articles/53138/ (well, one in five of them, anyway; there's
lots of noise). Companies caught in this bind deserve a sympathetic
treatment.

Hope that helps!
Thanks,
Dan

My letter to the Forbes author

Posted Oct 14, 2003 16:15 UTC (Tue) by andrel (guest, #5166) [Link]

Dan,

I don't belive Linus ever granted the exception for binary modules, i.e. he never granted the right to package them together with the kernel and distribute as a whole work. A careful reading of Linux's COPYING file leads me to believe exactly the opposite -- it explicitly says that the source code of modules (that is the word used in Section 3) distributed with the kernel must be provided. Where do you think Linus granted such an exception? Please, a URL to the exact document, not comments by someone who thinks they remember Linus saying something similar.

Besides, why is it in Linus's perview to grant such an exception? He's only one of many copyright holders on the kernel -- that was a deliberate decision on his part. Many other copyright holders would balk at such a change in the licensing terms.

Thanks,
--Andre

My letter to the Forbes author

Posted Oct 14, 2003 17:06 UTC (Tue) by Ross (subscriber, #4065) [Link]

I also remember Linus creating an exception for modules. I remember RMS
saying that it was a bad idea because it would lead to confusion and
blurred boundries later. That was when I used to read the kernel mailing
list. I guess it was sometime in 2000 or 2001, but maybe earlier. He
later made a more official statement which I think he said was
intentionally put in the public record to avoid problems down the road.

Later I remember Linus clarifying that modules which used hooks which
were added into the kernel were taboo because they were expanding the
interface that modules use past the point where they act like separate
pieces of software.

More recently there has been the tainting mechanism and the GPL vs. non-
GPL symbol marking. Around this time Linus made a statement to the effect
that binary-only module authors shouldn't assume that they are not
derivative works for reasons outside of the GPL. I assume he was making
a statement intended for specific companies that may have been copying
code and thinking the binary-only module exception would protect them.
Maybe he meant it in a more general sense.

Does anyone remember specific quotes? If not I'll check google but there
are a ton of messages on LKML so it might be hard to find the statements
in question.

My letter to the Forbes author

Posted Oct 14, 2003 17:30 UTC (Tue) by dlang (subscriber, #313) [Link]

all that Linus has said is that a modules is not nessasarily derived from the kernel.

that means that if the module code was really developed independantly then making it a loadable module doesn't mean that it must be GPL'd.

this doesn't mean that all binary modules are Ok, becouse it's very easy for them to BE derived from other kernel code.

Linus has said that the module boundry is a defined interface and things that use that interface are not derived from the kernel by the mere fact that they use that interface any more then userspace programs are derived from the kernel. He has clarified his position in the last couple of years by pointing out that just becouse it is a module (or a userspace program) it doesn't mean that it ISN'T derived from the kernel, what makes the difference is how it was developed.

talk to your lawyers, if you feel that you can make the case that your code was developed independantly from the kernel then a binary module is Ok, if you developed your code by copying from other drivers that it probably isn't Ok, and if you require modification to the kernel for your drivers to work then you are probably not Ok

My letter to the Forbes author

Posted Oct 14, 2003 17:47 UTC (Tue) by Ross (subscriber, #4065) [Link]

Yes, that's what I am referring to. He said that modules are not always
derivative works. He even said it was ok and that he supported the
ability to port binary-only drivers to Linux. I'm not talking to any
lawyers about this because I have no intention of writing any binary only
modules.

My letter to the Forbes author

Posted Oct 15, 2003 8:06 UTC (Wed) by macc (subscriber, #510) [Link]

As far as i remember
Mosix were the first to trigger this.

their code comprised of hooks in the kernel
and of kernel modules that used these hooks.

They published the gluecode they had added to the kernel
and kept the module code closed.

this was deemed not proper.

NonGPL Modules are permissible as long
as they only use the standard interface to the linux kernel.

Posted Oct 14, 2003 17:19 UTC (Tue) by stuart (subscriber, #623) [Link]

Like andrel, I point out to you that Linus is not the sole copyright holder of the Linux Kernel so basically whatever he says is largely irrelevant, as any intelligent person could work out.

Furthermore, these companies have a thing called 'lawyers.' I think lawyers should be able to understand copyright law. Therefore the company is well equipped to realise that it can steal all the public domain code it wants but hands off GPL'd code unless it complies with each copyright holder's wishes.

Just because I give apples away for free, it doesn't mean that you can take apples from my friend's tree without his permission.

Stu.

My letter to the Forbes author

Posted Oct 15, 2003 15:30 UTC (Wed) by Wol (guest, #4433) [Link]

Put simply, Linus granted an exception (with kernel 0.99, if you read the COPYING that accompanies the kernel, iirc), for modules which ABIDE BY THE PUBLISHED OFFICIAL INTERFACES.

The point here is that Broadcom MODIFIED THE INTERFACE, and did NOT publish the modification. So their mods to the kernel are most definitely in violation of the GPL. As these were needed to link to their binary module, this places their binary module in breach of the GPL too.

Now that they've published the mods (they were *forced* to, sadly), they'd like to say "it's a published interface" and close off the chapter. Unfortunately, that isn't good enough - it's not the official kernel interface and as such isn't exempt.

Don't forget, when Broadcom were *pushed* into releasing source, it was discovered that compiling the source did not result in a kernel identical to that in the routers. THAT SCREAMS BAD FAITH!

So tough. If Broadcom want to scream, that's their problem. If they're stupid enough to try to pull a fast one, why should we be bothered when they don't look where they're going and run full tilt into a wall? That's their problem, not ours.

Cheers,
Wol

Linux's Hit Men (Forbes)

Posted Oct 14, 2003 17:53 UTC (Tue) by iabervon (subscriber, #722) [Link]

What I find interesting is that Forbes keeps presenting IBM as a free software company and industry outsider, despite the fact that it is a licensee of commercial UNIX, is and has always been the biggest computer company in the world, and has been in the Forbes 10 since the dawn of time (or at least, as far back as Forbes has the list online).

Furthermore, they talk about the risk that companies take in basing their products on GPL code, but ignore the risk that companies take in basing their products (such as AIX or Irix) on commercial code.

On the other hand, I think that this article makes a good point about the FSF's enforcement actions: the outcomes ought to be publicized. There is now a substantial amount of precedent as to what can and cannot be done with GPL code in the FSF's opinion, and a substantial amount of experience in how to recover from mistakes made in making products which include GPL code. However, none of this (as actually negotiated) is available to companies which are trying to plan the use of GPL code. It's very difficult to get a clear statement of the responsibilities of a company using GPL code in a commercial product with proprietary portions, despite the fact that this has clearly been hashed out hundreds of times by the FSF with individual companies. Beyond that, the lack of information about these cases means that companies don't necessarily realize that the GPL is enforced, and developers don't realize that the FSF actually does much legal work.

Personally, I think the FSF really ought to have a simple set of guidelines: "When in doubt, don't base anything on GPL code. Whenever possible, be the end user of GPL code. To interact with GPL code, write to published APIs and don't modify the GPLed code." Really, that's what you need to know first. The free software movement as a whole is based on convincing people to GPL their code and no longer be in doubt. But it benefits no one for people to base their products on GPL code without understanding the obligations this entails.