It is a big shame that so many people here seem to be eager to dismiss the book. Perhaps you forget to realize the fact that the MS Press edition was a thorougly updated and much reduced revision of the original. The original CL/DM was published in 1974, and at that time it truly must have been revolutionary! Its original format and style was inspired very much by the "Whole Earth Catalog" by Stewart Brand. The new edition in 1987 was not intended as an expression of new material or ideas (although it did mention Macs and PCs; also the creative - but by todays standards amateurish - use of MacDraw seems to permeate it) but to make the by then widely acknowledged work, which had been part of the foundation of hypertext studies for more than a decade, accessible to a wider audience, as the original edition was simply too rare. (The original was self-published, and I suspect the number of copies wasn't very large.)
I frankly no longer remember if I bought the MS Press Edition (1987) before or after I attended a course on Hypertext given at the CS department of the University of Aarhus in 1991 or so. (A course that was give by Randall Trigg, BTW. He wrote the first PhD thesis on hypertext, and participated in developing NoteCards at Xerox PARC.) Probably before. Revelation may be too strong a word, but reading it definitely came close! Absolutely inspiring!
In 1987 hypertext had just become so acceptable in academic circles that the first conference was held, at the University of Chapel Hill, North Carolina. Ted Nelson was here recognized as the father of hypertext. I had the great fortune to be able to attend Ted Nelson's keynote at the CMF2006 conference in Aarhus last year, and though I don't agree with all of his ideas (not that anyone would care), I wouldn't hesitate to call him a brilliant thinker, maybe even a genius, and he's definitely not finished with contributing to the field. Many of the insights from that 1987 conference, as reflected in its proceedings, are still as valid as ever, and anyone who has read them will know that an abundance of criticism - but also a little praise - can be directed at a certain Tim Berners-Lee.
Considering CL/DM "stupid" or "overhyped" is far beyond absurd! And anyone calling Nelson a nutcase obviously has nothing to support such a description, and only demonstrates a total and pathetic lack of knowledge of the field of hypertext and the importance of Nelson's work. I suppose such a person would think of stuff like "XML", "AJAX" and "Web 2.0" as "cool", which only goes to show what a sad world we're living in.
No, if you make a binary library, you are free (as in freedom) to distribute it (anyway you like, even for lots of money) you don't have to distribute it for free (as in beer). So much is clear.
Further, either a) or b) applies: a) as this is a derived work, you can distribute it under your own terms, provided you include the license in the required places (docs). b) as a derived work of BSD code, you can not add your own terms, and whomever you sold it to may then redistribute it, even for free (as in beer.)
I am quite sure that a) applies, as that's precisely what allows companies like Apple and Microsoft to use BSD code. I am sure they can afford lawyers that have verified the validity of this argument.
However, for the source, it is clear that redistribution of modified or unmodified source is allowed if and only if it is done with the BSD license. This means that all the distributed source is redistributable under the same terms. As you are not required to offer it free of charge, you can try to profit from your redistribution, but the recipient is not bound by any of your terms, and may in turn redistribute for free (as in beer.) You could just as well distribute it for free (as in beer) yourself.
I think you could fairly say that whereas the GPL is a virus, in that any project using it gets infected, the BSDL is simply resistant - you can't infect it with other terms, but you can use it symbiotically under other terms. Mitochondrial DNA, anyone?
Wikipedia has a paragraph under the Public Domain entry which says, quoting Nimmer: "It is axiomatic that material in the public domain is not protected by copyright, even when incorporated into a copyrighted work...."
This speaks volumes to me. It means that even when you create a derived work using a PD work, the PD work may be extracted as being PD. Actually this comes as a bit of a surprise to me, as I know that Project Gutenberg has spent great effort to reproduce copyright-expired texts from books that were old enough to be out of copyright themselves. If the Nimmer quote above holds, then you could simply type the text in from any readily available print.
Anyway, my guess is that if this applies to PD work, then it would also apply to BSD work similarly incorporated into a work of different license. That's the reason I call it symbiotic: you can use it, even profit from it through your own license terms, you just can't restrict any part of it that isn't yours (including modifications to such parts even if they are technically yours).
Pardon me for mixing into this discussion, but I'd like to thank you for those links - a very interesting and noble cause in any case, although I doubt I would agree with all his (her?) ideas. Designification (?) of symbols is something that I think will become very important in the future, and I'd like to explain why I think so.
Symbols are objects that are used to represent ideas of any kind. Although there are theoretically an infinity of objects that can be brought to such use, some objects are more practical, and some contexts severely limit the number of usable objects, even if the same object can be used in different subcontexts to signify different things. Take DNS for example. So for practical reasons, we do need to reuse objects for new symbols. However, the world of text (culture) becoming more and more complex, means there is a growing need for more and more symbols. Also, as symbols tend to last much longer, the binding between the object and the signified gets much stronger, WWII was recorded and imaged so much, both at the time, and afterwards in fictional and documentary form, that its symbols are almost totally frozen in the collective mind.
A Danish bar owner wrote a letter to a newspaper last week, in which he questioned the validity of recent rules on smoking in bars and restaurants, asking "how can I explain to my customers that they are not allowed to smoke in restaurants that will readily accept a Danish 500 kroner banknote with a picture of Niels Bohr smoking his pipe." The answer of course is that the two things need not have anything whatsoever to do with each other, it's all in the mind. Why make it into a problem, it doesn't have to be? Niels Bohr smoked pipe, so the image is perfectly representative. And, just like Magritte's "This is not a pipe", the image of Niels Bohr does not cause lung cancer.
The swastika didn't kill any jews. That's a fundamental fact that many people seem to have great difficulty to grasp. Don't confuse the symbol with the thing it represents; ideas in people's minds cause killing and suffering, not the symbols we use to represent them. On the contrary, we NEED the symbols to DISCUSS such harmful ideas openly, without prejudice. Likewise with bomb recipes, cartoons of Muhammed, porn, etc.
It is strange and ironic that even religions originally recognized the need for distinguishing between the symbol and the symbolized, for example the commandments about making idols or wrongful use of the name of God. Yet religious objects of so many religions are considered sacred by themselves: holy books, names, crosses, flags, places, temples, plain tap water (!), alcoholic beverages and dry tasteless biscuits. The absurdity is impressive. The jews actually seem to have handled this matter best (at least some of them), whereas both Christians and Moslems have for example confused their human founder with the deity they claim to believe in: Christians by fusing Jesus with God and whatever the Holy Spirit may be, into the Holy Trinity, and Islam by effectively putting God aside and worshipping Muhammad (through his alleged writings) instead, although they will profusely deny that fact.
You can't ban ideas, and thinking you can do so indirectly by banning symbols is stupidity of the worst kind.
Alas, even in the socalled "scientific" and "skeptical" circles, there is a tendency to raise certain ideas to a status of being unquestionable, and identify these ideas with certain symbols. Fortunately this is rare, and I think it's mostly confined to theories about the "beginning" of the universe and other fundamental ideas, and perhaps just with some people, who do not fully understand those theories, and therefore are more prone to abuse them by overinterpretation. Advances in neuroscience may be helpful here. I have no problem with the possibility that there are limits to what we can understand about the world we live in, and I can accept such limits without resorting to the creation of an imaginary omniscient demiurge or an infinity of parallel
Incredible, but nonetheless true, I see, and stand corrected. A search gives plenty of hits that indicate that the Lisa was actually capable of running Microsoft (!) Xenix. However, I doubt that it ran Xenix with any form of Apple-like GUI.
I was a hardcore Mac fanboi until the coming of OS X. I never understood why the super-intuitive direct-manipulation interface of Finder, with its 1-1 correspondence between Folder windows and Folders (ie: one specific folder only ever showed up as one window), was abandoned. As I tend to be spatially organized, I need things to always pop up in the same locations. The Finder did that. Windows and OS X "file browser" never did.
The Mac had one thing that Lisa should have had (only more of them): square pixels. At least that's what I understand from reading about it, I only saw an actual Lisa once, at a computer convention in Copenhagen in 1984.
I am afraid you are mixing up a few things. At least I never heard of Unix for Lisa. However, A/UX came out in 1987-88 or so. I suppose that's what you were thinking of.
For GUI consistency, I don't think there's anything in current use (even OS X) that holds a candle to system software 7.x. To this day, I haven't found a newsreader I like half as much as NewsWatcher, so that's what I continue to use - now under BasiliskII emulation running on NetBSD. Likewise with Word 5.1a, which beats any later version of Word - as well as OOo - easily IMO. (It was the last Mac Word that didn't share codebase with Windows, and therefore had at least a somewhat Mac-like UI, if not completely adhering to HIG.)
A little search tells me that MacTCP (the TCP/IP stack for Apple Macintosh) goes back at least to 1987-1988. (As does A/UX, Apple's Unix, BTW.) Eudora, according to the Wikipedia article on Steve Dorner, was developed in 1988 (and of course Mac-only), so this precedes even Win3.1 by a couple of years. For about a half decade, Macintosh had plenty of free internet tools (FTP clients, mail client, NNTP news clients, WAIStation, TurboGopher, games) whereas the DOS/Windows platform didn't even have a standard IP-stack. In the beginning, the available tools for Windows were but faint imitations of their Mac counterparts.
The Internet Revolution was already over twice (first for servers/Unix, second for desktop systems/workstations/Macs) by the time Microsoft catched up. Heck, in 1993, there still was a belief that TCP/IP and Internet would be superseded by OSI protocols like CLNP/TP4, and Apple at that time release AOCE, which integrated an API that allowed any application to gain protocol-neutral (SMTP/POP/IMAP, X.400, proprietary mail protocols such as QuickMail, AppleLink etc) mail service capabilities. This was fully integrated in the OS and the GUI.
Apple was way ahead of Microsoft when it came to understanding, utilizing, and innovating network use in general and the Internet in particular, on the desktop.
In fact, the WAIS project, under the leadership of Brewster Kahle (now famous for Internet Archive), was a joint research project between Apple, Thinking Machines Inc, Dow Jones and KPMG Peat Marwick. It's purpose was to create a system for Wide Area Information Servers, a distributed, loosely linked network of dissimilar and greatly varied information services, which were made uniformly searchable across the net through the Z39.50 protocol. WAIS, at least in spirit (and together with its Gopher-based counterpart, Veronica), is a direct ancestor of search services such as Google, and could easily have developed into something that would perhaps have been far better than WWW. Funny thing is, I *never* saw any mention of Microsoft anywhere in those wonderful days. I think IBM was mostly absent as well.
If only Apple had focused on the Internet and enhanced it with the elegance of AppleTalk instead of wasting time on overly generalized network APIs, it would have been a major factor on the Internet.
If only Apple had pursued A/UX (preferably with more of BSD underneath) instead of repeatedly getting into failed OS projects (Pink, Taligent, Copland) it would have blown Win95 away in 1996, with a system that would be very much like todays OS X. Probably better.
Apple failing so many times is what gave Microsoft the chance to become big. Also a little luck, and of course a lot of cheating and other nasty behaviour. One thing that definitely hasn't played any major role in MS' success, is merit. Therefore, MS claiming to have laid a foundation for anything, other than BG's money bin, is simply ridiculous.
Sorry for this late a reply, I hope you still see it, and if you don't, well, it's not that important. At first I thought it was unnecessary to reply, because all that needs to be said, has been. After a few days of consideration, I think it may be relevant to summarize after all. Not because I want to claim "last post", but because I think we have reached the very core of the issue.
It says it must retain *the text*, not the rights.
I agree that that is what it says. The question is whether retaining the text means retaining the meaning of the text, which means retaining the rights also. More specific, does it make any sense to understand it to mean otherwise, that the text is retained without its original meaning? It would seem that the common interpretation has been yours, but that does not mean that it was correct, and now we have Theo saying he thinks it isn't. I am saying that the wording of the license - to me at least - makes more sense when understood as that the text of the license must be retained with its meaning preserved, applying to all the source that is redistributed. If this is not the meaningful interpretation, it definitely is a meaningful interpretation.
Now, what consequence does this requirement have?
Note that taking your viewpoint literally (since it has to retain a copyright notice) would mean that you'd have to gift your copyright to whoever the original author was, because it says (c) Original Author right at the top of the file.
You weren't explicit about the right carrying onto derived works.
I was explicit. Redistribution of works derived from works licensed under B are required to be licensed with B. That's almost what the BSD license says, although I *was* more explicit about the license actually applying to all of the derived work.
As for the pointlessness of redistributing unmodified source, I would like to retract that sentence. Although you would simply have to retrieve the work from a distribution site authorised by the author outside the terms of the license, that way it wouldn't really matter much. If I am not mistaken however, Dan Bernstein avoids the whole issue of rights to derived works precisely by only allowing unmodified distribution, forcing any derived works to be distributed as patches. So you are right, it is not completely pointless. In fact, Bernstein posits that modification and use are implicit rights attained by legally being in possession of the work (if I read his writings correctly) and therefore the only right the license has to deal with is the license to redistribute, because the rights to use and modify a lawful copy cannot be restricted based on copyright law.
No reply necessary, as I think we have reached a good understanding of the points. Again, thanks for it all.
Now we are really getting somewhere, I'd like to thank you very much for your patience with me.
The BSD license requires few limitations - the BSD license has to stay on the original files, and in some version you have to provide attribution. However, there's no limitation on the license *of the derived work*.
The BSD license doesn't use words like "derived work" and "license". It says "Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer." These are the parts that constitute the license, yet they are not called that.
So you are saying that "retain" can't possibly be understood as to apply to the redistributed (supposedly modified and thus "derived") source, and thus imply a requirement for the derived work to use the same license?
Or are you saying that (re)distribution of a derived work is not the same as redistributing unmodified source, and the requirement only applies to unmodified source?
But it does say "Redistributions of source code must retain (the license)". To me that means what goes out, not what came in. And if it doesn't include modified source, then the license only grants redistribution of unmodified source, quite pointless. I almost fear I still don't understand it. After all, what would be the point of retaining the license, unless it is meant to apply to the derived work?
In your "hypothetical" example, the two licenses are identical except that G provides one less right than B does. Therefore, a combined license is effectively equivalent to G.
But the terms of B say that you have the additional right, and that redistributed work should have license B (with that right.) How can that be compatible with G?
What would be the consequences of a BSD license variant that had just the one additional restriction, that derived works distributed as source are limited to the same license? (IOW a license that works the way I thought the BSD license might.)
I am sorry, but I still don't get it. How can the GPL get away with requiring derived works to be GPL'ed, then?
As for the double license, I am also still confused as to what this means.
I know you are saying that the following is actually not what the BSD license says, although it's what I understand from it. Please disregard that disagreement for a moment, and let's look at two paraphrased, quite similar licenses, B and G.
B: This license grants you rights to use, modify, redistribute source alone or binaries alone or both. If you redistribute source, including derived works, you must do so with this license B.
G: This license grants you rights to use, modify, redistribute source alone or both source and binaries. If you redistribute source, including derived works, you must do so with this license G.
If a new combined license is formed from B and G, then what rights does it grant? And under which conditions? Can a combined license be formed at all?
The BSD license doesn't say this. If it did, it'd be (essentially) the GPL. The BSD license allows to create derived works, and to distribute those derived works under any license you want, so long as the BSD notice and copyright stay on the source files (and attribution clause, for those versions of the BSD license). This is why you can use BSD code in closed source projects.
We agree that the license requires retention of the copyright notice, the conditional grant of rights to use, modify and redistribute, and the disclaimer of warranty (together making up the BSD license.) What would be the point of this requirement, if not the implication that the same license applies to the modified and/or redistributed file? Then how can you say it doesn't say that it doesn't
So the BSD license permits redistribution under another license so long as the BSD notice remains intact.
This is selfcontradicting. You are saying it permits redistribution under another license so long as it is redistributed under the BSD license. As long as the BSD notice is intact, the file IS licensed under the BSD license!
My "theory" is that if you try to license your work with a dual license, and forget to grant the right to relicense under either of the licenses, IOW delete the other, then your licensee has no choice but to redistribute it with the entire dual license.
Whether the copyright owner actually intended to allow such relicensing doesn't matter, if he didn't explicitly allow it, my bet is it's just not allowed. If he didn't do what he intended to do - and maybe even thought he did - then that's his own fault. It's the license, not the copyright holder's intention, that grants the licensee the rights.
You are right that this interpretation makes BSD very close to GPL. It's just not quite as infective, and it doesn't require source to be distributed.
If by "this debate", you mean the exchange you have been so kind to pursue with me, then I certainly owe you an explanation. I chose the subject because I think the word "relicensing" not only matters, but makes the entire difference. Unless we know what relicensing is, and whether it is a right that has been granted by the license, we don't really know anything.
If my interpretation is right - and it very well may not be, but this is what is so unclear - then the BSD license does not grant a right to relicense, it merely grants you a right to redistribute to a third party the licensed work with the same rights and conditions the copyright owner has granted you, IOW the BSD license. So you can't relicense, you can't change the license, and moreover, you agree to redistribute any modifications with the same license. Therefore there simply is no need to have an explicit clause forbidding relicensing under GPL, because relicensing never could be an issue when using the BSD license.
If I am right, then the BSD license is "freeer" than anyone seems to have imagined. You can do modify the code as you please, you mustn't even distribute anything. Only if you do, then you have to use the BSD license if you are redistributing (modified) BSD code. You can freely distribute it with other stuff using any license, including GPL or proprietary licenses. It doesn't infect anything except direct modifications of BSD source, and as there are no restrictions on use, as long as you keep the license intact, it goes along well with anything. The only thing you can't do is make it less free! (Except, of course, if you are the copyright owner.)
If the BSD license didn't allow relicensing, it would be worthless because it would effectively be a "look but don't touch" license, in that you could see the source code but would be prohibited from modifying and redistributing it. And if it only allowed for relicensing under the same terms of the BSD license, well, it'd be the GPL.
I am tempted to repeat the subject in-line. The way I understand the word "relicensing", it means licensee distributing the licensed material under a different license. This is a right which is apart from any right to use, modify or (re)distribute (subjecting the recipient to the same license as the one granting the distributor/licensee the right to distribute.)
And the BSD license doesn't allow relicensing. Do we agree that any right not explicitly granted in the license by the copyright owner to the licensee is not granted to the licensee? The 3-clause BSD license grants this, and only this:
* Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * * Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * Neither the name of the <organization> nor the * names of its contributors may be used to endorse or promote products * derived from this software without specific prior written permission.
A BSD license that adds restrictions on what you can do with derivative works is essentially the same as the GPL in that regard, because it forces you to license source code--if you release it--under the same license. Otherwise, you could license your derivative work under the less restrictive "traditional" BSD license, then use that version in a GPL project. So, if it is to be enforceable, it is no longer a true BSD license. In fact, it might as well be the GPL.
Almost exactly. My point is that this is in fact how it is now - BSD does restrict derivative works. The only two significant differences between GPL and BSD are that BSD does not require 1) redistribution of source in case of redistribution of binaries and 2) that any code statically linked with licensed code also be same licensed.
Actually, it is unclear whether the second clause could be construed as infecting the entire binaries with the BSD license.
I just found out that back in January, Brendan Scott, an Australian OSS lawyer, published a very interesting paper about issues with the BSD license. As I am not a lawyer myself, nor in any way trained in understanding legalese, I am not sure i get all the points of the paper right, but I think it confirms at least some of what I have been saying.
If you want to make the BSD license viral, then do it. "You need not distribute source. But if you do, it must be licensed under the BSD license".
You mean:
* Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. [...]
BSD is viral in that sense already. It's just a benign virus, compared to GPL, which also infects other code, and demands redistribution, instead of just permitting it.
The SSL license is interesting. They are explicitly stating what I believe is already implicit in the BSD license. This hinges on whether any right not explicitly granted in the license is transferred, especially the right to relicense under different terms.
muuh-gnu, you are absolutely right about the BSD license doing exactly the same as the GPL. Except two things: 1) there is no obligation to also redistribute source if you redistribute in binary form. 2) Using BSD code with your own code doesn't infect your code with the BSD license.
What he is saying is not relevant. What matters is the actual license he released his work under. He may believe that it means what he says now, but unless he in fact explicitly granted the licensee the right to relicense the work under either license, all a licensee can do is to redistribute the work under the same (dual) license.
I am deeply sorry, I had absolutely no intention to come across as rude, nor do I think you are an idiot at all.
You are, however, unclear about who you think can relicense the code. As a licensee of BSD source code, I can redistribute it according to the license, but not being the copyright owner, I wouldn't dare to relicense it under a different license.
As I see it, modified BSD licensed source code is still BSD licensed source code. You cannot remove the copyright notice, the conditions or the disclaimer without violating the license. You could add a new copyright notice for the parts you modified or added, I suppose. The BSD license grants you a right to use, modify and redistribute, but not to relicense. "Redistribution and use in source and binary forms, with or without modification" is all that the BSD license grants the licensee.
I don't know if you do it on purpose, but your paragraph starting "My point is..." is really confusing. Which modified BSD license do you mean? And what do you consider the "original" BSD license? Do you mean the 4-condition with the advertising clause, or simply the particular BSD license a work was originally licensed under? What do you mean by "unenforcable" and "self-defeating"?
Most importantly: what right do you believe a licensee who is not the copyright owner has to relicense a BSD licensed work? In what way has the licensee been granted that right by the original copyright owner?
I really feel like an idiot myself, because there seems to be a broad concensus that indeed such relicensing is allowed, but I just can't get into my head how this is granted in the BSD license. Please have patience with me and be very clear and explicit in your explanations. Presume that I'm just not very smart.
How can the BSD license be superseded? OK, sure, the original author can change the license of course, but as there is no termination clause, the already distributed copies are still fully licensed and thus remain redistributable under the BSD license. Right?
The BSD license says that a modified source file must be redistributed *as BSD*. So it is simply not within the rights obtained through the license to apply GPL to the changed file. NO CAN DO! Please note that I am talking here about the generic case of software initially released as BSD, addressing the misconception that you can "close", "GPL" or "render unfree" BSD licensed code. Only the original authors can do so, and already distributed copies remain distributable under the original conditions.
Regarding what has happened in the present case, we are not in great disagreement.
I will admit that at first I thought that the original author, like seemingly so many many people, did not understand the respective implications of the BSD and GPL licenses, and "dual licensed" them by simply saying "use either one". I agree that it is his right to change the license. I do not agree that anyone but the author, given just a dual license with a choice of distribution under BSD or GPL, could modify the license, to remove the choice given (consciously or not) by the original author. If the original author's license conditions are simply "BSD or GPL", then this is a composite license which must be retained on any copy, including the choice. I doubt the GPL is different than BSD when it comes to terms like redistribute and relicense: a licensee can redistribute the work under the exact same terms as the original, but not modify the license. Only the copyright owner has the rights to define a license. Of course, the copyright owner could grant the licensee the right to redefine the license, so that a licensee could redistribute with a different license - IOW relicense it - but I hope that a permission to relicense like that would have to be explicit in the license. In fact this seems to be what the Artistic License v.2.0 is doing, for example.
Therefore, I agree that the change of license on ath5k_base.c is absolutely fine. I am not sure whether its original dual license allows relicensing by anyone but the copyright holder, but as it was changed by the copyright holder this issue is moot, unless a licensee is going to redistribute the version before this change.
The change to the file ath5k.h, was not OK according to the license. Which I understand has already been acknowledged and resolved.
I will readily admit that IANAL, that my interpretation is based on simple common sense, and therefore of course I could be completely wrong. But I haven't yet seen anything to convince me of that.
I'm a big fan of the BSD and BSD-style licenses. [...] Does the BSD license allow people to make extensions and GPL the base code plus those extensions? Absolutely. Do BSD-style developers, then, have a right to be miffed if this is what happens? It's a hard question.
*SIGH!*
It's not a question at all: (from the wikipedia article on BSD licenses)
[...] * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. [...]
This effectively says that IF you redistribute source, you MUST distribute as BSD. Plain and simple. You are not - unlike with GPL - OBLIGED to redistribute source, but IF YOU DO YOU MUST LEAVE THE BSD LICENSE INTACT.
Why is everyone - even selfproclaimed BSD fans - having such a hard time understanding this?
OK; playing Devil's advocate (or would that be God's advocate in this case?). In principle you could wrap the entire notice like this: THE FOLLOWING RESTRICTIONS DOWN TO "end-of-disregard" DO NOT APPLY: [BSD license] end-of-disregard
But that is definitely not what the license intends, or the "spirit" of the license. It depends on how literal you interpret the word "retain". Does it mean "textually include" or "retain as the rightful conditions for licensed use of this copyrighted work"?
What on earth is this weird concept of "relicensing" all of a sudden? The right to license a work is precisely the right of the copyright holder. The BSD license doesn't talk about transferring any right to relicense, only the right to modify and redistribute, and I am quite sure the GPL is not much different in this regard.
It seems most people don't have a problem understanding that BSD source code can be used for closed source products and that's just swell.
People also understand that GPL source, when redistributed, must be GPL.
What so many people inexplicably seem to fail to understand is, that BSD source, when redistributed as source, has the same "stickyness" as GPL code. If you redistribute BSD source code - an action which, unlike GPL source code, isn't mandatory - you MUST redistribute it with BSD license. You can't redistribute BSD licensed source under GPL, although you CAN redistribute it freely - even in your GPL project - if you want to. Actually, slapping on the GPL (or any other license) is just about the ONLY thing you can't do with it. So GPL-guys cannot redistribute modified BSD code under GPL as it seems they would so much like to do. What they might be able to do is to put GPL on their modifications, which would not be a very friendly thing to do, but at least legal.
It simply doesn't make sense to equate "illegally redistributing with a different license" with "not redistributing source at all" under the label "close off the source".
I don't see how source code once licensed under BSD can somehow be made "unfree". Maybe we have to look a bit closer at what's going on.
Let's say we have a developer T, who develops his code on the machine beastie, and wants to release all his source code under the BSD license.
Further we have a developer L, who develops his code on the machine tux, and releases all his source code under the GPL. (Let's ignore reassigning copyright to FSF for now, in some places you can't even do that.)
Last, we have a company M (all developers here do work for hire, so the copyright belongs to M), developing code on the machine mtdoom, not releasing any source code at all, but distributing binaries.
T develops foob.c, L develops foog.c, and M develops foom.c.
case 1) If M wants to use foob.c in his project, he is allowed to do so, provided he acknowledges this use. He is not required to distribute the source, and his own code is not infected with the BSD license. Fine, a copy of foob.c is now found on mtdoom. A discreet notice of the use of foob.c goes into the documentation.
case 2) If M wants to use foog.c in his project, he is allowed to do so, provided he also distributes a copy of the file. Further, his own code must probably also be released with the GPL license. M doesn't want that because of his modus operandi, so effectively he cannot use foog.c. Let's just hope foog.c is not to be found on mtdoom
case 3,4) If L or T want to use foom.c - tough luck. They need to get M to distribute the file under some license they can use. No other way can they legally get a copy of foom.c onto their machines and into their project.
case 5) If L wants to use foob.c in his project, he is allowed to do so. He can redistribute the file freely. No reason why foob.c shouldn't be copied to tux. L can even distribute foob.c as is, as part of his GPLed project. foob.c is now on tux.
case 6) If T wants to use foog.c, he is allowed to do so. But like case 2, depending on various things, he may have to release the rest of his code under GPL as well. T, just like M, doesn't want that, and is thus effectively unable to use foog.c. Of course, a copy still might find it's way onto beastie, but it can not be linked with any of T's own code. T can merely redistribute this code under GPL just like any other friend of L. (This actually applies to case 2 as well.)
So effectively, the only file that could be freely copied everywhere is foob.c. Now what can actually be done to these copies of foob.c?
As mtdoom is a proprietary machine, we can't tell for sure. But for M's own sake, let's hope the license is retained, lest they forget to keep their minuscule obligation to T. M is allowed to modify it of course.
L however, likes to modify foob.c and redistribute the modified foob.c. Again, this is granted. However the license says * * Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer.
So L has to leave at least the license text intact if he modifies the file.
The questions that remain are: can L "infect" foob.c with GPL? NO! foob.c on beastie cannot be infected and will remain accessible and distributable. However, I don't know if L could slap a GPL onto his copy of foob.c given he did sufficient modification? And would that GPL "override" the BSD license already attached? It would seem certain that it could not _replace_ the BSD license, as that's what allows the use and modifications in the first place!
Particularly relevant for this discussion: let's imagine a hybrid developer H, releasing fooh.c on his machine hoohoo, who for some reason wants to leave the choice of BSD license or GPL up to the licensee. Does it even make any sense, any difference at all, to have a dual license? What would the point be? What does that choice mean? Does it really mean that L can make a copy of fooh.c on tux, and just REMOVE the BSD license and redistribute it under GPL only? Without
Anyway, the geek equivalent: If you have a sequence of bits, and don't know the decoding algorithm (if one exists), how can you tell whether the sequence is random, or actually interesting information that just happens to be compressed/encrypted? Maybe you could invent some new decoding algorithms, and see if anything interesting comes out -- but then : is the interesting stuff a result of applying your algorithm to this specific random data, or the information originally encoded in the data? You can only know the right answer if you cheat: cBwfCYG lr0 bGCkx TYJScPAtB yxF8jhtr0 XYYSjL iz9so CreZfPX pjzpUW9Cj1vP AoKEhc6hAF wA3 J2Y kDi5q 11KzWsm 9Mp5ZQvGo.
Somehow my mind wanders off to think about monkeys and Shakespeare...
As for me, I don't care much about infinity: the finite world is more than big enough. Infinity was invented to understand in finite small terms the very large. Mistaking it for something real is merely foolish abuse - or religion, which is just another name for "foolish abuse".
The number of jokes that can be told in three lines I need not hear all
Instead of wasting effort on space exploration, we should waste it on reducing suffering. Just as futile, only much nicer.
-Lasse
Re:optical mice have their own issues.
on
Mouse or Trackball?
·
· Score: 4, Funny
OK, your method ensures a clean desk, which is neat, but perhaps you should try lifting the mouse and reposition it at the opposite edge of the mousepad when you reach the edge and the pointer isn't "there yet". It won't hurt anything, if that's what worries you. I am curious: what would you do if you added another monitor? Upgrade to an A2 sheet?
Slashdot Mirror
It is a big shame that so many people here seem to be eager to dismiss the book. Perhaps you forget to realize the fact that the MS Press edition was a thorougly updated and much reduced revision of the original. The original CL/DM was published in 1974, and at that time it truly must have been revolutionary! Its original format and style was inspired very much by the "Whole Earth Catalog" by Stewart Brand. The new edition in 1987 was not intended as an expression of new material or ideas (although it did mention Macs and PCs; also the creative - but by todays standards amateurish - use of MacDraw seems to permeate it) but to make the by then widely acknowledged work, which had been part of the foundation of hypertext studies for more than a decade, accessible to a wider audience, as the original edition was simply too rare. (The original was self-published, and I suspect the number of copies wasn't very large.)
I frankly no longer remember if I bought the MS Press Edition (1987) before or after I attended a course on Hypertext given at the CS department of the University of Aarhus in 1991 or so. (A course that was give by Randall Trigg, BTW. He wrote the first PhD thesis on hypertext, and participated in developing NoteCards at Xerox PARC.) Probably before. Revelation may be too strong a word, but reading it definitely came close! Absolutely inspiring!
In 1987 hypertext had just become so acceptable in academic circles that the first conference was held, at the University of Chapel Hill, North Carolina. Ted Nelson was here recognized as the father of hypertext. I had the great fortune to be able to attend Ted Nelson's keynote at the CMF2006 conference in Aarhus last year, and though I don't agree with all of his ideas (not that anyone would care), I wouldn't hesitate to call him a brilliant thinker, maybe even a genius, and he's definitely not finished with contributing to the field. Many of the insights from that 1987 conference, as reflected in its proceedings, are still as valid as ever, and anyone who has read them will know that an abundance of criticism - but also a little praise - can be directed at a certain Tim Berners-Lee.
I just noticed that the Hypertext2007 conference took place last week in Manchester, UK. Ted Nelson was a keynote speaker, not surprisingly.
Considering CL/DM "stupid" or "overhyped" is far beyond absurd! And anyone calling Nelson a nutcase obviously has nothing to support such a description, and only demonstrates a total and pathetic lack of knowledge of the field of hypertext and the importance of Nelson's work. I suppose such a person would think of stuff like "XML", "AJAX" and "Web 2.0" as "cool", which only goes to show what a sad world we're living in.
-Lasse
No, if you make a binary library, you are free (as in freedom) to distribute it (anyway you like, even for lots of money) you don't have to distribute it for free (as in beer). So much is clear.
..."
Further, either a) or b) applies:
a) as this is a derived work, you can distribute it under your own terms, provided you include the license in the required places (docs).
b) as a derived work of BSD code, you can not add your own terms, and whomever you sold it to may then redistribute it, even for free (as in beer.)
I am quite sure that a) applies, as that's precisely what allows companies like Apple and Microsoft to use BSD code. I am sure they can afford lawyers that have verified the validity of this argument.
However, for the source, it is clear that redistribution of modified or unmodified source is allowed if and only if it is done with the BSD license. This means that all the distributed source is redistributable under the same terms. As you are not required to offer it free of charge, you can try to profit from your redistribution, but the recipient is not bound by any of your terms, and may in turn redistribute for free (as in beer.) You could just as well distribute it for free (as in beer) yourself.
I think you could fairly say that whereas the GPL is a virus, in that any project using it gets infected, the BSDL is simply resistant - you can't infect it with other terms, but you can use it symbiotically under other terms. Mitochondrial DNA, anyone?
Wikipedia has a paragraph under the Public Domain entry which says, quoting Nimmer:
"It is axiomatic that material in the public domain is not protected by copyright, even when incorporated into a copyrighted work.
This speaks volumes to me. It means that even when you create a derived work using a PD work, the PD work may be extracted as being PD. Actually this comes as a bit of a surprise to me, as I know that Project Gutenberg has spent great effort to reproduce copyright-expired texts from books that were old enough to be out of copyright themselves. If the Nimmer quote above holds, then you could simply type the text in from any readily available print.
Anyway, my guess is that if this applies to PD work, then it would also apply to BSD work similarly incorporated into a work of different license. That's the reason I call it symbiotic: you can use it, even profit from it through your own license terms, you just can't restrict any part of it that isn't yours (including modifications to such parts even if they are technically yours).
-Lasse
Pardon me for mixing into this discussion, but I'd like to thank you for those links - a very interesting and noble cause in any case, although I doubt I would agree with all his (her?) ideas. Designification (?) of symbols is something that I think will become very important in the future, and I'd like to explain why I think so.
Symbols are objects that are used to represent ideas of any kind. Although there are theoretically an infinity of objects that can be brought to such use, some objects are more practical, and some contexts severely limit the number of usable objects, even if the same object can be used in different subcontexts to signify different things. Take DNS for example. So for practical reasons, we do need to reuse objects for new symbols. However, the world of text (culture) becoming more and more complex, means there is a growing need for more and more symbols. Also, as symbols tend to last much longer, the binding between the object and the signified gets much stronger, WWII was recorded and imaged so much, both at the time, and afterwards in fictional and documentary form, that its symbols are almost totally frozen in the collective mind.
A Danish bar owner wrote a letter to a newspaper last week, in which he questioned the validity of recent rules on smoking in bars and restaurants, asking "how can I explain to my customers that they are not allowed to smoke in restaurants that will readily accept a Danish 500 kroner banknote with a picture of Niels Bohr smoking his pipe." The answer of course is that the two things need not have anything whatsoever to do with each other, it's all in the mind. Why make it into a problem, it doesn't have to be? Niels Bohr smoked pipe, so the image is perfectly representative. And, just like Magritte's "This is not a pipe", the image of Niels Bohr does not cause lung cancer.
The swastika didn't kill any jews. That's a fundamental fact that many people seem to have great difficulty to grasp. Don't confuse the symbol with the thing it represents; ideas in people's minds cause killing and suffering, not the symbols we use to represent them. On the contrary, we NEED the symbols to DISCUSS such harmful ideas openly, without prejudice. Likewise with bomb recipes, cartoons of Muhammed, porn, etc.
It is strange and ironic that even religions originally recognized the need for distinguishing between the symbol and the symbolized, for example the commandments about making idols or wrongful use of the name of God. Yet religious objects of so many religions are considered sacred by themselves: holy books, names, crosses, flags, places, temples, plain tap water (!), alcoholic beverages and dry tasteless biscuits. The absurdity is impressive. The jews actually seem to have handled this matter best (at least some of them), whereas both Christians and Moslems have for example confused their human founder with the deity they claim to believe in: Christians by fusing Jesus with God and whatever the Holy Spirit may be, into the Holy Trinity, and Islam by effectively putting God aside and worshipping Muhammad (through his alleged writings) instead, although they will profusely deny that fact.
You can't ban ideas, and thinking you can do so indirectly by banning symbols is stupidity of the worst kind.
Alas, even in the socalled "scientific" and "skeptical" circles, there is a tendency to raise certain ideas to a status of being unquestionable, and identify these ideas with certain symbols. Fortunately this is rare, and I think it's mostly confined to theories about the "beginning" of the universe and other fundamental ideas, and perhaps just with some people, who do not fully understand those theories, and therefore are more prone to abuse them by overinterpretation. Advances in neuroscience may be helpful here. I have no problem with the possibility that there are limits to what we can understand about the world we live in, and I can accept such limits without resorting to the creation of an imaginary omniscient demiurge or an infinity of parallel
Incredible, but nonetheless true, I see, and stand corrected. A search gives plenty of hits that indicate that the Lisa was actually capable of running Microsoft (!) Xenix. However, I doubt that it ran Xenix with any form of Apple-like GUI.
I was a hardcore Mac fanboi until the coming of OS X. I never understood why the super-intuitive direct-manipulation interface of Finder, with its 1-1 correspondence between Folder windows and Folders (ie: one specific folder only ever showed up as one window), was abandoned. As I tend to be spatially organized, I need things to always pop up in the same locations. The Finder did that. Windows and OS X "file browser" never did.
The Mac had one thing that Lisa should have had (only more of them): square pixels. At least that's what I understand from reading about it, I only saw an actual Lisa once, at a computer convention in Copenhagen in 1984.
-Lasse
I am afraid you are mixing up a few things. At least I never heard of Unix for Lisa. However, A/UX came out in 1987-88 or so. I suppose that's what you were thinking of.
For GUI consistency, I don't think there's anything in current use (even OS X) that holds a candle to system software 7.x. To this day, I haven't found a newsreader I like half as much as NewsWatcher, so that's what I continue to use - now under BasiliskII emulation running on NetBSD. Likewise with Word 5.1a, which beats any later version of Word - as well as OOo - easily IMO. (It was the last Mac Word that didn't share codebase with Windows, and therefore had at least a somewhat Mac-like UI, if not completely adhering to HIG.)
-Lasse
Indeed.
A little search tells me that MacTCP (the TCP/IP stack for Apple Macintosh) goes back at least to 1987-1988. (As does A/UX, Apple's Unix, BTW.) Eudora, according to the Wikipedia article on Steve Dorner, was developed in 1988 (and of course Mac-only), so this precedes even Win3.1 by a couple of years. For about a half decade, Macintosh had plenty of free internet tools (FTP clients, mail client, NNTP news clients, WAIStation, TurboGopher, games) whereas the DOS/Windows platform didn't even have a standard IP-stack. In the beginning, the available tools for Windows were but faint imitations of their Mac counterparts.
The Internet Revolution was already over twice (first for servers/Unix, second for desktop systems/workstations/Macs) by the time Microsoft catched up. Heck, in 1993, there still was a belief that TCP/IP and Internet would be superseded by OSI protocols like CLNP/TP4, and Apple at that time release AOCE, which integrated an API that allowed any application to gain protocol-neutral (SMTP/POP/IMAP, X.400, proprietary mail protocols such as QuickMail, AppleLink etc) mail service capabilities. This was fully integrated in the OS and the GUI.
Apple was way ahead of Microsoft when it came to understanding, utilizing, and innovating network use in general and the Internet in particular, on the desktop.
In fact, the WAIS project, under the leadership of Brewster Kahle (now famous for Internet Archive), was a joint research project between Apple, Thinking Machines Inc, Dow Jones and KPMG Peat Marwick. It's purpose was to create a system for Wide Area Information Servers, a distributed, loosely linked network of dissimilar and greatly varied information services, which were made uniformly searchable across the net through the Z39.50 protocol. WAIS, at least in spirit (and together with its Gopher-based counterpart, Veronica), is a direct ancestor of search services such as Google, and could easily have developed into something that would perhaps have been far better than WWW. Funny thing is, I *never* saw any mention of Microsoft anywhere in those wonderful days. I think IBM was mostly absent as well.
If only Apple had focused on the Internet and enhanced it with the elegance of AppleTalk instead of wasting time on overly generalized network APIs, it would have been a major factor on the Internet.
If only Apple had pursued A/UX (preferably with more of BSD underneath) instead of repeatedly getting into failed OS projects (Pink, Taligent, Copland) it would have blown Win95 away in 1996, with a system that would be very much like todays OS X. Probably better.
Apple failing so many times is what gave Microsoft the chance to become big. Also a little luck, and of course a lot of cheating and other nasty behaviour. One thing that definitely hasn't played any major role in MS' success, is merit. Therefore, MS claiming to have laid a foundation for anything, other than BG's money bin, is simply ridiculous.
-Lasse
Sweet.
Give Putin a cup.
-Lasse
I agree that that is what it says. The question is whether retaining the text means retaining the meaning of the text, which means retaining the rights also. More specific, does it make any sense to understand it to mean otherwise, that the text is retained without its original meaning? It would seem that the common interpretation has been yours, but that does not mean that it was correct, and now we have Theo saying he thinks it isn't. I am saying that the wording of the license - to me at least - makes more sense when understood as that the text of the license must be retained with its meaning preserved, applying to all the source that is redistributed. If this is not the meaningful interpretation, it definitely is a meaningful interpretation.
Now, what consequence does this requirement have?
You are almost right. However, you also could do one other thing (perhaps the only alternative), which is add a copyright notice for your modifications at the top. It could be as simple as "Portions copyright © 2007 Foo Bar", or you could mention explicit which portions have a different copyright owner. Whether you could then put a different license on those parts, is the question above. Again, I would expect the "natural" scope of the license to be the file containing it, and thus, you'd have to put your code in a different file to be able to license it differently.
I was explicit. Redistribution of works derived from works licensed under B are required to be licensed with B. That's almost what the BSD license says, although I *was* more explicit about the license actually applying to all of the derived work.
As for the pointlessness of redistributing unmodified source, I would like to retract that sentence. Although you would simply have to retrieve the work from a distribution site authorised by the author outside the terms of the license, that way it wouldn't really matter much. If I am not mistaken however, Dan Bernstein avoids the whole issue of rights to derived works precisely by only allowing unmodified distribution, forcing any derived works to be distributed as patches. So you are right, it is not completely pointless. In fact, Bernstein posits that modification and use are implicit rights attained by legally being in possession of the work (if I read his writings correctly) and therefore the only right the license has to deal with is the license to redistribute, because the rights to use and modify a lawful copy cannot be restricted based on copyright law.
No reply necessary, as I think we have reached a good understanding of the points. Again, thanks for it all.
-Lasse
Indeed. It's way out on the taurokoprometric scale. If only they could find a bullshit-absorbing mineral.
-Lasse
The BSD license doesn't use words like "derived work" and "license". It says "Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer." These are the parts that constitute the license, yet they are not called that.
So you are saying that "retain" can't possibly be understood as to apply to the redistributed (supposedly modified and thus "derived") source, and thus imply a requirement for the derived work to use the same license?
Or are you saying that (re)distribution of a derived work is not the same as redistributing unmodified source, and the requirement only applies to unmodified source?
But it does say "Redistributions of source code must retain (the license)". To me that means what goes out, not what came in. And if it doesn't include modified source, then the license only grants redistribution of unmodified source, quite pointless. I almost fear I still don't understand it. After all, what would be the point of retaining the license, unless it is meant to apply to the derived work?
But the terms of B say that you have the additional right, and that redistributed work should have license B (with that right.) How can that be compatible with G?
What would be the consequences of a BSD license variant that had just the one additional restriction, that derived works distributed as source are limited to the same license? (IOW a license that works the way I thought the BSD license might.)
-Lasse
I am sorry, but I still don't get it. How can the GPL get away with requiring derived works to be GPL'ed, then?
As for the double license, I am also still confused as to what this means.
I know you are saying that the following is actually not what the BSD license says, although it's what I understand from it. Please disregard that disagreement for a moment, and let's look at two paraphrased, quite similar licenses, B and G.
B: This license grants you rights to use, modify, redistribute source alone or binaries alone or both. If you redistribute source, including derived works, you must do so with this license B.
G: This license grants you rights to use, modify, redistribute source alone or both source and binaries. If you redistribute source, including derived works, you must do so with this license G.
If a new combined license is formed from B and G, then what rights does it grant? And under which conditions? Can a combined license be formed at all?
-Lasse
We agree that the license requires retention of the copyright notice, the conditional grant of rights to use, modify and redistribute, and the disclaimer of warranty (together making up the BSD license.) What would be the point of this requirement, if not the implication that the same license applies to the modified and/or redistributed file? Then how can you say it doesn't say that it doesn't
This is selfcontradicting. You are saying it permits redistribution under another license so long as it is redistributed under the BSD license. As long as the BSD notice is intact, the file IS licensed under the BSD license!
My "theory" is that if you try to license your work with a dual license, and forget to grant the right to relicense under either of the licenses, IOW delete the other, then your licensee has no choice but to redistribute it with the entire dual license.
Whether the copyright owner actually intended to allow such relicensing doesn't matter, if he didn't explicitly allow it, my bet is it's just not allowed. If he didn't do what he intended to do - and maybe even thought he did - then that's his own fault. It's the license, not the copyright holder's intention, that grants the licensee the rights.
You are right that this interpretation makes BSD very close to GPL. It's just not quite as infective, and it doesn't require source to be distributed.
-Lasse
If by "this debate", you mean the exchange you have been so kind to pursue with me, then I certainly owe you an explanation. I chose the subject because I think the word "relicensing" not only matters, but makes the entire difference. Unless we know what relicensing is, and whether it is a right that has been granted by the license, we don't really know anything.
If my interpretation is right - and it very well may not be, but this is what is so unclear - then the BSD license does not grant a right to relicense, it merely grants you a right to redistribute to a third party the licensed work with the same rights and conditions the copyright owner has granted you, IOW the BSD license. So you can't relicense, you can't change the license, and moreover, you agree to redistribute any modifications with the same license. Therefore there simply is no need to have an explicit clause forbidding relicensing under GPL, because relicensing never could be an issue when using the BSD license.
If I am right, then the BSD license is "freeer" than anyone seems to have imagined. You can do modify the code as you please, you mustn't even distribute anything. Only if you do, then you have to use the BSD license if you are redistributing (modified) BSD code. You can freely distribute it with other stuff using any license, including GPL or proprietary licenses. It doesn't infect anything except direct modifications of BSD source, and as there are no restrictions on use, as long as you keep the license intact, it goes along well with anything. The only thing you can't do is make it less free! (Except, of course, if you are the copyright owner.)
-Lasse
I am tempted to repeat the subject in-line. The way I understand the word "relicensing", it means licensee distributing the licensed material under a different license. This is a right which is apart from any right to use, modify or (re)distribute (subjecting the recipient to the same license as the one granting the distributor/licensee the right to distribute.)
And the BSD license doesn't allow relicensing. Do we agree that any right not explicitly granted in the license by the copyright owner to the licensee is not granted to the licensee? The 3-clause BSD license grants this, and only this:
It talks about redistribution, not relicensing. Unless you can tell me where the relicensing comes into play (a form of "use"? Nah, don't think so) as a right for the licensee, I will maintain the view that any derivative work from a BSD licensed work, has to remain at least partially BSD licensed. You can put your own license into any new files you add and for the project as a whole, and you can possibly add "portions copyright © <year> <your name>" in BSD licensed files that you modified. But these files would still be BSD licensed, including your modifications, because of the conditions under which you are allowed to produce your derivative work.
Almost exactly. My point is that this is in fact how it is now - BSD does restrict derivative works. The only two significant differences between GPL and BSD are that BSD does not require 1) redistribution of source in case of redistribution of binaries and 2) that any code statically linked with licensed code also be same licensed.
Actually, it is unclear whether the second clause could be construed as infecting the entire binaries with the BSD license.
I just found out that back in January, Brendan Scott, an Australian OSS lawyer, published a very interesting paper about issues with the BSD license. As I am not a lawyer myself, nor in any way trained in understanding legalese, I am not sure i get all the points of the paper right, but I think it confirms at least some of what I have been saying.
-Lasse
You mean:
BSD is viral in that sense already. It's just a benign virus, compared to GPL, which also infects other code, and demands redistribution, instead of just permitting it.
-Lasse
The SSL license is interesting. They are explicitly stating what I believe is already implicit in the BSD license. This hinges on whether any right not explicitly granted in the license is transferred, especially the right to relicense under different terms.
muuh-gnu, you are absolutely right about the BSD license doing exactly the same as the GPL. Except two things: 1) there is no obligation to also redistribute source if you redistribute in binary form. 2) Using BSD code with your own code doesn't infect your code with the BSD license.
-Lasse
What he is saying is not relevant. What matters is the actual license he released his work under. He may believe that it means what he says now, but unless he in fact explicitly granted the licensee the right to relicense the work under either license, all a licensee can do is to redistribute the work under the same (dual) license.
-Lasse
I am deeply sorry, I had absolutely no intention to come across as rude, nor do I think you are an idiot at all.
You are, however, unclear about who you think can relicense the code. As a licensee of BSD source code, I can redistribute it according to the license, but not being the copyright owner, I wouldn't dare to relicense it under a different license.
As I see it, modified BSD licensed source code is still BSD licensed source code. You cannot remove the copyright notice, the conditions or the disclaimer without violating the license. You could add a new copyright notice for the parts you modified or added, I suppose. The BSD license grants you a right to use, modify and redistribute, but not to relicense. "Redistribution and use in source and binary forms, with or without modification" is all that the BSD license grants the licensee.
I don't know if you do it on purpose, but your paragraph starting "My point is..." is really confusing. Which modified BSD license do you mean? And what do you consider the "original" BSD license? Do you mean the 4-condition with the advertising clause, or simply the particular BSD license a work was originally licensed under? What do you mean by "unenforcable" and "self-defeating"?
Most importantly: what right do you believe a licensee who is not the copyright owner has to relicense a BSD licensed work? In what way has the licensee been granted that right by the original copyright owner?
I really feel like an idiot myself, because there seems to be a broad concensus that indeed such relicensing is allowed, but I just can't get into my head how this is granted in the BSD license. Please have patience with me and be very clear and explicit in your explanations. Presume that I'm just not very smart.
-Lasse
How can the BSD license be superseded? OK, sure, the original author can change the license of course, but as there is no termination clause, the already distributed copies are still fully licensed and thus remain redistributable under the BSD license. Right?
The BSD license says that a modified source file must be redistributed *as BSD*. So it is simply not within the rights obtained through the license to apply GPL to the changed file. NO CAN DO! Please note that I am talking here about the generic case of software initially released as BSD, addressing the misconception that you can "close", "GPL" or "render unfree" BSD licensed code. Only the original authors can do so, and already distributed copies remain distributable under the original conditions.
Regarding what has happened in the present case, we are not in great disagreement.
I will admit that at first I thought that the original author, like seemingly so many many people, did not understand the respective implications of the BSD and GPL licenses, and "dual licensed" them by simply saying "use either one". I agree that it is his right to change the license. I do not agree that anyone but the author, given just a dual license with a choice of distribution under BSD or GPL, could modify the license, to remove the choice given (consciously or not) by the original author. If the original author's license conditions are simply "BSD or GPL", then this is a composite license which must be retained on any copy, including the choice. I doubt the GPL is different than BSD when it comes to terms like redistribute and relicense: a licensee can redistribute the work under the exact same terms as the original, but not modify the license. Only the copyright owner has the rights to define a license. Of course, the copyright owner could grant the licensee the right to redefine the license, so that a licensee could redistribute with a different license - IOW relicense it - but I hope that a permission to relicense like that would have to be explicit in the license. In fact this seems to be what the Artistic License v.2.0 is doing, for example.
Therefore, I agree that the change of license on ath5k_base.c is absolutely fine. I am not sure whether its original dual license allows relicensing by anyone but the copyright holder, but as it was changed by the copyright holder this issue is moot, unless a licensee is going to redistribute the version before this change.
The change to the file ath5k.h, was not OK according to the license. Which I understand has already been acknowledged and resolved.
I will readily admit that IANAL, that my interpretation is based on simple common sense, and therefore of course I could be completely wrong. But I haven't yet seen anything to convince me of that.
-Lasse
*SIGH!*
It's not a question at all:
(from the wikipedia article on BSD licenses)
This effectively says that IF you redistribute source, you MUST distribute as BSD. Plain and simple. You are not - unlike with GPL - OBLIGED to redistribute source, but IF YOU DO YOU MUST LEAVE THE BSD LICENSE INTACT.
Why is everyone - even selfproclaimed BSD fans - having such a hard time understanding this?
OK; playing Devil's advocate (or would that be God's advocate in this case?). In principle you could wrap the entire notice like this:
THE FOLLOWING RESTRICTIONS DOWN TO "end-of-disregard" DO NOT APPLY:
[BSD license]
end-of-disregard
But that is definitely not what the license intends, or the "spirit" of the license. It depends on how literal you interpret the word "retain". Does it mean "textually include" or "retain as the rightful conditions for licensed use of this copyrighted work"?
-Lasse
What on earth is this weird concept of "relicensing" all of a sudden? The right to license a work is precisely the right of the copyright holder. The BSD license doesn't talk about transferring any right to relicense, only the right to modify and redistribute, and I am quite sure the GPL is not much different in this regard.
-Lasse
It seems most people don't have a problem understanding that BSD source code can be used for closed source products and that's just swell.
People also understand that GPL source, when redistributed, must be GPL.
What so many people inexplicably seem to fail to understand is, that BSD source, when redistributed as source, has the same "stickyness" as GPL code. If you redistribute BSD source code - an action which, unlike GPL source code, isn't mandatory - you MUST redistribute it with BSD license. You can't redistribute BSD licensed source under GPL, although you CAN redistribute it freely - even in your GPL project - if you want to. Actually, slapping on the GPL (or any other license) is just about the ONLY thing you can't do with it. So GPL-guys cannot redistribute modified BSD code under GPL as it seems they would so much like to do. What they might be able to do is to put GPL on their modifications, which would not be a very friendly thing to do, but at least legal.
It simply doesn't make sense to equate "illegally redistributing with a different license" with "not redistributing source at all" under the label "close off the source".
-Lasse
I don't see how source code once licensed under BSD can somehow be made "unfree". Maybe we have to look a bit closer at what's going on.
Let's say we have a developer T, who develops his code on the machine beastie, and wants to release all his source code under the BSD license.
Further we have a developer L, who develops his code on the machine tux, and releases all his source code under the GPL. (Let's ignore reassigning copyright to FSF for now, in some places you can't even do that.)
Last, we have a company M (all developers here do work for hire, so the copyright belongs to M), developing code on the machine mtdoom, not releasing any source code at all, but distributing binaries.
T develops foob.c, L develops foog.c, and M develops foom.c.
case 1) If M wants to use foob.c in his project, he is allowed to do so, provided he acknowledges this use. He is not required to distribute the source, and his own code is not infected with the BSD license. Fine, a copy of foob.c is now found on mtdoom. A discreet notice of the use of foob.c goes into the documentation.
case 2) If M wants to use foog.c in his project, he is allowed to do so, provided he also distributes a copy of the file. Further, his own code must probably also be released with the GPL license. M doesn't want that because of his modus operandi, so effectively he cannot use foog.c. Let's just hope foog.c is not to be found on mtdoom
case 3,4) If L or T want to use foom.c - tough luck. They need to get M to distribute the file under some license they can use. No other way can they legally get a copy of foom.c onto their machines and into their project.
case 5) If L wants to use foob.c in his project, he is allowed to do so. He can redistribute the file freely. No reason why foob.c shouldn't be copied to tux. L can even distribute foob.c as is, as part of his GPLed project. foob.c is now on tux.
case 6) If T wants to use foog.c, he is allowed to do so. But like case 2, depending on various things, he may have to release the rest of his code under GPL as well. T, just like M, doesn't want that, and is thus effectively unable to use foog.c. Of course, a copy still might find it's way onto beastie, but it can not be linked with any of T's own code. T can merely redistribute this code under GPL just like any other friend of L. (This actually applies to case 2 as well.)
So effectively, the only file that could be freely copied everywhere is foob.c. Now what can actually be done to these copies of foob.c?
As mtdoom is a proprietary machine, we can't tell for sure. But for M's own sake, let's hope the license is retained, lest they forget to keep their minuscule obligation to T. M is allowed to modify it of course.
L however, likes to modify foob.c and redistribute the modified foob.c. Again, this is granted. However the license says
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
So L has to leave at least the license text intact if he modifies the file.
The questions that remain are:
can L "infect" foob.c with GPL?
NO! foob.c on beastie cannot be infected and will remain accessible and distributable. However, I don't know if L could slap a GPL onto his copy of foob.c given he did sufficient modification? And would that GPL "override" the BSD license already attached? It would seem certain that it could not _replace_ the BSD license, as that's what allows the use and modifications in the first place!
Particularly relevant for this discussion: let's imagine a hybrid developer H, releasing fooh.c on his machine hoohoo, who for some reason wants to leave the choice of BSD license or GPL up to the licensee. Does it even make any sense, any difference at all, to have a dual license? What would the point be? What does that choice mean? Does it really mean that L can make a copy of fooh.c on tux, and just REMOVE the BSD license and redistribute it under GPL only? Without
Oh, I thought that was Berkeley?
Anyway, the geek equivalent: If you have a sequence of bits, and don't know the decoding algorithm (if one exists), how can you tell whether the sequence is random, or actually interesting information that just happens to be compressed/encrypted? Maybe you could invent some new decoding algorithms, and see if anything interesting comes out -- but then : is the interesting stuff a result of applying your algorithm to this specific random data, or the information originally encoded in the data? You can only know the right answer if you cheat: cBwfCYG lr0 bGCkx TYJScPAtB yxF8jhtr0 XYYSjL iz9so CreZfPX pjzpUW9Cj1vP AoKEhc6hAF wA3 J2Y kDi5q 11KzWsm 9Mp5ZQvGo.
Somehow my mind wanders off to think about monkeys and Shakespeare...
As for me, I don't care much about infinity: the finite world is more than big enough. Infinity was invented to understand in finite small terms the very large. Mistaking it for something real is merely foolish abuse - or religion, which is just another name for "foolish abuse".
The number of jokes
that can be told in three lines
I need not hear all
Instead of wasting effort on space exploration, we should waste it on reducing suffering. Just as futile, only much nicer.
-Lasse
OK, your method ensures a clean desk, which is neat, but perhaps you should try lifting the mouse and reposition it at the opposite edge of the mousepad when you reach the edge and the pointer isn't "there yet". It won't hurt anything, if that's what worries you. I am curious: what would you do if you added another monitor? Upgrade to an A2 sheet?
-Lasse