// Patrick Louis

Licenses on Unix

Licenses on Unix

(Transcript of the podcast)

Intro

The world of licenses is the legal world, a world where the literal meaning of words is important and where all the crevasses are exploited.

I’m not a lawyer, nor have I studied laws, and whatever I say will be based on what I understood from my reading.

In this episode we’re going to do a small overview of the topic of licenses on Unix. But beware, a “small” overview in the legal world is quite heavy!

What’s a license

Let’s first define what’s a software license.

From wikipedia:

A software license is a legal instrument (usually by way of contract law, with or without printed material) governing the use or redistribution of software.

The software license can be distributed on paper or in digital form and usually contains rules granting or removing rights, imposing restrictions, that the users of the software should agree to or otherwise would commit copyright or license infringement. Licenses can also contain some lines about liability and responsibility from both party, the software provider and the users.

In sum it’s a contract.

Legality falls under the laws of the country and so they might vary from one place to the other. All countries can have their own jurisdictions.

For instance the Marshall Islands have 0 copyright laws.

Beyond that, there are some international treaties that are signed by some countries to mutually respect copyrighted materials such as the treaty of the Berne convention and the World Intellectual Property Organization.

Let’s not discuss too much the legality and more about some interesting stories and stuffs related to Unix that you might fancy.

The first licenses used on Unix-like systems - History

In the 50s till the 70s-80s it was common for computer programs to be shared with their source code, it was the norm. Everyone had the permission and the ability to modify the software for their specific usage.

There were even organizations of users and suppliers that got formed around sharing the softwares such as SHARE and DECUS.

This changed around the end of the 60s when the competition in the software world got tighter. Usually it was the hardware manufacturer that shipped the software with the hardware it was producing but independent software distributors started selling the software untied to the hardware.

A court case filled in 1969, started in 1975, and finally dropped in 1982: US vs IBM, stated that bundling hardware with software was anticompetitive and brought monopoly to the marketplace (Tell that to the DRMs of today).

And so some of the software industry of the 70s and early 80s began distributing binaries and proprietary packages - Taking measures to limit the copying and spreading of their softwares so that they could keep their marketplace.

While the software is distributed this way, as binaries that is, it cannot be studied nor changed.

Unix wise, before 1975, all the Research Unix that Bell Labs produced were for internal use only.

It was only in 1975 that the first source license for UNIX V5 was sold to the University of Illinois Department of Computer Science, UIUC.

This was a great move for the university of Illinois because they started working on implementing ARPANET on those Unix machines, creating the RFC 681, for NETWORK UNIX, with FTP and telnet clients. The first Unix later to included them was BSD.

At the time, Bell Labs wanted to sell licenses but for non-university users it wanted to sell them at $20K which was very expensive.

The V6 version of UNIX did turn that around and was the first commercially successful UNIX. It remained the most widely used version into the 80s.

V7 superseded it later in 1978, but V6 still remained the most used.

As we’ve said, at those time the source code was available to anyone who got the software, and so it was a great tool to teach in universities.

The license they provided with the software didn’t explicitly said anything special about sharing the source, licenses for softwares didn’t mean much yet.

One of the university who got hold of UNIX V5 in 1974 was the university of Berkely. Now you might wonder, “that’s before 1975”, yes it is, because no license was sold to Berkeley, at least not yet. It all happened because a teacher named Bob Fabry was teaching a course about OS principles and took UNIX as an example. It was more like an informal notice written by the developers themselves.

One year later, Ken Thompson, one of the creator of UNIX, visited the university and helped install UNIX V6 on the machines they had.

Some students got quite found of the system and began improving it and adding softwares they liked. Bill Joy was one of those students.

Because other universities became interested in the work they were doing with UNIX at Berkeley, in 1977, Bill Joy and friends compiled a distribution of all the addons over V6 and distributed it under the name BSD: Berkeley Software Distribution.

Finally in 1980 the US established in its law that computer softwares were copyrightable. Before that softwares were not considered nor intellectual property nor copyrightable, they were open ideas for everyone.

Actually, this process didn’t really start in 1980 because in 1976 there were already copyright acts related to softwares it was just that they weren’t all agreed on yet. Let’s say it started in 1976 and got settled in 1980 because they wanted to make sure that softwares embodied an “author’s original creation”

So from this point onwards you could claim that a software was yours instead of distributing binaries or do nasty tricks to deter your users from sharing it. Now you have the laws behind you, at least in the US.

In 1983, a hacker at the MIT AI lab, Richard Stallman was a bit frustrated with how the software industry and users were closing themselves. And so he announced a project called GNU, which aim was to release a free operating system. In 1985, he released the GNU Manifesto, which was an explanation of the GNU philosophy, the base definition of free software, and the “copyleft” ideas.

But still in the 80s UNIX was the corporate norm, it was what organisations liked.

And what do organisations do when something becomes popular, they try to regularize it.

It needed regulations because everyone was customizing their own versions and brand of Unix-like system. It was creating too many rivalries and doing more harms than good.

There were mainly two groups that appeared trying to do that:

  • In 1984 there was the X/Open Company, which was a consortium of several European UNIX systems manufacturers. Their aim was to create a single specifications for all UNIX. The members were big players.

  • In 1988 there was the OSF, Open Software Foundation, a not-for-profit organization in the U.S. also to create open standard for implementing UNIX systems.

The two of them joined way later in 1996 to form the Open Group, with 500 members. They hold a sort of public approved certificate saying what is UNIX and what isn’t, the term is trademarked by them.

Anything that has the term UNIX is licensed under their system to have this name, which comes with royalty fees.

And this Open Group brand fee isn’t cheap…

To register your product to the trademark license you have to pay at least $2.5K plus a $1K annual fee and more payments for a bunch of other things.

It’s good to know that in 1997, the Open Group got hold of the responsibility of the X Window System too.

Now back in time, in 1989, after all the legislation and standardisation happening, the guys at Berkeley were still distributing their softwares, the BSD, it had already reached version 4.3

However, they were paying fees to AT&T, the parent company to the Bell lab, that got very expensive.

Yep, mobile companies are all tycoons.

Now if you remember the University of Illinois made the specs for Unix networking and it got implemented first in BSD. Many people requested this networking code without the hassle of licensing and so that’s what the BSD did, they crafted a custom license - the BSD license - to distribute their Net/1 code.

The BSD license is very permissive and allows the source code to be modified, reused and to be sold, which was perfect for the industry.

In 1991, they got fed up with AT&T and rewrote all the standard Unix utilities under the BSD license so that they would not be tied to any copyright.

They named it 386BSD.

But you don’t get away from a titan without a scar, AT&T sued them for infringing copyrighted material, they filled the lawsuit in 1992: USL v. BSDi.

This dealt a blow to the BSD communities and slowed their development.

BSD wasn’t a good choice for corporates now.

A year before that lawsuit, in 1991, a student named Linus Torvalds was working on a kernel named Linux, a UNIX clone, which he decided to release as free under the license Stallman created, the GPL - GNU General Public License. Together joined with the userland utilities that the GNU project had it created the GNU+Linux operating system.

It had no ambiguity with the license and attracted the attention of volunteer programmers and spread because of the open community and sharing of ideas, much like the first days of Unix with university students.

Then, one year after it had started, in 1993, the case USL v. BSDi lawsuit was settled out of court and the FreeBSD and NetBSD projects emerged.

Once the BSD code base was released it got widely adopted as the base of many proprietary Unix variants, as it did before, all of that because of how easy and permissive the BSD license is with commercial use.

Still today several operating systems choose to use softwares under the BSD license for the same reasons, macOS and iOS being two well known examples.

In 1995, the Apache HTTP Server, commonly referred to as Apache, was released under the Apache License 1.0 which is also a permissive BSD-like license that allows commercial use.

It was used to undergird the advent of the world wide web.

In 1997, Eric Raymond published The Cathedral and the Bazaar, which was more or less putting into words what the hacker community loved about in the sharing of softwares.

This book was one factor that incited Netscape to release their internet suite software under a free license. Which is why we have the whole Mozilla ecosystem today, and a key event that influenced others to release free and open source softwares after that.

The Mozilla license is GPL compatible, very permissive but at the same time doesn’t allow to change the software without sharing the source code, this is the idea of “copyleft” that the GNU project envisioned.

In 2002 a company named Santa Cruz Operation, SCO, released the source code of some Ancient UNIX relicensed under a BSD license.

SCO was the “first Unix company”, founded in 1979 and filled bankruptcy in 2007, it used to port many UNIX softwares to many platforms.

One fun use of that release was on the Nintentdo Game Boy Advance which was made to run V5 UNIX using a PDP-11 emulator.

It is also a great way to learn about operating system theories.

…And this is where we are now.

You might be questioning yourself a lot about licenses.

Just like BSD has relicensed itself, and SCO relicensed their UNIX, how is relicensing possible? What do those licenses really mean? What are the ones used today? What should I know?

Licenses today

First Thoughts

According to the Free software foundation there are only two big categories of software licenses, they are either proprietary or free and open source.

That’s a very narrow view of the matter, a black and white fallacy, as we’ll delve into later on.

But for the moment let’s just presume it is true.

According to this entity’s chosen distinction, the difference between a free and open source software and a proprietary one lies in the rights granted to the users.

A free and open source license being one that allows both modifying and reusing the software, which as a premise implies that the source will be somehow available.

Then, again according to their distinction, a proprietary software is everything else.

But what about softwares without licenses, an unlicensed software.

Is it outside of the copyright protection, is it in the public domain?
What if that software is an internal trade-secret software used inside the company?

There are no correct answer to this, it depends first on the country and second on the court choice during the legal appeal.

In sum: it’s complicated.

In the US, a software distributed without license is fully copyright protected and thus you can’t legally use it until the right has been granted.

That’s as long as there is an owner to the software, which is usually before the copyright expires. Some countries like the United States have a limited lifetime for copyright. When the copyright or patent expires it falls under the public domain.

A patent being a right to a certain type of invention, which is again a vague term. There are other legal mechanism too such as legislation and trademark.

What about softwares distributed on Github without license, you may ask. Well, yes, beware, that this also applies here.

Under most international treaties, such as the Berne convention, an author automatically obtains the exclusive copyright to anything they have written, even rights over derivative works.

So remember to always add a license to your projects when pushing them.

However it’s also not as simple as that, someone using a software without license can win a court appeal and the copyright can be waived away. The legal border is thin because releasing anything without license is not like releasing with a license.

The only true way to relinquish a software into the public domain is to do it explicitly via a license such as the CC0, creative common zero or the WTFPL, what the fuck public license.

That way you’re assured to have completely lost ownership, trademark, and patent to the software.

You must be explicit about whatever you do so that you stay in accordance with the country’s laws.

In a license

So what’s inside a license, what varies from one to the other?

Licenses are contracts that or extend the legislation of the citizen of a certain country over the software or widen their rights over the software - rights that would otherwise be disallowed by default copyright.

So yeah, you’ve got the base laws of the country and the license over it.

Many licenses only enumerate those basics rights that the user already has under law. But keep in mind that the rights you have as a citizen cannot be taken away if they conflict with a license, that’s simply not possible.

Free software licenses are said to be more or less non-contractual extensions as they aren’t an agreement between parties but that remains to be contested as they are still an agreement. We’ll see later on in this podcast, someone has to be held accountable for infringement.

For instance, citizens in the US that owns a software have the explicit right to use the software with a computer even if that software accidentally makes copy of itself. You are legally entitled to own copies of a software you own.

There’s also a first-sale doctrine in the US that grants middleman, resellers the right if they own something to resell or rent it even if they are copyrighted material as long as they own it, which is not the case with proprietary software. This is called “distribution right”. However it’s vague how it applies to digital goods as an item can be copied indefinitely.

In the European Union the laws differ where a copyright holder cannot oppose the resale of a digitally sold software.

Where that is more of a play-on-words is when an end-user isn’t the owner of the software. That is the precise definition of a proprietary software, a software you are borrowing and not owning.

The software technology might be patented for instance.

The intellectual property rights remain with the entity that provided the software and thus a proprietary software doesn’t give you the same right as a citizen as a software you own.

In that case the user has to sign an EULA, end-user license agreement, with the digital good owner to be able to use it, otherwise without accepting the license the user may not be able to use the software at all.

An EULA is yet another contract between the end-user and the licensor that establishes what rights the user has.

Most of the time it is about liability, holding the software distributor non-liable for damage or the user liable for infringement, and restrictions, which can range from forbidding reverse engineering, benchmarking, to limiting the number of users and the number of devices the software can be used on, etc..

Free softwares also have EULA that is part of the license, we’ll see those in a bit.

A nasty way of including the EULA is to do it in a shrink-wrap manner. The term “shrink-wrap” means that the EULA cannot be read before the product is actually bought. Which means that if the users don’t accept the license they won’t be able to use the software and would have paid for nothing.

So what’s in a license really?

  • The basic rights of your country
  • Who owns it
  • What rights and restrictions the end-user has such as modifying, reselling, benchmarking, sharing it
  • Who is held liable
  • What responsibility the owner has
  • What responsibility the user has

The scale

Depending on the way those previously mentioned liabilities, rights, and restrictions are set the license is said to be more open and free or less open and free.

This is a better realistic approach than the mundane dichotomy the free software foundation discusses.

There’s actually a lot of nagging over the definition of what is truly “free software”.

The arguments are about those: how permissive the license is, the requirement for redistribution, if the license is copyleft/share-alike or not, if the license allows derivative work, if the license allows selling the software or using it for advertisement, etc..

We often hear about “Free and open source software” which is a actually a vague umbrella term for softwares that provide liberty to their control and are open source.

It’s truly confusing because the distinction between open source and free software is nebulous, it’s more philosophical than practical because they describe almost the same things.

Depending on who defines them they might mean different things. It’s a philosophical war and using “Free and open source software” doesn’t make anyone cry and that is why it’s used a majority of the time instead of simply “open source” or “free software”.

Some examples organizations and groups that put those definitions together are the free software foundation, FSF, the OSI, open source initiative, the Debian project, the BSD project, etc..

They all have different guidelines and definitions for “free and open source” software licenses.

But generically they all allow to use the code for any purpose, modify it, and redistribute it, which is normally forbidden by default copyright.

What they also have in common is how they’re perceived by the communities using it for how they decrease costs, increase security and stability, privacy, and control.

Free and open source softwares often live surrounded by a community of contributors.

Now let’s talk about some of the philosophies.

The Open Source Initiative is an organization that derives it’s license norms from the Debian Free Software Guidelines.

They hold a list of licenses that they “certify” open-source.

Their definition is:

An open-source license is a type of license for computer software and other products that allows the source code, blueprint or design to be used, modified and/or shared under defined terms and conditions.

Most of the other popular free and open source licenses are approved as open source, for instance:

  • Apache License 2.0
  • BSD 3-Clause “New” or “Revised” license
  • BSD 2-Clause “Simplified” or “FreeBSD” license
  • GNU General Public License (GPL)
  • GNU Library or “Lesser” General Public License (LGPL)
  • MIT license
  • Mozilla Public License 2.0
  • Common Development and Distribution License
  • Eclipse Public License

On the side of the FSF, which also holds a list of licenses agreeing to their views, they perceive everything that isn’t in their definition of free software as an unfair shadow of what they are trying to achieve by bringing liberty in the software world.

They define free softwares in terms of liberty and not price.

For them liberty is based around community sharing and respect of individuals, which lies, in practicality, inside licenses as the concept of copyleft aka share-alike.

A copyleft is an additional clauses that requires users who modify the code to also redistribute the modifications under the same license as the original software.

This way it protects all derivative works from drifting and losing the original permissions and be used in proprietary programs.

Now what about the BSD side.

The BSD camp has yet another position on licensing.

They despise copyleft licenses and dismiss them as overly complicated and restrictive.

It is true that copyleft licenses has an insane amount of clauses in their licenses compared to the usual 4 that you find in the BSD license family.

The BSD license is simple and permissive, it’s simply about acknowledging that there was an original author and to keep this license so that the author is credited for his or her work.

Some even consider the term “permissive license” is synonym with anything that is a BSD license.

The BSD license family is liked because it can be use in proprietary softwares, as we’ve mentioned before.

The supporters of the BSD license argue that it’s more free than copyleft licenses because it grants the right to do anything with the source as long as there is attribution to the authors.

But then according to this definition the public domain is more free than the BSD license.

In the end it’s all philosophy.

License compatibility

Because the BSD license only requires you to give credits to the authors without imposing any limitation on the usage it can be used in proprietary softwares, as we mentioned before.

That means the license can be modified by adding new clauses to it as long as they don’t conflict with the previous clauses.

For instance MATLAB is based on MathWorks which is distributed under the BSD license but has an extra clause limiting it’s usage to the proprietary MATLAB which renders all contributions useless outside of MATLAB.

BSD licensed code can also be included in copylefted softwares by added the clauses related to copyleft. That’s what we refer to as compatible licenses.

The FSF indeed has the BSD license under its list of licenses that are considered free.

But it doesn’t work the other way around, code under GPL cannot be relicensed under a BSD license without the consent of the copyright holders.

For this reason the BSD camp avoids at all cost to include anything under the GPL license in their distribution. They’ve recently switched from GCC to the Clang/LLVM compiler.

Another reason is that GPLv3 includes new clauses about patent and trademark which some individuals try to avoid. Some stay with the softwares under the v2 some switch to the BSD licenses.

Several major FOSS projects like the Linux kernel, MySQL, BusyBox Blender, VLC media player decided against adopting the GPLv3.

According to some stats, which are contested, between the 90s and 2000s, during the dotcom bubble, the GPLv2 was the most widely used license with a 42.5% usage rate.

In 2015 according the same company who made the first stats accompanied by Github this time said that it’s now the MIT license that is the most used and GPLv2 being second.

Other stats from Fedora in 2016 shows that the most used licenses in their packages are under the GPL license, and second place the MIT license.

Those stats shows that depending on where they are used some licenses are more favored than others.

Some odd questions

Let’s answer some odd questions.

  • Can an open source software be used for commercial purposes?

Sure as long as the license doesn’t limit you for what you want to do commercially. But beware that commercial is not equal to proprietary some licenses don’t allow you to place extra restrictions on the user.

  • Can I call my program “Open Source” or “Free Software” or “X”?

It’s probably better to avoid that if your license isn’t listed as an approved on that organization or group website. You have the right to call it whatever you want but that would confuse people too much.

That applies even if you have just changed a few words or added one clause, any small change implies it’s now a different license.

  • Can I use the code that a minor releases as open source?

This falls back to the laws of the country related to liability and copyright. In a lot of countries the copyright is usually transferred to the legal parents who will sign and handle everything, just like with the parents of child stars. So when using the software of a minor you are actually dealing with his or her parents.

  • Can I have many licenses on a single software?

What it’s referring to is called multi-licensing and the answer is yes but that depends.

Multi-licensing is the practice of distributing a software under two or more different set of terms and conditions.

If someone wants to use the software they can choose whichever license pleases them.

Where it becomes tricky is when that license is permissive and allows adding terms to it.

The original distributor is then not allowed to include the changes of a contributor that changed the license until all the original licenses are compatible and not contradictory with the original one.

The GPL goes a long way with that not allowing code under GPLv3 to be used with code under GPLv2 is it doesn’t include the “any later version” statement for instance.

Multi-licensing is common in scenario where a free software company wants to make business, they release a software under a free license and another license which allows to create proprietary softwares but has royalty fees.

This is a sort of business model of companies that use free softwares and want to remain the single-vendor of it.

There are many examples of multi-licensed softwares, MySQL, FreeType are good ones.

  • Can I swap the license along the way?

The short answer is yes, but again that depends.

If you own all the code then you can re-release the software under any license you want. That includes suddenly making a software under GPL into a proprietary one.

The problem that arises is when you include patches, fixes, and parts of code you don’t own into your first released free software.

Now if you can prove those contributions are benign/not-significant and that the free and open source license discussed or didn’t discuss the topic of contributions then you might be able to prove the code is still all yours, the original author.

Otherwise, you will have to contact the other contributors so that they relinquish their software to you, a legal agreement so that they give you ownership of the code they wrote.

That’s why in some communities, like the FreeBSD and OpenBSD camps and also the Linux kernel, copyright assignment is considered bad taste so there’s none.

What if someone still uses the prior free and open version?

Then this person is still able to make changes to the software as they wish, they are able to use it just like the prior license implied, the new license doesn’t apply to the old one.

That is one of the reason sometimes projects get forked.

A good recent example is Oracle that now owns MySQL and so the community forked it as the MariaDB project.

There are many examples of legal battle fought over free and open source licenses. Free software licenses have indeed been tested in court and their validity hasn’t been contested.

BusyBox is one interesting example.

Busybox is a wrapper of many core utilities that is under the GPL.

In 2007 the Software Freedom Law Center (SFLC) announced it has filed the first ever U.S. copyright infringement lawsuit based on the violation of the GPL on behalf of its clients, two of the original developers of BusyBox, against MoonSoon Multimedia Inc.

This arose because MoonSoon Multimedia publicly said that they were using BusyBox but were not releasing the source code of it to comply with GPL copyleft clauses.

The case was settled and BusyBox won, making MoonSoon Multimedia pay fees and release the related source code. This was the first time compliance with GPL was brought to court.

Tips and tricks

Let’s give out some tips to choose the right license.

There are websites which make it easy to search amongst licenses and understand their terms without having to read all the crevasses in them. However, like all summaries they don’t go into all the details.

Checkout: tldrlegal and choosealicense.

There’s also a tool called fossa which helps keep track of what licenses you are using so that you comply with them.

Conclusion

Legality is a deep subject.

There are two books in the show notes that go into more details:
“Understanding Open Source and Free Software Licensing”

Which is a freely available book I haven’t read but skimmed through. It goes over all the different licenses and discusses their ins and outs.

The second one, also free, is “Fundamentals of software licensing”.

This podcast was truly insightful for me, it answered a lot of questions I had and I hope it did for you too.

Cheers.





If you want to have a more in depth discussion I'm always available by email or irc. We can discuss and argue about what you like and dislike, about new ideas to consider, opinions, etc..
If you don't feel like "having a discussion" or are intimidated by emails then you can simply say something small in the comment sections below and/or share it with your friends.