LWN.net Logo

OSI approves two new licenses

From:  Russell Nelson <nelson@crynwr.com>
To:  lwn@lwn.net
Subject:  OSI approves two new licenses
Date:  Fri, 27 Sep 2002 12:26:18 -0400 (EDT)

The Board of Directors of the Open Source Initiative (OSI)
approved two licenses Thursday.  These licenses, written by Larry Rosen,
attorney for OSI, are intended to be encapsulations of the best of
their class of open source licenses.

One approval is a technical revision of the Academic Free License.
This class of license puts very few restrictions on use of the code.  It's
designed to be a compatible upgrade for legacy academic-style licenses like
MIT and BSD. 

The other is a completely new license, the Open Software License (OSL).
First introduced at the O'Reilly Open Source Convention in July, the
OSL is a reciprocal license, just like the GNU General Public License (GPL).
This class of licenses requires a reciprocal sharing of code.  
You get, you give.

-- 
-russ nelson              http://russnelson.com |
Crynwr sells support for free software  | PGPok | businesses persuade
521 Pleasant Valley Rd. | +1 315 268 1925 voice | governments coerce
Potsdam, NY 13676-3213  | +1 315 268 9201 FAX   |


(Log in to post comments)

OSI approves two new licenses

Posted Sep 27, 2002 23:37 UTC (Fri) by JoeBuck (subscriber, #2330) [Link]

The OSL seems almost identical in effect to the GPL, but there are marginal differences. Section 8, for example, repeats RMS's original trick (distributing or modifying implies acceptance) almost exactly. It appears to be GPL-incompatible, though IANAL.

The OSL may be tougher than the GPL: the GPL says it applies to code linked into the same executable, but the OSL says that any derivative work must be OSLed; there is no "mere aggregation" exemption. That might mean I can't put something into the same tarball as an OSL work unless I want to OSL it, but again, IANAL. The part about using the legal jurisdiction of the Licensor may be a problem for a derivative work, as then any trial would have to occur in two different places! The issue of how to handle a piece of software, like the Linux kernel, whose copyright owners live in many disctinct jurisdictions is not addressed.

Is the intent just to have something that works like the GPL but doesn't have Richard Stallman's rhetoric in it?

Random thoughts / exegesis on the new licenses

Posted Sep 27, 2002 23:54 UTC (Fri) by Peter (guest, #1127) [Link]

Interesting. Random thoughts on the OSL:

  • Explicit license for any patent needed to use the software - nice. Prevents someone from distributing software under OSL and then ambushing the public later with patents.
  • Unlike the GPL, licensees do not need to redistribute source code - it appears to be sufficient to post a pointer to a third-party web site. No "offer good for three years" necessary, either. Note, however, that if you have modified the distribution in any way, your copy is a distributed work and you do have to provide your own copy of the source code for your licensees. Again, though, a web site is sufficient - no need to send them a CD-ROM, or an offer to provide a CD-ROM, or anything.
  • Section 5: "external deployment". Interesting in that if you have, say, a web service using OSL-ed components, you have to provide pointers to the source to your web clients. If you have modified the web service, you have to provide the source directly (see previous point).
  • Unlike the GPL, the OSL does not attempt to define "derivative works", leaving that to copyright law. I'm not sure this is a good idea. I don't think there is much if any case law on things like "if you dynamically link to a library, is your binary a derivative work of the library?" (The GPL would assert "yes", but since you aren't actually distributing the library, many people would say "no".) So it's unclear what you are or are not allowed to do under the OSL. Of course, many people are confused by the GPL as well, so perhaps this isn't really tractable.
  • I am a bit confused by the distinction between "display the work publicly / perform the work publicly" (section 1d and 1e) and "external deployment" (section 5). Section 5 invokes the viral clause (section 1c) whereas sections 1d and 1e do not ... but don't they really cover the same activities?

Random thoughts on the AFL:

  • It has the dreaded BSD advertising clause...
  • If you redistribute the original work you have to "reproduce all copyright notices ... both in the Original Work and in any documentation and/or other materials provided". However, if you create a derived work, it seems the above does not apply. (The advertising clause does still apply.) Looks like a loophole - is it intentional, I wonder.

Finally, the "mutual termination for patent action" paragraph appears in both licenses and it is particularly interesting. Basically, the license terminates if you file any lawsuit anywhere against anyone related to patent infringement by any open-source software that has this clause in its license. So if you write a program that encodes GIF files, for example, and distribute it under one of the two new licenses, and Unisys sues you, they are automatically prohibited from using any software under either of those two licenses, no matter who wrote it or what it does.

Random thoughts / exegesis on the new licenses

Posted Sep 28, 2002 9:15 UTC (Sat) by lordsutch (subscriber, #53) [Link]

Actually, this isn't the "dreaded" BSD advertising clause; that clause requires the acknowledgement in all advertising, hence all the crap the *BSD people had to put on their CD covers until the UC Regents struck the clause from the license.

Even the current (3-clause) BSD license has the (paraphrased) "you may not use the name of XXX to endorse your product without our consent" clause, which isn't a PITA.

Random thoughts / exegesis on the new licenses

Posted Sep 28, 2002 10:51 UTC (Sat) by Peter (guest, #1127) [Link]

Actually, this isn't the "dreaded" BSD advertising clause

My mistake. You are correct, milordsutch.

Re: Random thoughts...

Posted Sep 29, 2002 21:35 UTC (Sun) by Phaeton (guest, #3579) [Link]

Okay, what I do not understand is why we need the AFL.

The BSD license is doing well, in fact there are some
operating system distributions who use exactly that
license.

It has the big pro that it's short, understandable for
all those who'd say: IANAL, and basically goes down
smoothly.

You don't try to gulp hooks like with the new AFL, where
you'd consult a lawyer first to find out whether you really
should be using that license.

For the OSL, it's even longer than the AFL. *shriek*
Do people really think users and most 3rd party developers
actually read such manifests ?

Re: Random thoughts...

Posted Sep 29, 2002 23:14 UTC (Sun) by lordsutch (subscriber, #53) [Link]

I think the point of the two OSI licenses is to stop the proliferation of mutually-incompatible licenses that have patent clauses; i.e., so you can take code from e.g. Sun and IBM and combine it under the same license, rather than dealing with the fact 2 teams of lawyers were involved (thus creating two incompatible license texts).

I don't know if this is likely to be successful, but it is worth trying.

OSI approves two new licenses

Posted Sep 28, 2002 4:23 UTC (Sat) by frazier (subscriber, #3060) [Link]

I fail to see why we need a GPL-like license. Even if the new one is slightly better in whatever nuances, if it is not GPL-compatible (haven't read through it and even if I did, I avoid legal anything when possible), that makes it far worse than whatever improvements it may bring thanks to license conflicts.

OSI approves two new licenses

Posted Sep 28, 2002 12:21 UTC (Sat) by rknop (guest, #66) [Link]

I fail to see why we need a GPL-like license. Even if the new one is slightly better in whatever nuances, if it is not GPL-compatible

Indeed, this is a huge problem. The GPL and the OSL both require that derivative works be distributed under the terms of that licence. As such, it now becomes illegal to combine GPL and OSL code together.

Pain in the butt. Things would be so much easier of the OSI just adopted the GPL (the most popular Free Software licence) as the fiducial "give back to the community" free licence.

What does the OSL solve that the GPL doesn't already solve? Unless there's something significant, the OSI is shooting us all in the foot by putting forward this new licence.

I can only see a couple of possibilities. (1) The patent clause, which is trying to slap people who abuse the patent system. (2) The "People hate the FSF" problem. The FSF has gotten an increasingly bad name as a result of being (depending on your point of view) the "moral center" or the "uncompromising extremists" of Free Software. The GPL is associated with the FSF, and so therefore suffers from "guilt by association". There are probably some afraid to touch the GPL because they think that they may tacitly be agreeing with something Richard Stallman said that somebody else objected to-- and people always want to avoid too much controversy.

(Indeed, I think that 80% of the bad taste that the FSF has left with many would go away if they would just get a clue and realize that they lost the GNU/Linux naming battle years ago, and stop making so much noise and trouble about it. Disclaimer: I agree with them, they're right. On the other hand, they did lose that battle long ago, and it's not important enough to keep fighting. All it does now is divide the community and make them look like whiners. There are too many very crucial battles out there, many of which are in nearly as much danger of being lost, that it just is tacitcally unwise get hung up over this minor issue.)

-Rob

OSI approves two new licenses

Posted Sep 30, 2002 21:27 UTC (Mon) by Peter (guest, #1127) [Link]

As such, it now becomes illegal to combine GPL and OSL code together.

Agreed, especially given the huge corpus of GPL-licensed stuff. Next year are we going to see Mozilla go for a world record by quadruple-licensing their codebase? (:

I can only see a couple of possibilities. (1) The patent clause, which is trying to slap people who abuse the patent system.

This is what I like best about both new licenses, actually.

(2) The "People hate the FSF" problem.

...which is one specific instance of the more general "Not Invented Here" problem.

(3) Complexity. The GPLv2 is, what, 17 kilobytes? It includes a lot of explanatory text, which makes it quite easy to read, but the length alone will guarantee that most people don't really understand it. (Anyone who has ever participated in any interactive forum concerning the GPL will know that this not hypothetical.) The new OSL is a lot shorter, yet includes most of the same conditions. That is a Good Thing. Of course, it leaves out some of the more esoteric conditions, such as "offer good for at least three years", but that is probably a Good Thing too.

Myself, I will continue to use the GPLv2, mainly because of "license compatibility". Hey, I like being able to link to libreadline, y'know? (That being the foremost example of a GPL'd product you might expect to be LGPL'd.)

Too much licenses will be kill Free Software

Posted Sep 28, 2002 6:38 UTC (Sat) by adulau (guest, #1131) [Link]

Complexity is the evil in Computer Science...and also in the legal area. So why to create new licenses ? The GNU GPL (version 3 will solve some issue, like redistribution) and Berkeley style is enough.

Too much "Free Software" licenses is not good at all.

Too much licenses will be kill Free Software

Posted Sep 29, 2002 21:45 UTC (Sun) by Gnulix (guest, #1399) [Link]

So why to create new licenses ? The GNU GPL

Perhaps because some people are unwilling to support the FSF, directly or indirectly... Although I like the idea of my software being available for anyone to use and improve - I don't like the idea of RMS being able to hijack and/or take credit for my work, just because I've used the FSF's license or some GNU tools during my development. Everything released under the GPL isn't part of the GNU project, no matter what RMS claims...

Too much licenses will be kill Free Software

Posted Sep 30, 2002 4:03 UTC (Mon) by jamesh (subscriber, #1159) [Link]

RMS doesn't claim that all software covered by the GPL is part of the GNU project. I don't think I have ever heard him claim that.

The use of FSF licenses or FSF copyrighted build tools also does not affect your ownership of your work. In cases where the GPL may be a bit unclear (such as for gcc), clarrifications to the licenses are included with the package to make the intentions clear.

Too much licenses will be kill Free Software

Posted Sep 30, 2002 21:45 UTC (Mon) by Peter (guest, #1127) [Link]

The use of FSF licenses or FSF copyrighted build tools also does not affect your ownership of your work. In cases where the GPL may be a bit unclear (such as for gcc), clarrifications to the licenses are included with the package to make the intentions clear.

I wish there were some way to make this point clearer. There are company lawyers out their advising their employers to steer clear of the GNU development tools for exactly this reason - just to be on the safe side, perhaps.

I have looked at the gcc source just to verify the matter, and sure enough, libgcc1.c and libgcc2.c (the parts you would link into your program) explicitly give you unlimited distribution rights. Similar provisions are made for things like the Bison parser skeletons and the autoconf support files. It is obviously the intent of the developers not to trap you by the use of their tools (which I would consider a bait & switch tactic). But I can understand how a lawyer, who does not have the necessary background to know which bits of a program are included in its output, would be reluctant to state unequivocally that these tools are indeed "safe from GPL-pollution".

It would be nice if these tools would come with an obvious and short README that details the effort that has been made to prevent such "pollution" - and directs readers to the specific source files affected. Then, at least, a lawyer can quickly see the intent of the developer, and, at the very least, can be confident of a "bad faith" defense.

Copyright © 2002, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds