Recently, the Apache Software Foundation (ASF) celebrated its 10th birthday. As the day approached, I spoke with several Foundation representatives and enjoyed hearing their stories of their accomplishments and their hopes for the future. As I listened to their aspirations, I could not help but wonder whether the time for open-source foundations has come and gone.

In the early days of the movement, the Free Software Foundation (FSF) was the key player in establishing an open-source foothold. It was the pointy end of the lance, with lots of emphasis on “pointy end.” Success, which took a solid two decades, was hard won and important.

Since then, however, the contributions of the FSF have primarily consisted of revisions to licenses, campaigning against patents, and revving existing products. No new products with the sweep and obvious usefulness of emacs or gcc have come to market via the FSF for a very long time.

The ASF, in my view, is in a similar point in its trajectory. Great, new Apache projects are a thing of the past. The Apache Web server and the Apache Portable Runtime were important contributions when they first came out. So were Jakarta and the Apache Commons and, supremely so, the business-friendly Apache license.

But the ASF has increasingly become a mere host as its role has changed. It has become primarily an organization for taking in projects from other sources. It gives those projects tools, a website, a process for establishing community, and visibility.

The problem is that many of the projects it has accepted were moribund or likely to become so (for example, Hivemind, a codebase donated by BEA when that company could find nothing else to do with it). It is a standout in this “dumping ground” category. Its last release was in late 2006, but it took three more years for the project to be officially retired.

Other Apache projects are such omnibus propositions that it’s nearly impossible to fill all the needed roles. Apache POI, the project to read and write Microsoft Office files, is a prime example. The impenetrably name HWPF subproject works on reading and writing Microsoft Word files. That project has been looking for a lead developer for years, and the work on other POI subprojects advances slowly.

Moreover, the ASF is willing to accept multiple projects that do the same thing. I am not talking about Apache Ant vs. Apache Maven, but projects that are essentially duplicates. For example, the ASF’s two continuous integration (CI) servers: Apache Continuum and Apache Gump. While competition is indeed desirable, hosting competing products under the aegis of the same organization cannot be a plus. (Both CI servers are likely to be rendered irrelevant by Hudson, which is a hub of intense activity and increasingly the preferred open-source CI server.)
The ASF once stood as the host for top projects. As the years have passed, its ability to control the quality of the incoming projects and to breathe added life into them has steadily diminished. It still attracts good projects (Hadoop), but for each one of those, it also picks up its share of mature projects that are unlikely to develop much further (Subversion). It is becoming increasingly like a pure open-source hosting site with the limitation of a bureaucratic acceptance process.

The upshot is that other sites are emerging to become hot spots of the open-source universe. Among these is Codehaus, which hosts Groovy, Grails, Jetty, JRuby, Drools and the majority of plugins to Apache Maven. Another site is Google Code. The latter has become an important player due to the many open-source contributions of Google itself. Google Chrome (browser), V8 (JavaScript engine), Google Web Toolkit, Google Guice (dependency injection), and the Go language are all stored in this repository.

A final site of interest is GitHub, which hosts the code for Ruby on Rails, jQuery, git itself, and many other projects.

There is a clear transition going on. The bazaar of open source used to be much more centralized than it is today. It used to gather around certain oases such as the FSF and the ASF. But those organizations are slowly being converted into just hosts as their original contributions of actual code have declined. Their principal role going forward is providing licensing terms, legal opinions and, to a lesser extent, serving as expressions of the larger open-source movement. But the movement itself is now highly heterogeneous and much more oriented toward good hosting sites, rather than foundations and institutions.

Andrew Binstock is the principal analyst at Pacific Data Works. Read his blog at