NIPY logo

Site Navigation

NIPY Community

Github repo

This Page

Why we should be using BSDΒΆ

John Hunter - 16 Dec 2004

I’ll start by summarizing what many of you already know about open source licenses. I believe this discussion is broadly correct, though it is not a legal document and if you want legally precise statements you should reference the original licenses cited here. The Open-Source-Initiative is a clearing house for OS licenses, so you can read more there.

The two dominant license variants in the wild are GPL-style and BSD-style. There are countless other licenses that place specific restrictions on code reuse, but the purpose of this document is to discuss the differences between the GPL and BSD variants, specifically in regards to my experience developing matplotlib and in my discussions with other developers about licensing issues.

The best known and perhaps most widely used license is the GPL, which in addition to granting you full rights to the source code including redistribution, carries with it an extra obligation. If you use GPL code in your own code, or link with it, your product must be released under a GPL compatible license. Ie you are required to give the source code to other people and give them the right to redistribute it as well. Many of the most famous and widely used open source projects are released under the GPL, including linux, gcc and emacs.

The second major class are the BSD and BSD-style licenses (which includes MIT and the python PSF license). These basically allow you to do whatever you want with the code: ignore it, include it in your own open source project, include it in your proprietary product, sell it, whatever. python itself is released under a BSD compatible license, in the sense that, now quoting form the PSF license page

There is no GPL-like “copyleft” restriction. Distributing binary-only versions of Python, modified or not, is allowed. There is no requirement to release any of your source code. You can also write extension modules for Python and provide them only in binary form.

Famous projects released under a BSD-style license in the permissive sense of the last paragraph are the BSD operating system, python and !TeX.h

I believe the choice of license is an important one, and I advocate a BSD-style license. In my experience, the most important commodity an open source project needs to succeed is users. Of course, doing something useful is a prerequisite to getting users, but I also believe users are something of a prerequisite to doing something useful. It is very difficult to design in a vacuum, and users drive good software by suggesting features and finding bugs. If you satisfy the needs of some users, you will inadvertently end up satisfying the needs of a large class of users. And users become developers, especially if they have some skills and find a feature they need implemented, or if they have a thesis to write. Once you have a lot of users and a number of developers, a network effect kicks in, exponentially increasing your users and developers. In open source parlance, this is called competing for mind share.

So I believe the number one (or at least number two) commodity an open source project can possess is mind share, and mind share means you want as many damned users using your software as you can get. Even though you are giving it away for free, you have to market your software, promote it, and support it as if you were getting paid for it. Now, how does this relate to licensing, you are asking?

Most software companies will not use GPL’d code in their own software, even those that are highly committed to open source development, such as enthought, out of legitimate concern that use of the GPL will “infect” their code base by its viral nature. In effect, they want to retain the right to release some proprietary code. And in my experience, companies make for some of the best developers, because they have the resources to get a job done, even a boring one, if they need it in their code. Two of the matplotlib backends (FLTK and WX) were contributed by private sector companies who are using matplotlib either internally or in a commercial product – I doubt these companies would have been using matplotlib if the code were GPL. In my experience, the benefits of collaborating with the private sector are real, whereas the fear that some private company will “steal” your product and sell it in a proprietary application leaving you with nothing is not.

There is a lot of GPLd code in the world, and it is a constant reality in the development of matplotlib that when we want to reuse some algorithm, we have to go on a hunt for a non-GPLd version. Most recently this occurred in a search for a good contouring algorithm. I worry that the “license wars”, the effect of which are starting to be felt on many projects, have a potential to do real harm to open source software development. There are two unpalatable options. 1) Go with GPL and lose the mind-share of the private sector 2) Forgo GPLd code and retain the contribution of the private sector. This is a very tough decision because their is a lot of very high quality software that is GPLd and we need to use it; they don’t call the license viral for nothing.

The third option, which is what is motivating me to write this, is to convince people who have released code under the GPL to re-release it under a BSD compatible license. Package authors retain the copyright to their software and have discretion to re-release it under a license of their choosing. Many people choose the GPL when releasing a package because it is the most famous open source license, and did not consider issues such as those raised here when choosing a license. When asked, these developers will often be amenable to re-releasing their code under a more permissive license. Fernando Perez did this with ipython, which was released under the LGPL and then re-released under a BSD license to ease integration with scipy and other enthought products. The LGPL is more permissive than the GPL, allowing you to link with it non-virally, but many companies are still loath to use it out of legal concerns, and you cannot reuse LGPL code in a proprietary product.

So I encourage you to release your code under a BSD compatible license, and when you encounter an open source developer whose code you want to use, to do the same. Feel free to forward this document on them.