Running a free software project can be a rewarding experience if you begin with your eyes open. In my personal experience, starting a free software project with only a head-on view of a few existing free software projects is not really enough. Some basic background information can really help get you started in the right direction.

Historical perspective

In 2001, I was working on the eDirectory development team at Novell. We needed to add service advertising functionality to the next version of eDirectory and we chose to go with OpenSLP—a free software implementation of the IETF Service Location Protocol. My manager asked me to write a patch for OpenSLP to add DHCP support—a feature that was slated for some future release of OpenSLP. As I was working with the OpenSLP project administrator on integrating this feature, he asked me to become a developer on the OpenSLP project, and I accepted. To shorten what could be a really long story, I eventually found myself running the project—simply because everyone else dropped out.

I’d downloaded, compiled and executed various free software projects in the past, and like many people who have reached this level, I thought I was pretty hot stuff. I didn’t have a clue; but I learned. Because of my new position of responsibility (and luckily, it wasn’t a very popular project), over the next 5 years, I gradually learned about everything from package creation and maintenance to mailing lists, forums, and web site management. This article is an attempt to ease the burden of discovery for programmers new to the free software world. I hope I succeed—to some degree at least—because running a free software project properly is a rather intense and detail-oriented process.

To run a free software project, you have to be the proverbial Jack-Of-All-Trades. Before we get started, I’d like to make a special note of one excellent resource for information similar to that found in this article: Karl Fogel’s free online book entitled, “Producing Open Source Software”.

Project setup

Setting up a project is actually rather trivial—well, the initial portion of the setup, anyway. However, you soon find that you’ve started a small snowball down a long hill. But let’s just start at the beginning. First you need to find a free software hosting service that you like. Here is a short list of popular services:

• http://www.sourceforge.net
• http://code.google.com
• http://dotsrc.org
• http://developer.berlios.de
• http://www.freepository.com
• http://gforge.org
• http://icculus.org
• http://savannah.gnu.org
• http://www.seul.org
• http://www.bountysource.com

I’ve listed these services in no particular order—with one exception: SourceForge.net is generally considered to be the definitive standard. Hosting over 120,000 free software projects, it offers a very extensive feature set to free software project administrators. I’ve heard it said that SourceForge is the largest repository of “dead” projects on the internet. However, it should be noted that to gain such notoriety SourceForge must, in fact, be a very popular free software hosting service. Otherwise, the main difference between these services is the feature sets offered. You will simply have to play with a few in order to determine which features you want to provide with your project. After reading the remainder of this article, you will be in a better position to judge the relative value of the various features offered by these services.

SourceForge.net is generally considered to be the definitive standard

In addition to this abbreviated list of general purpose public services, there are also several services hosted by various companies such as Novell, Microsoft, Sun, HP, IBM, etc, which provide free software hosting portals. Generally, these services are more closely affiliated with corporate goals and are intended to host projects that highlight technology specific to the host company. However, most of these services will allow any type of project to be hosted (as long as it’s not derogatory or derisive of the company providing the services, of course).

Some services have very specific goals in mind, and in fact, enforce these goals with very specific rules and regulations. The Eclipse project (which is sponsored and administered by IBM), for instance, wants to host projects that extend or otherwise enhance the Eclipse development environment framework. For this reason, since Eclipse is written in Java, nearly all of the code hosted at eclipse.org is written in Java. Prospective Eclipse projects are also scrutinized closely for relevance to the Eclipse framework, and every Eclipse project must use a free software license that was designed specifically for Eclipse.

Figure 1: Starting a new project on SourceForge.net

Once you’ve chosen a hosting service (the link will point you to a good favourite web hosting search web site), you will then have to go through the process of setting up your project by entering some initial information into a web form. Generally, the hosting service will ask a number of questions about the nature of your project, and details such as a short (file system compatible) name for the project, a longer name, and a short description. The service may also ask you what you believe the value of your project is to the free software community. I’ve set up several projects in this manner, and never had one rejected, so I’m inclined to believe that the process is a bit of a formality. Another piece of information that is usually requested at this time is the free software license that you wish to use for your project.

Choosing a free software license

To properly choose a free software license, you should be familiar with the general features of each type of license available. I strongly recommend that you read the overview information on each of the major free software licenses available at OpenSource.org. The concept of the free software license was originally invented by the Free Software Foundation under the guidance of Richard Stallman. Often these licenses were referred to as copyleft, implying “copyright with a twist”. Indeed, the rationale behind every free software license is fundamentally to keep the software freely available to the public. They effectively prohibit people or companies from stealing it and claiming it as proprietary, while at the same time allowing them to use it for their own purposes.

Each major type of free software license has features that make it either better or worse for a given purpose. For example, the GNU General Public License is a very good general purpose free software license, providing an airtight legal right to the general public to use any GPL licensed software freely and without charge. However, it comes with the often rather painful caveat that any software that consumes GPL software in any way must also be made freely available—in source code form—to the general public. This feature is sometimes labeled “viral” by software companies considering using some free software within their proprietary software products. They simply can’t use GPL software for this purpose because it forces them legally to give away their intellectual property.

On the other end of the spectrum, the MIT license is generally considered to be a very good free software license for projects that simply want to remain free, without imposing any significant restrictions on consumption. Basically, consumers must include the license in their proprietary offerings in a manner that allows the license to be publicly viewed, either at installation time, or in an “about” dialog of some sort, thereby giving credit to the providers of the free software that is being consumed.

So, if MIT is such a good license, then why not use it for everything? Ironically, the MIT license is often a little too loose for proprietary software companies wishing to release portions of their code base. Sometimes proprietary software companies would like to make code available to the “little guy” for free, but establish a dual license agreement with larger customers wishing to make more proprietary use of their free software offerings. I say this is ironic because there’s little doubt that Richard Stallman never intended GPL to have this effect on software. His goals were very idealistic—by forcing consumers of GPL software to also make their consuming software GPL, he clearly hoped to eventually make all software free.