Going not gone

Before we “light out for the territories”, we thought we’d take a quick look back at an open source community in which we’ve spent a lot of time and learned a lot over the last several years–Eclipse.


We got involved in Eclipse back in 2003, long before Cloudsmith was started. We were developing Buckminster–the technology that helped put us on the road to Stack Hammer–and needed an open source community in which it could thrive. When we founded Cloudsmith, we decided to put as much of the technology we were developing into open source as we could and made a bigger bet on Eclipse. So we became strategic members and served on the board for several years, in addition to developing most of our foundational technology as open source at Eclipse.


To be honest, it wasn’t easy at first, since we were competing for mindshare in a community that was dominated by big companies like IBM. But it’s actually true that the community is merit-driven. Acceptance isn’t immediate, but if you have good ideas, produce good technology and are willing to contribute, you’ll earn your way in.


For us, the turning point was jumping into p2, Eclipse’s dependency management and provisioning framework created by a large team at IBM. Initially, it was seen as competitive with Buckminster, and to some extent it was. We pruned Buckminster in a few areas and started contributing major extensions to p2. After a few months, we were welcomed as the first active p2 committers outside IBM. (“Committer” is an Eclipse term for a core member of the project team.) And Buckminster continued to thrive.


Similarly, we helped the foundation automate the build of the first annual Eclipse release in 2007. By last summer, that system had evolved into the b3 aggregator, which is widely used throughout the Eclipse community, most visibly to assemble the work of roughly 500 committers and 40 separate projects into Eclipse’s official release-stream. Meanwhile, we’d spent lots of time facilitating Buckminster adoption, with over half of all active Eclipse projects (and nearly all of the roughly 30 or so modeling projects) building their contributions with Buckminster.


When Kenn Hussey, lead of the Model Development Tools project, joined the team full-time last year to run development (complementing Ed Merks, who’d been our “modeling guru/cons”), we added modeling to our portfolio of activities in the component build-assemble-deploy domain. This came in really handy when were looking for ways to abstract away dependencies between the server-side technologies in Stack Hammer and the GWT (Google Web Toolkit) technology we were using in our web interface. And so we ended up making major contributions to the Eclipse modeling core to support development of GWT applications and other RESTful web UIs.


And so after several years of hard work, which had made us kind of the “the dudes” in some really important technical domains at Eclipse, why have we been scaling back our involvement? Limited bandwidth and the need to focus, really. In order to spend more time using (and contributing to!) some new technologies, we have to cut back on the time we spend on Eclipse.


There’s still plenty of great Eclipse technology inside our service, and we’ll still be contributing as much as we can (i.e. still plenty) to those Eclipse technologies on which we or others depend. But to free up room to do something new, we have to cut back elsewhere.


Therefore, to all our past, current and future friends at Eclipse, it’s been great to meet you and has been/is/will be great to work with you! And we’re still here (maybe just a little less so).