OSNews

...why should one do so when the act itself (the hijacking of the rights to the code in question) never should have happened and since the news value of this is legitimate enough for them to discuss openly in a forum?

Never attribute to malice that which is readily explained by stupidity. In this particular case, the driver had BSD bits and GPL bits in the source and someone removed one of the notices by accident (presumably not realizing where the BSD-only chuck started and ended).

Seeing as how the issue was fixed within a few minutes of it being pointed out, I don't think it's a big deal.

There doesn't seem to be any stealling going on at all.

It seems that is was originally licenced under the GPLv2 (and also under the BSD) and it's been included in the -mm kernel under the GPLv2.

I don't see what the problem is.

http://kerneltrap.org/node/14243

Maybe this clarifies it. There is no "it seems" in copyright law.

Theo is wrong about the removal of the BSD-part of Sam Leffler's file (ath5k_base.h). The moment you use, modify or distribute the file you have the right to choose one of the licenses (or both if you want to do keep the dual-licensing).

Theo claims incorrectly that the right to choose one of the licenses don't exists UNTIL a modification has been made. Truth is that it kicks in the moment, you use, modify and/or distribute the file. So everything has been done 110% correct for the BSD-licensed parts. The problem is the three files NOT under the BSD and NOT under the GPL. Those three files have had their license violated. That's clear. But Theo is rambling like a stupid maniac in regard to the dual-licensed files.

Yes, the patch as emailed would be a violation of the BSD license. However, the patch has not been accepted into the Linux kernel, and as we can see from the ensuing LKML thread, the licensing details will be ironed out before anyone even thinks about merging it.

LKML is a public mailing list. Anyone can submit patches, and the only thing the kernel community can do about bad patches, including those with licensing issues, is to reject them with or without comments. The kernel community has already concluded that this patch cannot be accepted as written.

The difference between this situation and the BCM thing is that BCM was committed to OpenBSD's official CVS repositories. Furthermore, the OpenBSD project had no legitimate use for that code, whereas the Linux community can use ath5k under an appropriately derived license.

The Linux project is permitted under the BSD to add the GPL to this code. Depending on how it is added, the resulting code could be dual-licensed or GPL only. When used under the GPL (in either case), the clauses retained from the BSD are redundant. That doesn't mean they can be removed, however.

So, functionally, the Linux developers are correct. They can redistribute ath5k under the GPL or under a BSD/GPL dual-license. But the patch as emailed is not the correct way to do this. The clauses from the BSD must be retained in order to satisfy the redistribution requirements.

It's likely that the act of emailing improperly licensed code to a public mailing list constitutes copyright infringement on the part of the sender. However, is the project or its various hosting and mirroring services liable for code appearing on the mailing list? Can't they simply disclaim responsibility like OSNews does for comments?

BTW, the BSD community has repeated held that BSD code can be relicensed by adding restrictions (GPL, proprietary, etc). Theo is dead wrong when he says "none else can add a GPL to it." Proprietary vendors do this with BSD code all the time.

Edited 2007-08-29 15:30

can't we give the parties involved some minimal amount of time to sort this out *before* posting this flamebait?

why? when this sort of thing happen in April of this year between Michael Buesch and OpenBSD. Michael Buesch posted all over the mailing lists when he was accusing Openbsd of GPL license and
Copyright violations of the bcw driver. he did not give time to work things out.

It would also be nice if you didn't make it sound like your title, 'Linux Developers Steal OpenBSD Code for Wireless Driver', is a quote from the original when it isn't. The original headline is simply 'Linux Driver Copyright Violation', nothing about stealing.

again back in april,Buesch made sound like openbsd was stealing. when it turned out to be only an honest mistake on the dev part

Edited 2007-08-29 16:56

I seem to remember there were 60 something commits to that developers tree in violation of the license. Maybe 5-10 would be accidental, but 60+ *different* commits?

No, that was intentional on the BSD developer's part. The only sad part is that he quit working on it when the linux bcm43xx developer offered to dual license the code.

Some people call this *news* ;-)

Some people call it tabloid press, gutter press etc. In other words, a sensational, gossip-filled publication that bends the truth and particular circumstances, or pretends to stumble upon them at the exact point in time that will provide maximum impact; common sense and civility be damned.

There seem to be entirely too many of this kind of stories on OSNews lately.

The Internet has never been a bastion of journalistic integrity, or civility for that matter.

This is democracy in action. Issues and news acted on by the masses without any filtration or abstraction. As an an Internet denizen I would expect nothing less then rabble rousing, sensationalism, and shoddy fact finding.

Seriously, this is better the soap operas.

Meanwhile, I submit important news about OOXML's difficult road to ISO standardization (such as the fact taht India and Brazil have rejected it) and they get rejected...

Oh well, to stay on-topic, it does seem as if there were some improper stuff going on here, but that in the end the Linux kernel devs will sort it out so as to stay on the legal side. I do agree that the headline is too sensationalistic (and that Theo de Raat has no manners), but I don't disagree that this is in fact news - if a bit gossipy.

Not to mention the bribing of supporters in Sweden...

Ah yes, a proud moment for my homeland :/

Don't cry baby

http://www.idg.se/2.1085/1.118680

That's exactly what I had expected from a Scandinavian country. The end times ain't here yet

"""
but I don't disagree that this is in fact news - if a bit gossipy.
"""

Everything that happens in the Open-Source fish bowl is "news".

If Theo de Raat encounters hemorrhoid problems which affect his work, that's "news".

I'm just not sure that a story about it on OSNews would be appropriate.

If such a proctological issue should become serious and chronic, then a story as to its possible effects upon the greater community might be warranted.

I should hasten to point out that I have no evidence... beyond what the rest of you have available to you, to suggest that Theo might have such an... indelicate... rectal issue.

For all I know, it could just be occasional irregularity. ;-)

Edited 2007-08-30 16:16

"For those parts, the Linux dev was perfectly entitled (legally) to choose only the GPL licence for his derived work."

For his derived work yes, remove the BSD license from the file, no. Only the copyright owner can remove it or give permission for its removal.

Wrong. Dual-licensing means that you have to obey *either* the BSD licence *or* the GPL licence. If you choose the BSD licence, then you're free to do all those things that the GPL says you can't, like releasing binary-only versions; correspondingly, if you choose the GPL, then you're free to do the practically the only thing that the BSD licence says you can't, which is to remove the licence text.

As I said before, the Linux dev did nothing wrong with the dual-licensed bits of the code (though obviously he was wrong to do this for the BSD-only parts).

You are wrong.

For the dual-licensed files you can remove one of the licenses the moment you use them. You don't have to make changes in order to change the licenses. You can remove the license you don't want to use the software under. So Jiri Slaby was perfectly in line for the two dual-licensed files. Especially the first one, where he is the sole author.

The problem is solely the three files NOT licensed under the GPL and NOT licensed under the BSD (but under a modified MIT-like license, neither MIT nor BSD). Those three files had their licenses violated.

Theo is right out wrong and he knows it!

- * 1. Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer,
- * without modification.


Just one little snippet that clearly shows that the removal of the BSD license notes was a violation of the BSD license terms.

As for changing from dual licensed to making it GPLv2 only for future changes, I think it's fairly immoral considering both licenses are supposed to be open source licenses.

But only if you choose to use it under the terms of the BSD licence; in this case, the Linux dev chose to use it under the terms of the GPL licence instead. Removing the BSD licence text is perfectly legal under the GPL.

If you're saying that you have to adhere to *both* licences, then suddenly the BSD bit becomes almost worthless, because you now can't link to proprietary software, you can't release binary-only versions, etc etc etc.



Immoral, perhaps. But not illegal.

Isn't the BSD license a superset of GPL v2?

If only.

Isn't the BSD license a superset of GPL v2?

It's the other way around.

One is an Open Source license, the other is a Free Software license.

No, both are open source, only gpl is 'free software'

Both the BSD license and the GPL are both open source and free software licenses. What makes you think they are not?

Well, if we're going to get into it, even the FSF themselves refer to the (non-ad-clause) BSD licence as a "simple, permissive non-copyleft free software license"[1]...


[1] http://www.fsf.org/licensing/licenses/

Sorry , my mistake, but anyways I was just wanting to note that the parent was wrong to mark only one open source and the other only free software license.

Wrong, the OSI considers them both open source licenses.

http://opensource.org/licenses/alphabetical

While it is true you can call the gpl "a free software license". It is just as much open source as the BSD only more restrictive.

I think you're getting confused by thinking both licenses apply simultaneously. That's impossible, for obvious reasons. Perhaps it would help if you thought about it as two different packages, with identical source code in them, but different licenses: one BSD, one GPL.

Removal of BSD licensing from the GPL version is perfectly legal and logical, since it doesn't make sense anyway. The same things goes for the other way too.

It's obviously a problem if different people want to work on the same code but want to license it under uncompatible licenses. It can be overcome if they work together and come to an agreement.

But if they're going to get pissed at some point by the "other guys" doing stuff that's specific and perfectly normal under the "other" license, then I don't see the point in keeping it up. Releasing stuff with dual licensing will only create confusion.

Edited 2007-08-29 14:32

DOH!

That doesn't matter since the license for the files clearly states that you can choose EITHER! It's okay to remove the BSD-license if you keep the GPL-license, and it is okay to remove the GPL-license if you keep the BSD-license.

It is so simple it beats me why you can't understand something this simple.

http://undeadly.org/cgi?action=article&sid=20070901041657

Still want to argue your invalid point?

Oops... Sam Leffner, the copyright holder, agrees with the Linux-devs. Oops.

Wrong. Anybody can choose which license to remove as long as the other license is kept. The license clearly allows for this. The copyright holder has given everybody the right to choose between either BSD or GPL.

It is just Theo being an ass this time - but that's no surprise. It comes along with being strongminded.

The problem is the three other files not dual-licensed. But people tend to forget them since they are not exactly under the BSD.

Thanks for clearing that up. I wasn't aware that it worked that way.


Same old same old then. The day I hear of a delicate situation involving Theo where he is not being an ass is the day I check my ass for frost bite.


Correct me if I'm wrong but I read something about that in one of the posts in the original article. So it seems that it ain't really about GPL/BSD, but about some of the code being released under a third license, then relicensed under the GPL even though they were not dual licensed in the first place.

Wow, talk about license hell.

So true.

The license for the three files are the ISC-license btw. Extremely simple so the whole issue is weird. But great... I got headache from this... damn licenses!

Theo does get allot of peoples backs up. Thing is, both situations seem to be genuine mistakes, I don't see the point in starting a grudge match about it.

Come to think of it, starting a grudge match over Theo's inability to be civil has kinda been done to death ;-)

There was quite a difference there. The BSD developer used a very insubstantial part of the code as a placeholder, while developing within the CVS tree. Was this a mistake? Obviously. Did the linux johnnies goes entirely overboard? You bet they did.

In this case, the developer took the BSD code, relicensed it wholesale with the GPL, and even replaced attribution with his own name. It's unclear if this guy even wrote one line of code before doing this. Was this a mistake? Yes. Does it seem it was intentional? Absolutely. Did the OpenBSD developers publicly call this guy a thief? Nope. It was only discussed within the OBSD lists and happened to get posted on undeadly.org (read only by OpenBSD users/devs and the occasional troll from slashdot).

Incorrect. He NEVER replaced attribution with his own name. He removed the BSD-license from two dual-licensed files. Perfectly correct of him. The dual-license allows for this.

The problem is the three files not under the GPL and not under the BSD. There is no stealing of BSD-code because the BSD-code was also under GPL and he could freely choose which license to use. The three files under the MIT-like non-BSD license did however have their conditions removed which is a copyright violation. But the attributions were NOT removed.

Stop lying.

You get to chose license for your derived work, you dont get to remove the BSD license from the file.



Indeed they should even though he can be harsh.

Edited 2007-08-29 13:24

Yes, you do.

Take the dual-licensed file. Choose to use it under the GPL. Edit the source to remove the BSD licence text (perfectly legal under the GPL). Voila.

Well, wrong sort of. You have the right to use, modify and distribute the file under the license you choose.

You don't have to make modifications to the sources only.

Besides, removing one of the licenses create a derivative anyway, so your argument (which is copied directly from Theo) is moot

Theo is wrong on this one, and won't admit it. That's the negative part of his strong mind, which I usually admire.

Well, since this (even though it has been solved) has already been blown way out of proportion by sites likes this and slashdot (and interestingly enough not by Theo himself) lets just say I think you're the one who's wrong and Theo is right and leave it at that.

Well, Alan Cox agrees with me, and so does an enlightened minority among the BSD-group. Theo's very creative interpretion of the Dual Licensing Scheme is enough to prove his wrong. If your and Theo's interpretion is correct (which it is not) then the Dual Licensing Scheme is incompatible with GPL since it adds an extra stipulation - apart from running counter the optional nature of the license.

Right now I've got other things to do, but I'll fry you and Theo later.

The dual licensed part has not been fixed btw. - it's only the ISC-licensed part that has been solved.

It doesn't allow you to claim you wrote it yourself.

OMG, but that's so restrictive! Isn't it supposed to be so "open" and "free"?

I'm joking, but I just used it as an example of the same reasoning that some people apply in reverse to GPL.

It's probably because those people don't understand that freedom needs to be actively defended and enforced. That means certain restrictions. There's no such thing as "pure freedom". If something was completely free, it would imply allowing someone to come along and say "it's not free anymore". Which is silly and defeats the purpose. So at the very least you have to make restrictions against things that would make the stuff non-free.

GPL wants stuff to be free and to keep being free. The GPL license specifies terms that work towards that goal.

BSD wants stuff to be free and to give credit where it's due. The BSD license specifies terms that work towards that goal.

I think you can see what the difference is. BSD doesn't care if stuff that was free is made non-free. That is why some people prefer GPL. End of story.

Edited 2007-08-29 14:43

Nobody has done that!

The copyright notices have not been removed. The conditions have been removed. This is legal for the two dual-licensed files, but a copyright violation for the three not dual-licensed files.

You haven't read the diffs I can see. RTFA!

This is about copyright, not about license.

And the copyrights haven't been touched. Only the conditions. Which means this is all about licenses and not about copyright

Yes you have a good chance. Most distros support it. I know ubuntu,mandriva support my D-link DWL-G520 PCI wiress-card right out of the box.

Have a look at the list of manufacturers who use the atheros chipset:

http://atheros.rapla.net/

A good OSS project around atheros based cards is MadWifi:

http://madwifi.org/

With atheros chipset based wifi cards you have a fairly good chance the card will be detected on *BSD,linux most of the times.

Edited 2007-08-29 14:13

Question: Does this mean that it is now possible to use Atheros based wlan devices right out of the box on GNU/Linux?

Yes. With the new mac80211 stack we will have a lot more drivers available in kernel. Although atheros support has been quite good on Linux for years now with the madwifi project, I would love to have in kernel drivers for atheros based wireless cards.

The BSD license was stripped from two files - both situations legally correct, since both files were dual-licensed.

The problem is the three other files which are not under the GPL, nor under the BSD, but under a MIT-like license. That's the real problem and not the perfectly legal removal of the BSD-license from the two dual-licensed files.