Slashdot Mirror
FSF: New Apache License not GPL-Compatible
__past__ continues "The new version of the Apache license will apply to all Apache projects, including the popular web server and many Java libraries like Xerces and Log4J, and making it easier to integrate Apache- and GNU-licensed code was one of the primary goals for its development. With the new license being GPL-incompatible (just like the older Apache licenses were), it is not possible to distribute programs that use libraries covered by under it and others covered by the GPL.
Apparently, the FSF does not actually consider the patent-related clauses a bad idea, let alone non-free - it is just that they impose a restriction that the GPL does not, and that makes the license automatically incompatible. It might even be that GPL Version 3 will include similar statements or at least allow them, as a message from FSF legal counsel Eben Moglen indicates. Additionally, prominent Apache hacker Roy Fielding claims that it doesn't really matter what the FSF thinks about the matter, because according to the Apache Software Foundation, derived works can just be distributed under the GPL."
Slightly concerned that we'll look back and say "Well, 2004 would've been the year Linux arrived in a big way . . . EXCEPT THAT WE TORE OURSELVES APART AT THE SEAMS."
I don't mean to panic-monger or scream that the sky is falling without due cause - but this is all starting to get a bit worrying. Open Source has enough problems right now without actively helping its opponents.
--Ryv
So since a good portion of XF86 is apparently GPL (right?)...Will they have to change the license back or be faced with dropping all of the GPL'd portion?
I feel like i've been sleeping for months, all of a sudden this is all hitting the fan? What, will we all have to run some Hurd variant soon to be fully compliant?
CB
free ipod and free gmail!
1)Add some new comments to the Apache code
2)Recompile
3)...
4)Profit!!!
So that's the answer then, we just change a few comments, recompile, and call it a derived work? Surly it can't be that simple?
SCO.com uses Linux
I believe that you can distribute a program under the GNU General Public License and a seperate Trademark license. That is what AbiSource does with AbiWord. And I don't think it restricts the freedom of the user since it is still allowed to distribute derived works. What does not seem to be compatible with the GPL is trying put any further restrictions on the user by invoking normal copyright law. By trying to use copyright law in the Apache license to restrict the rights of recipient to use an arbitrary word in the use of their derived work doesn't seem to be compatible with the GPL. (That arbitrary word would of course be Apache in this case :) Although if you have a real trademark on that word then clause
4 and 5 could be seen as just stating that the Apache License does not
grant someone the right to use the word Apache since it is a trademark.
But if that is the case then you could easily take away all confusion by
using something like the following instead of clauses 4 and 5:
This license does not grant you the right to use any of the trademarks
of the Apache Software Foundation. "Apache" is a trademark of the Apache
Software Foundation and products derived from this software may not be
called "Apache", nor may "Apache" appear in their name, without prior
written permission of the Apache Software Foundation.
I think that would make sure that the new Apache License is compatible with
the GPL.
The problem is not one of the GPL being problematic, but rather people not thinking through what they're doing with their licensing. One of the GPL's purposes is to ensure that the code stays available to those that are interested in it as long as they are so interested. Pretty much only the GPL and LGPL do this of all the licenses out there. The other variants do this to some or no degree (MPL and APSL do some of this, but they're not as strong as the GPL is in this regard and the BSD/MIT/Artistic variants tend to not protect you from propritization of the code at all...).
In the case of the new XFree86 license, it's a stupid play on their part to try to get more recognition (there's other ways to do this, folks- not a single player in the FOSS community is claiming that they're the ones that produced XFree86 at this point in time.) and I'm sure that the Apache license is probably another example of something not being quite thought out in the ramifications department.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
Probably. And that's exactly why the FSF suggests saying that you may redistribute under GPL version 2 *or later*.
Opus: the Swiss army knife of audio codec
Has OSS reached a stage in the lifecycle that few siblings are big enough to branch out now?
In the early days, all everyone cared about is making OSS, and the whole model is the production of best software based on peer-review.
Nowadays, some OSS distributors are gaining enough cash reserve, support and momentum and it's sad to see other agendas come into play.
These few weeks we have seen enough incidents already.
Rock that crushes, Paper & Scissors that don't matter.
Many developers have strong convictions about which license they wish to use when releasing their code. However, I think that they'd often rather reuse and extend an existing library that does not use their license of choice than be burdened with re-implementing that functionality within their products or creating a new project with their license-of-choice. These kinds of incompatibilities encourage duplication of effort and discourage collaboration on many projects.
I often wonder whether this problem could be mitigated or even solved by some creative license language. I'd like to license my software in such a way that it could be reused by projects using any of a majority of the other open source licenses. Also, I'd like to modularize it so that it could take advantage of high quality software released under otherwise incompatible licenses.
So there is some light at the end of the tunnel.
It seems they said long ago in their Halloween Documents that "The lack of singular, customer-focused management has resulted in the unwillingness to compromise between the different initiatives and is evident of the management costs in the Linux process."
In my opinion, this recent XFree86 (and now Apache) business is further proof that Microsoft was right about this. I'm not trying to bash open source as a whole--I am a big Linux fan. However, I think this problem MUST be solved if the OSS community is to move forward. We cannot go on having endless fragmentation of projects, proliferation of different (non) standards and forks and everyone-going-their-own-way. A truly usable desktop OS's bread-and-butter is its ability to have truly inter-operable (dare I say this--horizontally integrated) components.
Just my 2 cents worth.
I, for one, use plenty of non-GPL software in my day to day life. I enjoy GPL software, but I tend to release open source software under more permissive licenses, such as MIT. I also use (and write) a lot of (gasp) closed source software as well.
For the love of Pete, there are plenty of different software licenses out there. If you don't like the terms of a given license, don't use the software. If apache changes their terms because they think it makes them less likely to be sued, good for them. The GPL isn't the One True License, it's just one of the more restrictive ones that still claims to be "free". News flash: you can run software with virtually any license as long as you agree to the conditions. If some GPL-zealot distribution decides not to include Apache because of this, that's their problem.
F- the GPL. I need to write an updated version of this that follows a more legally friendly format, but:
http://people.freebsd.org/~seanc/ossal/
The GPL is the worst thing to happen to software development and stability in the last 20 years.
-- Sean Chittenden
> with all these problems maybe the GPL should change.
No, the GPL is doing *exactly* what it was intended to do: force everything
(well, actually, just everything you want to use together with any GPLed
code) to be licensed under a license that either *is* the GPL, is just *like*
the GPL, or is liberal enough to allow the code to be relicensed *under* the GPL.
If we want to get around this restriction, we're perfectly free to not use
GPLed code. Which IMO is exactly what the community should do: when there's
GPLed code that you'd like to use, but it's incompatible with the licensing you
want to use, rather than changing your licensing to support the FSF's extreme
idealistic stance, just replace the GPLed code with other code that's licensed
in a compatible fashion (e.g., BSD). This is exactly the approach that the FSF
takes to licenses with restrictions they don't like (e.g, proprietary licenses):
just don't use that code.
At this point, there's a *large* amount of GPLed software that we heavily rely
upon, so for practical reasons we would have to be gradual about moving away
from it. We'd have to replace Emacs (no small task), gcc (or just replace the
C and C++ development languages altogether...), the whole suite of Gnu tools,
and so on. But OTOH if we keep the GPL we will ultimately have to replace
everything that's got *different* restrictions, as we're seeing.
The BSD people have the licensing thing right: it's compatible with pretty
much *everything*, so pretty much everyone can use it. This isn't the tack
RMS wants to take, because he specifically wants to eliminate software that's
licensed under incompatible licenses. (He doesn't need to eliminate code
that's BSD-licensed, though, because it can be relicensed under the GPL at
will.)
I must disagree
If the copyright owner says "Sure you can do that", then you can. The only reason for all the paperwork people have is to prove agreements. In most countries permission under copyright law doesn't need to be written and signed.
Apache (as owners) said you can GPL derivative works if you want - end of discussion.
Trademark btw works very similarly except that its easier to create a promise by inaction ("estoppel"). You can in some cases lose the ability to enforce copyrights in some situations through estoppel too, but you don't lose the copyright per se just the ability to enforce it in some situations
Apache adding it as a footnote to their license would neaten it up but its hardly essential. Of course you then have openssl and other bits to consider. Really no standard loaded set of Apache packages has ever been GPL compatible, and except for the mysql4 problem nobody has had any problems due to it.
You are right, the situation has not become worse than it was. But it hasn't become as good as was expected either. The GNU project and the Apache Foundation are arguably the two most important FLOSS projects, and that interoperability between them keeps being hampered by incompatible licenses is just annoying, especially when it is only due to stupid legal details that are in principle non-critical to either party.
Programming can be fun again. Film at 11.
"- all changes are clearly labelled"
This would require someone using the source code to provide a HUGE list of updates. Where should this be clearly labelled? In the documentation? In the splash screen? In the About-box?
What if there were several companies providing changes to the original source code. Where would the end be? After 10 years, how many changes would be listed?
"-the original author(s) are credited as the creators of the original code"
Why only the original? What if someone contributed something that was more than a 100% of the original contribution? If the original contributors were credited, shouldn't them? And where should that be presented?
"-if this code is used as part of a commercial product, any modification of a source code file which was an original part of this software's source code must be made available under this license."
Eh.. So you are using the original source files as the judgement of modification? So if someone copies the source files completely, and writes around all your problems in other source files without distributing changes, then that is better than someone changing your source files without distributing the changes?
The thing is, a license is complicated if it tries to do something other than give away all rights..
GPL or Copyleft gives away most rights, but not all, and tries to make sure that the work done by the original contributors is kept open without ending up being used in closed source projects without the copyright holder being asked about it. I find that to be perfectly fine.
The problem is that you can't write something really simple in legalese. All the possible problems, of which I've mentioned only a few, have to be taken into account, or you might end up in a lawsuit not going your way, sooner or later.
Including, unfortunately, letting everyone use if for free for the first 15 years, then charging once it becomes a ubiquitous standard, such as happened with GIFs. [...] You can "undo" it for future releases, but whatever you already GPL'd stays that way forever.
Licensing is a form of contract. It works the same way for patents as for copyrights. If you explicitly allow a patent to be used for free under certain conditions, you can no more undo that than you can undo specifically allowing a copyright to be used for free.
The compression algorithm for GIFs never was offered for free in the first place. It just took the owner a long time to complain about its use.
Moderating "-1, Disagree" is simple censorship. Have the guts to post your opinion.
If I want to distribute a derived work made partially from Apache software under the Apache license version 2, and partially from Foobarco's software under the GPL version 2, the fact that Roy Fielding (or even the Apache Foudnation) is OK with it does NOT solve the problem. Distributing the derived work under the Apache license terms is a violation of the GPL, and Foobarco would have grounds for action.
Or even better yet, caudium http://www.caudium.info/
- as a past roxen user, I find caudium to be more free in every sense.
I've been using Linux since my first Slackware disk in 1994 or 1995. But version 2.0 of the Apache license, alone or in conjunction with Evan Moglen's assertion that FSF is going to revise the GPL to deal with patents, is enough to keep me from ever wanting to touch another Linux distribution.
First, exactly what harm has free software, and specifically, GPL-ed software, ever seen from software patents? Yes, I am aware of the Eolas decision against Microsoft, but has the owner of that patent given any hint of being interested in suing the writers or distributors of any GPL-ed software for money? Yes, I know about the GIF patent, but was anyone distributing or using free software actually sued over that patent? To my knowledge, the patent owner only demanded that the patent not be used in free products wihtout payment of a royalty.
Second, if you look at the damages available to a patent owner as compared to that of a copyright owner, you will see that many of the draconian remedies and huge statutory minimum damages available under copyright law just aren't available under patent law. In most instances, I would imagine that a patent holder would simply demand that the patented feature be taken out of the GPL-ed product until the patent expired. There would be very little incentive, if any, to go after ordinary users.
But more importantly, the Apache license as written seems to guarantee that Apache would act as a magnet for stolen intellectual property. Almost anyone with an axe to grind against some company (the fact that the company has software patents would likely be enough) and knowledge that GPL-ed products or Apache-licensed products are used in that company would be motivated to try to incorporate those patents into a code contribution as a matter of spite. Thus, the "free" software product with the patent license revocation clause would almost certainly become littered with stolen IP and be vulnerable to attack from many fronts.
Even worse, I understand that at least one company (i.e., IBM and possibly others) that supports free software selectively releases their software patents for use with GPL-ed software that they contribute, when the contribution would otherwise be covered by their own patent. How much longer do you think this company would be willing to contribute to free software if it could not enforce its other patents against someone who is maliciously stealing IP from the company?
I'm sorry, but I'm not interested in using or recommending Apache to anyone if the version 2.0 license is in effect. And if Evan Moglen allows FSF to incorporate an "in terrorem"-like clause concerning software patents into the GPL, then I will lose a great deal of respect for him as a practical attorney. I think clauses like this will increase the need for attorneys to study software licenses at companies, thereby increasing the cost of free software. And the term itself will inhibit both the adoption and advancement of free software.
Not so. Read the License. The Patent provision is only binding on the contributor's patents (i.e., if you own patents that will be necessarily infringed by your contributions to Apache, then you automatically grant a patent license for use in Apache (and derivative works, I assume) ). If the contributor then sues Apache for patent claims, then the license revokes their license to use the software.
It is a reach to suggest that someone stealing IP from a company and contributing it to Apache would bind that company to accept that their IP has been released. Furthermore, the license provision is only triggered by litigation against Apache, and it is unlikely that a company whose IP has stolen (except for a certain desperate Utah-based company) would start right off the bat with litigation.
That is manifestly NOT what happened in the case of the Unisys LZW patent. Unisys stated informally to everyone that they could continue to use LZW compression in software for GIFs or whatever, then reversed their decision and instantly made tons of shareware and gratis software authors liable for IP infringement.
I know that was a case of patent law and not copyright law, but I believe the principles are the same. Most IP licenses are revokable by the IP owner; unless they tell you otherwise up front, the IP owner can come back and tell you you have a completely different set of rights and costs to use their IP tomorrow. Formal licenses like the GPL can explicitly agree to be irrevokable; informal agreements like you are talking about usually do not include this. In other words, anyone can pull a Unisys and tell you one thing while quietly getting ready to extort licensing fees from you.
Secession is the right of all sentient beings.