The Open Software License, Version 3.0
Most of the problems found by this committee were related to terminology. Most open source licenses, for example, allow the licensed software to be redistributed. Under the European interpretation, however, "redistribute" has a narrower meaning; in particular, it does not include acts like making the software available for general download on the net. The essential right for this sort of redistribution is "communicate to the public." Without an explicit grant of the right to communicate the licensed code to the public, the possibility remains that some court, somewhere, could conclude that putting a tarball on a web site is a violation of the license.
"Virality" is another concern of the authors, who see the GPL is being rather more "viral" than the alternative licenses. In particular, the authors see dynamic linkage as a barrier over which the concept of a "derived work" cannot cross:
The Free Software Foundation, instead, does not feel that the type of linking used affects the copyright status of the resulting program. This distinction is important; it could, for example, affect the status of proprietary kernel modules. Because they disagree with the FSF's interpretation, the report's authors shy away from the GPL, even though other "copyleft" licenses contain similar language - and copyleft is what the authors say they want.
A few other details caught their attention. Licenses in Europe, for example, are generally not allowed to outlast the corresponding intellectual property protection period. The terms of a copyright license thus cannot be imposed after the covered work has gone out of copyright, should that ever be allowed to happen again. Some details in warranty disclaimers are different, and there are certain types of warranty which cannot be disclaimed.
In response to this report, Lawrence Rosen, the author of the Open Software License, has announced a draft version 3.0 of the OSL [PDF] for review. The draft is annotated so that it is easy to see what has changed from the current version (2.1). Most of the changes are fairly obvious given the discussion above: the OSL now explicitly grants the right to "communicate" the software, for example. The license is no longer "perpetual"; instead, the copyright and patent grants are for the copyright and patent protection periods, respectively.
There are a couple of new terms which might not be popular with all users of this license, however. The "acceptance" clause now includes the following text:
This language is a response to concerns about whether a license can truly be binding in Europe if the licensee has not explicitly accepted it. The "reasonable effort under the circumstances" might include an active copyright acceptance step required at download time or when the software is installed. It is unclear what might be expected of a distributor shipping OSL-licensed software mixed in with thousands of other packages.
The new license also adds:
This looks like the return of the unlamented BSD advertising clause. It is less onerous, however, in that it only requires attribution in places where the redistributor is asserting copyright claims. Still, a splash screen for an application built from several OSL-licensed libraries could get unwieldy. Mr. Rosen states:
It is not clear how much of a problem this has been in the real world, and whether it truly needs fixing.
The OSL is not a hugely popular license; Freshmeat claims that the OSL
applies to 0.15% of the projects listed there. There are some important
projects using the OSL, however, including Rails, Globus, ImageMagick,
and sparse. This license is well respected and carries a certain
influence. Its importance could grow if it comes to be seen as the license
to use for those who are especially concerned about adherence to European
law. So this proposed update is significant. For those who are
interested, the discussion is happening now on the Open Source Initiative's
license-discuss mailing list.
Posted Aug 15, 2005 19:25 UTC (Mon)
by elanthis (guest, #6227)
[Link] (1 responses)
Posted Aug 15, 2005 19:31 UTC (Mon)
by corbet (editor, #1)
[Link]
Posted Aug 15, 2005 19:47 UTC (Mon)
by JoeBuck (subscriber, #2330)
[Link] (9 responses)
Posted Aug 16, 2005 4:01 UTC (Tue)
by sanjoy (guest, #5026)
[Link] (8 responses)
Posted Aug 16, 2005 5:35 UTC (Tue)
by Ross (guest, #4065)
[Link] (7 responses)
As an aside, I wish they would avoid using the term "viral" to refer to copyleft where merging (though linking or header files) is considered a derivative work because it is misleading. In the case where more software licenses are involved, it would be a copyright violation. When the GPL is involved, is is a copyright violation. I see no difference other than in the GPL case you have the _option_ to dual-license your code under the GPL (or any compatible license) to avoid the infringement. You can always choose not to use the software in question.
Posted Aug 16, 2005 14:30 UTC (Tue)
by elanthis (guest, #6227)
[Link] (6 responses)
Posted Aug 16, 2005 16:31 UTC (Tue)
by arcticwolf (guest, #8341)
[Link] (4 responses)
Which you don't: you also have the option of removing the offending GPL'ed code. And that is of course assuming that you distributed the software at all in the first place, as opposed to using it for internal purposes only.
Posted Aug 18, 2005 18:04 UTC (Thu)
by bronson (subscriber, #4806)
[Link] (3 responses)
So how can you possibliy claim that the GPL is not viral? Sounds pretty clear-cut to me!
Posted Aug 18, 2005 19:32 UTC (Thu)
by oak (guest, #2786)
[Link] (2 responses)
Posted Aug 18, 2005 23:09 UTC (Thu)
by giraffedata (guest, #1954)
[Link] (1 responses)
I think there was some incorrect reference to "GPL-compatible" in this thread. If I have a piece of code A that I'm licensed only under GPL to distribute, and I write some code B and combine them to form program AB, and then I distribute AB, I must license all of the code, including my B code, to the recipients under GPL. Not GPL compatible, but GPL itself.
I can separately distribute just my B code under any license I please.
The GPL-compatible license comes into play if I want to add in code C, which someone else wrote. If I distribute ABC, I must distribute all of it -- A, B, and C -- under GPL. That means that the author of C must license C to me under some kind of license that gives me the power to redistribute it under GPL. Such a license is GPL-compatible.
Posted Aug 25, 2005 7:22 UTC (Thu)
by Wol (subscriber, #4433)
[Link]
You can distribute your combined work and just say "A is GPL, therefore B must be GPL too. But B is also X-licence, if the recipient wishes to extract it and use it elsewhere".
It's then the recipient's responsibility to make sure he doesn't accidentally include A when he takes B code to use in his product C.
Cheers,
Posted Aug 25, 2005 21:15 UTC (Thu)
by dmag (guest, #17775)
[Link]
Posted Aug 16, 2005 10:39 UTC (Tue)
by dw (subscriber, #12017)
[Link] (2 responses)
I don't know of any people using Rails yet in production, most are still only toying with it. It is probably a philosophical can-of-worms opener, but I can't help but wonder what your definition of important is. :)
David.
Posted Aug 18, 2005 18:19 UTC (Thu)
by bronson (subscriber, #4806)
[Link]
If importance only depended on installed copies then SPICE would be a non-starter. In reality, SPICE was one of the more influential programs to come out of the last 3 decades.
Posted Aug 22, 2005 20:30 UTC (Mon)
by BrucePerens (guest, #2510)
[Link]
Bruce
Posted Aug 16, 2005 16:57 UTC (Tue)
by sanjoy (guest, #5026)
[Link]
Posted Aug 16, 2005 17:41 UTC (Tue)
by zooko (guest, #2589)
[Link] (3 responses)
Posted Aug 18, 2005 9:04 UTC (Thu)
by ber (subscriber, #2142)
[Link] (2 responses)
Proprietary software is "viral" in the sense that you are often
drawing much more software in because it often works best with other
proprietary software. The motivation behind this is clear, it is a lot of
effort to make software work together and in the proprietary world you are
payed by new features and updates.
Posted Aug 18, 2005 18:26 UTC (Thu)
by bronson (subscriber, #4806)
[Link] (1 responses)
Posted Aug 21, 2005 16:42 UTC (Sun)
by man_ls (guest, #15091)
[Link]
Even regular copyright seems to abide closed-sourceness, if we read Adobe's attempts to enforce digital restriction management in PDF. This is embedded in the specification for the PDF format, which is otherwise open. It is hard to see how restriction management can be done in free software, since it might be easily removed.
According to your definition, many document formats, protocol specifications, etc. would be "viral".
Posted Aug 16, 2005 18:16 UTC (Tue)
by Max.Hyre (subscriber, #1054)
[Link] (1 responses)
I glanced at the headline, saw ``License'' and ``v.3'', and filled the blank with ``GNU General Public''. That's what I get by going for the story before my mind has caught up. Speaking of needing more coffee...
Posted Aug 18, 2005 0:52 UTC (Thu)
by pengo (guest, #7787)
[Link]
Posted Aug 18, 2005 7:30 UTC (Thu)
by tsr2 (subscriber, #4293)
[Link] (4 responses)
concerns about whether a license can truly be binding in Europe if the licensee has not explicitly accepted it. I wonder how the licensee can then be licensed, if they are using a piece of software without accepting the license. Is this really any different to the US? Surely it's illegal to use someone else's copyrighted work, without accepting the terms it's licensed under, regardless of whether you are in Europe or the US?
Posted Aug 18, 2005 9:10 UTC (Thu)
by job (guest, #670)
[Link]
Posted Aug 18, 2005 22:48 UTC (Thu)
by giraffedata (guest, #1954)
[Link] (2 responses)
I'm just guessing, but I'll bet the difference is that in Europe a copyright license is more of a contract: In exchange for the right to distribute my code, you promise to license your own code that you add to it to the public in certain ways. The damage you do by breaking your promise can be more than the damage you do simply by distributing my code without my permission. But if you didn't accept the "license" terms, there is no contract, and all you owe me is the damage you caused by the copyright violation.
Posted Aug 25, 2005 11:35 UTC (Thu)
by forthy (guest, #1525)
[Link] (1 responses)
Posted Aug 25, 2005 15:33 UTC (Thu)
by giraffedata (guest, #1954)
[Link]
Copyright licenses in the commercial world are typically tied to a contract of sale like this; it's special to the open source world that we give them out unilaterally (albeit with conditions attached).
In this case, I'm not entirely sure there's even a relevant copyright license. I think what you get in exchange for your promise not to check that box is the copy of Windows that you physically received.
That's the first I've heard of that theory, and I can't see any substance to it. I think if the hardware is unavoidably bundled with the software, then you're buying the bundle -- both the hardware and the software. Why would that one component of the computer system be considered not part of the merchandise?
Posted Aug 18, 2005 19:12 UTC (Thu)
by KaiRo (subscriber, #1987)
[Link] (2 responses)
Now, if I want to display the name "myapp" e.g. on the title bar, on my download site and other places, I always have to write "myapp based on mylib, hisapp, libhers, ..." and have a package name of "myapp-based-on-mylib-hisapp-libhers-and-others-0.1a.tar.gz"? Nice...
Posted Aug 25, 2005 7:28 UTC (Thu)
by Wol (subscriber, #4433)
[Link] (1 responses)
If my splash screen got unwieldy I have the choice of leaving it out. I don't think the BSD licence lets you do that!
The OSL merely requires that I give *equal* prominence to other peoples copyright as to my own. Should I change the flash screen to read "copyright assorted people, see manual for details", that's perfectly okay according to the OSL. But not according to the 3-clause BSD.
Actually, I think that's a damn neat way of doing things :-)
Cheers,
Posted Aug 31, 2005 22:37 UTC (Wed)
by sanjoy (guest, #5026)
[Link]
Then you'll be unhappy that the latest draft of the OSL 3.0
I think the OSL is an excellent license. Debian probably thinks
It's called the Open Software License, not the Open Source License.Open Software License
Sigh. I knew that, really did. Fixed now. Off for more coffee.
Open Software License
Does the OSL now have anything equivalent to the "mere aggregation" clause of the GPL? If not, it seems to still be a violation to ship an OSL work on a CD-ROM that also contains other works whose licensing terms are in any way contract the OSL (in particular, any GPL'd works).
The Open Software License, Version 3.0
It implicitly has an aggregation exception because the OSL copyleft applies toThe Open Software License, Version 3.0
derivative works, not to collective works (which you get
by 'mere aggregation'). Though I wish the license
explicitly stated this distinction (as the Creative Commons sharealike
licenses do) rather than leaving it to be inferred
from the defintional boundaries in (US) copyright law -- and
I brought up this point in an email on the license-discuss
list.
Not just US law, but whatever county it is being interpreted in. From the article it looks like one of the goals is to handle unusual aspects of other copyright laws (like Germany).Clarity and terminology
Quite a few open source licenses don't have that behavior, though. I can merge BSD code with code under any other license, for example. Compared to proprietary licenses, sure, GPL code is no more "dangerous" to mix in with your own, but compared to some other open source licenses, the GPL is unattractive to proprietary developers. Whether you actually care about those developers is the distinction between whether the GPL is good or bad, I guess.Clarity and terminology
No matter what you think of the GPL, it seems unfair to describe this characteristic as "viral", though - this term seems to be rooted in the misconception that when you incorporate GPL'ed code into your own project, you *have* to GPL your project as well.Clarity and terminology
You said it yourself. If I want to incorporate ANY GPL code into my project, I have to make my ENTIRE project GPL-compatible. The only alternative, as you say, is to not incorporate the GPLed code.Clarity and terminology
> If I want to incorporate ANY GPL code into my project, I have to make my Clarity and terminology
> ENTIRE project GPL-compatible.
I'm not sure what you mean by entire. If it's for example a Linux
distribution, you need to have as GPL compatible only the part that can be
legally interpreted as derivative works. (The definition of what is
derivative works is not so clear-cut though.)
> The only alternative, as you say, is to not incorporate the GPLed code.
Or ask the copyright holders for a license for the code that allows
incorporating the code. This is a base of e.g. business around Qt and
MySQL.
These are the same options as with proprietary code/products. With GPL
you just have the additional possibility of licensing your derived work as
GPL.
Calling GPL "viral" is about equal to calling proprietary products
"encouraging copyright infringement" because you don't have means to
legally distribute derivate works freely (free in the "free beer"
sense)...
That's a good point about the term "virus." A biological virus doesn't just tempt a cell to make more virus. It enters the cell by force and forces the cell to make more virus. GPL, on the other hand, is strictly quid pro quo.
Clarity and terminology
You don't have to SEPARATELY distribute your code B under a different licence at all.Clarity and terminology
Wol
If you are going to say that the GPL is viral, then you have to say Copyright is viral. Windows code is just as viral as the GPL. (Try incorporating code from Windows 95 into your application. Even if you only include some of the Windows 95 binaries, and not the source, you will still be in trouble.)Clarity and terminology
As a regular reader of LWN I've often wondered what you mean by "important" when applied to open source projects of various kinds. In this case, the important label is being applied to Rails, the Ruby web framework that only in the last 6 months or more has really gained the social spotlight.The Open Software License, Version 3.0
"Important" usually refers to mindshare, not installed numbers. Mailing list, IRC, and blog traffic are usually pretty good indicators. By these measures, Ruby on Rails has become very important in the past 6 months. But, if you want production sites, check out basecamphq.com or the rubyonrails.com homepage for more.The Open Software License, Version 3.0
David Heinemeier Hansson, author of Ruby on Rails, tells me he has never heard of the OSL, and intends to stick with the MIT license. What Rails were we talking about?NOT APPLIED TO RAILS
The acceptance clause is not new. It was in v2.1 but got moved around in the paragraph (so the red annotations show it as an addition in the new spot).The Open Software License, Version 3.0
A better term than "viral" is "transitive". Virusses self-propagate -- enter into other systems and infect them and reproduce. Licenses are not allowed to do that under current laws in the developed world, thank goodness, and the GPL certainly doesn't attempt to do that. The notable fact about the GPL is that some of the requirements are "transitive" -- you are allowed to use GPL'ed software under certain terms only if you extend the same terms to users of the software which you have combined with the GPL'ed software. This is a requirement on your use, and it is a transitive requirement. It is not viral.The Open Software License, Version 3.0
As the GNU GPL tries to preserve the freedom of software users and
developers, it has a "vaccinating" or "freedom preserving" aspect.
"vaccinating" or "freedom preserving"
That argument sounds pretty thin, don't you think? If a closed-source program required its files to be manipulated only by other closed-source programs then, yes, that program would be viral. Short of that, it's not."vaccinating" or "freedom preserving"
In practice, they often do. Either you must agree to keep information to yourself (signing an NDA), or you are forbidden from distributing crypto keys, or you cannot reverse-engineer an algorithm (because of the DMCA and equivalent European legislation).
"vaccinating" or "freedom preserving"
Blast, it's not /GPL/ v.3!
You weren't the only one :)Blast, it's not /GPL/ v.3!
The Open Software License, Version 3.0
Depends on where you are and what you mean with 'use'. Not all use requires a license. A text, for example, is something you normally can expect to have the right to read without adhering to any licenses. That, however, is not always true for computer programs. (IANAL, etc)The Open Software License, Version 3.0
That's a good point, which people always miss because of incorrect wording such as "whether a license can be truly binding." In the US, a copyright license is never "binding." Copyright law is binding, and the license unbinds. A license permits, it doesn't restrict. You don't violate the license, you violate the copyright.
The Open Software License, Version 3.0
Have you ever removed the Windows XP-Home registry entry that limits the The Open Software License, Version 3.0
functionality of XP-Home? An alert box pops up and tells you "by removing
this item, you violate the license". Afterwards, you can log in as admin
in normal mode, create "normal" users, and thus do the most basic things
to make XP-Home a bit more secure than as delivered. But you violate the
license.
Copyright, on the other hand, allows exactly this sort of patching. Well,
it's not even a patch, it's just removing a registry entry, it's use. It
does not touch copyright at all.
Microsoft considders licenses as contracts. They try to prevent
publication of benchmarks. They had wordings like "you aren't allowed to
speek badly of this piece of shit^Wsoftware" in some of their EULAs. They
even put some ban of free software into some of their licenses. They
actually think that copyright is too generous, and want their license to
restrict you further.
The fact that you usually don't buy Windows, but a piece of hardware,
unavoidable bundled with it, renders this "contract" unenforcible. But
that's IMHO and IANAL.
That sounds like another example of the same misnomer, then. I'm sure what Microsoft means is "you violate the end user license agreement." That's a contract wherein Microsoft gives you a copyright license in exchange for your promise to do (or not to do) some stuff. If you break that promise, you're no longer entitled to the license. Under ordinary contract law, you'd still have the license but would owe Microsoft money. But there may be some special provision for copyright license sales that says the license becomes automatically revoked.
The Open Software License, Version 3.0
The fact that you usually don't buy Windows, but a piece of hardware,
unavoidably bundled with it, renders this "contract" unenforcible. But
that's IMHO and IANAL.
The advertising clause looks interesting. Say that I have "myapp" licensed under the OSL3, which counts as a derived work of a bunch of other code from mylib, hisapp, libhers, etc (say they're all stically linked so that we can be sure we are derived) - and all that code is under the OSL3 as well.The Open Software License, Version 3.0
But you have a choice.The Open Software License, Version 3.0
Wol
> "That's a damn neat way of doing things"The Open Software License, Version 3.0
has taken out the new attribution sentence. You're still required
to preserve attribution notices in the source code, though, as
in the OSL 2.1.
it's nonfree due to the external deployment or patent-defense
clauses (and maybe even the 'express assent' clause) -- both of which I like (I'm not sure about the express-assent clause). But
that just means I think Debian's test for being free is too
strict. The OSL, unlike almost every other free software license,
works equally well for documents as for software, so you can
license your entire free-software package with one license, or you
can use it for textbooks (what I will use it for).