OSI Hopes To Decrease Number of Licenses

Noksagt writes "Various outlets report that the OSI may cut down the increasing number of Open Source licenses. Right now there are about 50 approved licenses; incompatible licenses confuse and impede developers and end users alike. The OSDL has been pushing hard for this at LinuxWorld. Sam Greenblatt, a member of the OSDL board, said 'Eventually there should be three licenses: The GPL, a commercial version of the GPL, and, of course, there will be the BSD because you can't rid of it.'"

Comments

Ein Volk

[pair-a-noyd]

Ein License

Re:Ein Volk

[squiggleslash]

GNU/Linux works as well as it does because developers can freely share code amongst one another. Incompatable licenses impede this. This is why we've seen a growth in the number of "dual licensed" projects, from Mozilla to QT, often (Mozilla, OpenOffice.org, etc) where the original project-specific license was open source, free software, compatable but not GPL compatable.

Yes, different developers have different needs. But much of the cruft that passes for alternative licenses these days ultimately is unnecessary and incompatable with other licenses for the sake of being so. A Solaris user will not, when the OS is opened, be able to include code from Darwin and redistribute the results. An X.org user cannot include parts of XFree86 and redistribute the results. There's little reason for this: Sun and Apple want to distribute closed code and aren't willing to work with each other. Some of XFree86's developers unilaterally decided that the usual copyright attributions weren't credit enough for their work. None of these really have much to do with the type of code being written.

It sucks. It's hard to figure something's "free software" if you're not allowed to include code from other "free software" and still treat it as free software. Incompatable licenses undermine software freedom.

Re:Ein Volk

[sepluv]

No. They are saying that there should be multiple licenses (for the different free-software distribution systems). However, we should also cut-down the number of licenses that are similar to each other by removing licenses that are functionally the same as another commonly-used license.

If your license genuinely embodies another distribution/licensing system for free software (other than copyleft and effective public domaining), I'm sure they won't have a problem adding your license to the list that they encourage people to use (after it is properly checked), but most licenses are just rewordings of old ideas (with a new person/company's name at the top).

The current number of licenses causes confusion (for prospective licensors and licensees); encourages people to write even more licenses (without properly considering alternatives, and without making sure they are legally watertight or make sense); and, worst of all, means that licenses exist which are effectively the same as each other but are incompatible (which discourages the mix-and-match creative commons which is the primary reason for software freedom in the first place).

How can they do this?

[nlinecomputers]

I can create any damn kind of license that I want. What are they going to do. Claim it is not "Open Source" by changing the definition of Open Source. Sure it is confusing but all the different licenses exist because someone finds the GPL or the BSD license doesn't support how they want software to be distributed. Fix people then you can fix this mess.

Re:How can they do this?

[Yobgod Ababua]

What they really want to do is find out what your licensing goals are and encourage you to use a pre-existing license that fits those goals, rather than creating your own possibly incompatible license.

Why they want to do this is because the fewer number of licenses that are in use, the more likely it will be that your software's license is compatible with another piece of software's license... permitting both of you to benefit from each other's work and allowing all of your end-users, in turn, to benefit.

Keeping a small number of licenses also makes it easier for people to regularly review them and make sure that, legally, they are doing what they are intended to do. This can be a much more difficult endeavor than you think, especially when we want licenses to work across international and legal boundaries.

Sure, you can always roll your own, and no one can stop you from offering your copyrighted work under whatever license you choose, but if you pick from a limited set of known and tested licenses, you benefit from knowing that it's solid, you benefit from knowing that it's compatible with all the other software released under the same license, and we all benefit from having fewer 'license glitches' get in the way of what we care about... making better software (some us want to charge for it, some don't, that's not really relevant).

Re:How can they do this?

[JoeBuck]

You're free to write whatever license you want; they are free to refuse to certify it.

The problem with the proliferation of licenses is that you can't mix and match software. Right now there are basically three types of open source (or free software) licenses:

• Copyleft. The GPL fits this category, as do a few other licenses like the OSL, IBM CPL, etc. Some of these exist for no other reason than to have something like the GPL without Stallman's rhetoric, but since the drafters weren't careful to maintain compatibility of conditions, you can't mix and match code.
• Copyleft restricted to a body of code; the code can be linked to other code that uses other licenses or is even proprietary. LGPL, MPL (Mozilla), and a whole host of licenses that differ just slightly from the MPL because some corporate lawyer wanted to fine-tune. Where the licenses conflict, you can still mix and match code as long as you preserve file or library boundaries.
• Non-copyleft (MIT/BSD style licenses).

These licenses differ from each other on technicalities, and on what happens with patents, or because someone wants to tweak a boundary case. Some of them give a privileged position to the original contributor, some don't.

The community would be better off if we could just get down to three basic license choices, and the use of "special exception" clauses where needed. For companies that want special privilege (like the ability to use code plus fixes using other licenses), they can ask for copyright assignment of contributions, and treat contributors well enough that they actually get it.

Re:How can they do this?

[pclminion]

That's fine, and I can refuse to certify theirs back. That doesn't change whether my license, or theirs, is open source.

True, but they can prevent you from hosting your project on SourceForge. SourceForge does not accept projects that don't use OSI-conforming licenses.

Stunts like the one we're discussing right now make it abundantly clear why relying on SourceForge to host the majority of OSS projects was a BAD idea. People come to rely on the services SF.net provides, but the OSI license requirement gives them a stranglehold over you.

Re:How can they do this?

[argent]

What they really want to do is find out what your licensing goals are and encourage you to use a pre-existing license that fits those goals, rather than creating your own possibly incompatible license.

That's great , but that's not going to get anywhere NEAR three licenses. At the very least you'll need:

1. Something like QMAIL's license.

2. The Aladdin license.

3. The two main forks of the BSD license.

4. The GPL.

5. The LGPL.

6. The non-transitive GPL-alikes.

7. No commercial use variations.

That's just off the top of my head.

LGPL?

[Sim9]

I'll admit, I'm not quite sure what the "Commercial GPL" is, but I really hope that LGPL isn't eliminated. [The LGPL allows users to use a library, and not release your code that uses the library. Changes to the library source itself must be released].

Let's say I have a write a game that uses the popular library, LibSDL (a rendering library). Though open-source may be great, why should I be *forced* to GPL my game code, which has little to do with LibSDL development?

Re:LGPL?

[JoeBuck]

My guess was that the LGPL was what the guy was thinking of by "commercial GPL".

Amusing

[bonch]

Though I understand the ideas behind all these licenses, it occurs to me how amusing it is that if something was truly 100% free, it wouldn't have or need a license at all. BSD comes closest to that.

Re:Amusing

[pclminion]

The BSD license does not restrict your freedom to see code which was already in the open. It simply makes it possible for people to make their OWN modifications private.

You make it sound like the BSD license could be used to close a previous open piece of software. That's impossible. BSD license simply gives you more powerful rights over your own modifications to that software. Some people see that as a flaw, others (including myself) don't see what the big deal is about allowing other people to profit as long as it doesn't restrict our own rights to use the code we've written.

Re:Amusing

[mpcooke3]

I don't know if you comment is a troll or not.

It's fairly well established that some people believe something is more free if it has a license that restricts users ability to make versions of the software non-free whilst some people believe that software is more free if you have the right to make non-free versions.

I think regardless of how you define "free" both the BSD and GPL style licenses have different purposes.

When you say that if something was 100% free it wouldn't need a license that might be true if the world had no laws or commercial interests. That extra waft in the GPL that makes it longer than the BSD license is to make it clear that the software can't be moved from the category of "free" software to the category of "non-free" software by commercial interests.

Imagine another world (as Stallman problably does) where the law by default rather than supporting commercial interests supported freedom of software. In this world the GPL would be short and the BSD license long because the BSD license would need to explain that future versions of the software could be taken by private companies and changes withheld unlike "normal software" where future versions of the free software must remain free by default.

Matt.

Re:4 Licenses, not 3

[k98sven]

He was talking about software licenses. Not licenses in general.

Uhhh.. How does this impact SourceForge?

[pclminion]

In order to host your project on SourceForge it must use an OSI-conforming license. If the list of OSI-conforming licenses is drastically reduced what will happen to all the projects on SourceForge which don't use the GPL or BSD licenses? Will they just be booted off the server? Forced to switch licenses?

Ah, standards.

[Tackhead]

> Right now there are about 50 approved licenses; incompatible licenses confuse and impede developers and end users alike.

But that's what's so wonderful about standards. There are so many to choose from. Besides, if you really have a problem with a certain license, you should have the right to view, modifiy, and release your own license based on the work of those who've written licenses before you.

Join us now and share the licenses!
You'll be free, lawyers, you'll be free!
You'll be free, lawyers, you'll be free!

Lobbyists get piles of money,
That is true, lawyers, that is true.
But they cannot help their neighbors;
That's not good, lawyers, that's not good.

When we have enough licenses
At our call, lawyers, at our call,
We'll throw out that dirty li*cough*software,
Ever more, lawyers, ever more.

Sorry, RMS, I had to. The muse knows what it wants, even if it wants to give me a first-class ticket to hell with window seating.

Yes. That's exactly why they'll do

[sterno]

What they are doing is branding the term "Open Source" and this will not change the meaning of "open source" (note small "o" and small "s"). One of the big problems in software licensing in general is that every license is different in subtle or sometimes huge ways. If you want to do any sort of development that involves integration of pieces of other software, it can get quite complicated quickly.

Does this mean that you can't make your own license? Of course not. What it means is that if you want their official seal of approval, you likely won't get it.

I think 3 licenses might pass as a sort of Platonic ideal, but I can't really see that covering all needs in the real world. However, establishing a base line of a few simple licenses could make life much easier for smaller developers that don't really have an interest in paying a lawyer to craft them something more complex.

Re:Commercial GPL

[Ark42]

Commercial GPL is something like the NPL/MPL licenese that let Netscape keep some code to themselves while still sharing and modifying the Mozilla codebase, I think.

Of course, I see a clear need for LGPL as well here, since that is different than just GPL or MPL or BSD, and very useful indeed.
LGPL is what lets me use things like libPNG or ZLib in my commercial application without giving away the unrelated source code to my entire program. LGPL is a good thing if you value PNG support in other programs that aren't going to be using GPL themselves.

We African Greys...

[African Grey]

prefer the BSD license. It's not that we don't want to publish the changes we make to source code, but a lot of parrots don't have decent net access, especially in Africa. IP over avian carrier just isn't very fast, and source is a lot bigger than binaries. If you humans would get your act together and bring some kind of decent connectivity to the jungles, you'd be able to see our coding prowess for yourselves!

Here's a different idea...

[StevenMaurer]

I think they're going to run into the same problem that DRM manufacturers have: there's no benefit to the people who are untimately in charge.

In this case, it isn't the 'paying customers', it's the developing free software engineers. The proliferation of licenses comes directly from the fact that developers have found some aspect of the GPL or LGPL to be too onerous to release under. And there is no way you're going to get them to alter their license just because Stallman thinks they should.

So here's a different idea. Instead of trying to reduce the number of Open Source licenses, people should instead come up with a comparison chart. Much like the Unix rosetta stone except for legalese, identifying general contract features in common (or different) between them.

That way developers can see the difference in a single place, and pick the best license for their particular purpose.

Put away your crack pipe

[QuantumG]

Maybe it's time for the "Open Source" movement to die. After all, the founders of this movement (Eric S. Raymond, Bruce Parens, I'm lookin' at you) havn't had anything official to say about Open Source in a while (oh wait, there was that Java thing, you're ok Bruce). I thought "to reduce confusion" was what the Open Source movement set out to achieve, being that Free Software just wasn't straight forward enough for them. The result of this mess has been one person after another putting the "openness" of the source code ahead of the freedom to modify and redistribute the source code (yes, Microsoft, Sun, X11, Apache, and that worm who wrote the packet filter the OpenBSD project rewrote in a week). It's amazing to me the number of people who have no problem understanding exactly what I'm talking about when I say Free Software, compared to the number of people who are now confused about Open Source. Maybe it's the use of capital letters. Ahh, what irony that is, we could have avoided endless debates about Free Software vs Open Source if we'd just capitalized "free".

Commercial != Proprietary

[jbn-o]

Sam Greenblatt, a member of the OSDL board, was quoted as saying something very unclear: "Eventually there should be three licenses: The GPL, a commercial version of the GPL...". The GNU General Public License (GPL) allows one to distribute copies of covered works for a fee. Many people have turned GCC (the GNU Compiler Collection), one noteworthy GPL-covered program, into a commercial work by distributing copies of it for a fee, some have also based for-hire consulting services on GCC. These consultants develop GCC as a business activity.

Most of the time when people say "commercial" in this context, they don't mean that. That word was just a poor choice which may stem from not fully understanding what software freedom entails. What they really meant to say was "proprietary", which is something different. In this case, I don't know what that other meaning would be; a proprietary GPL would not be the GPL, it would be a perverse opposite of what the GPL stands for and accomplished long before the open source movement existed. Thus I'm left thinking Greenblatt's statement is at best unclear, non-sensical at worst.

Been there, done that.

[sleepingsquirrel]

So here's a different idea. Instead of trying to reduce the number of Open Source licenses, people should instead come up with a comparison chart. Much like the Unix rosetta stone except for legalese, identifying general contract features in common (or different) between them.

You looking for something like this perhaps?

OSS is about choice

[lakeland]

... and you really can't stop people from making bad choices.

Since editors are overused as an example, lets try CD burners. There are two that most people will know: k3b and nautilus. Yet a quick search on freshmeat will return literally dozens of CD burners. Why did those authors write a CD burner when excellent ones already existed? Maybe for experience, maybe due to a missing feature... it doesn't matter. The point is they can, so they will.

Choosing an open-source licence is the same: There are a couple basic smart choices, but there is no way you're going to get everybody to agree to only use them. As a random example, one of the programs I use is only free if the kernel of the computer you run it on is open source, weird huh? It is the OSI's job to try and simplify things as much as possible so people can understand what's going on. Sure, they can discourage wacky choices, but they shouldn't be outlawing them from the OSS definition.

PS: A google for licen{s,c}e returns the GPL as the number one hit.

Wait this sounds like...

[xgamer04]

Three Licenses for the GNU-kings under the sky,
Seven for the Corporate-lords in their halls of stone,
Nine for Hacker Men doomed to die,
One for Bruce Perens and Eric Raymond on their dark throne.
In the Land of CVS where the Versions lie.
One License to rule them all, One License to find them,
One License to bring them all and in the darkness bind() them.
In the Land of CVS where the Versions lie.

Sorry, had to be done. :)

Re:Commercial GPL

[Samrobb]

GPL has an important restriction -- you MUST give your changes back to the community if you want to distribute them.

<disclaimer type="IANAL">

Nope. The GPL says nothing about giving changes back to some "community". It does say that when you distribute a binary to someone, you must make the source code for that binary avilable to them (either as part of the distribution of the binary, or on request within a limited period of time.)

Practially speaking, this does mean that changes can and do make it back to "the community" in a lot of instances. However, it isn't required by the GPL.

To use your own analogy - the GPL doesn't force you to put your changes on the village green. It allows you to put them on the village green, and it also allows anyone who receives them from you (directly or indirectly) to put them on the village green. That allowance greatly increases the chance that someone will put your changes on the village green, but it doesn't make it a requirement.

</disclaimer>