Short story, I've done bits and pieces.

Over the last 7 or 8 years, the term "open source" has become a huge buzzword. In some cases, it use actually holds more than mere buzz. In many cases, buzz is really all that it's about. As I sit around taking stock of my background while school slowly comes to an end, I've given some thought to some of my contributions to projects in the past.

Prior to the open source push, there were a handful of licenses that covered software and source code distribution. Principally, there Free Software Foundation's General Public License, a key piece of its push for software freedom, the various forms of the Berkeley Systems Division (BSD) license, and the MIT license (used by X11.) These originated before the Internet became a popular medium, and the result was that the concept of licensing was not well known.

How do I know this? Well, for one thing, I made great deal of use of a programming language called MOPS. It was distributed under the... Well, no license. It was released in the public domain. It was based on a package called Neon, which was released into the public domain. That spawned at least one other version called Yerk (which I also used.) You may note I still retain a credit, though for the life of me, I don't remember doing much significant work on the MOPS core:

Mops implemented by: Michael Hore
Able assistance from: Doug Hoffman, Nao Sacrada, Greg Haverkamp, Xan Gregg

Documentation: Mike Hore, Ed Williams, Nao Sacrada, Arthur W. Green
HTML:

Gnarlodious, Arthur W. Green, Jim Tittsler

I did make a few contributions through the principal newsgroup, comp.lang.forth.mac. See here and here.

Then things slowed down for a while. I started working in software companies, and in the mid-90s, most companies weren't too keen on the idea of giving away any intellectual property. For instance, a wrote a system to do firewall reporting for old Linux ipfwadm-based firewalls, while I was at Instinctive (now, basically, represented by this product of EMC.) When I asked to share this series of Perl scripts that generated some useful statistics and analysis -- developed on software licensed under the GPL -- you would have thought I wanted to give away the keys to the kingdom. I was lectured on how, as a small startup, all we had was our intellectual property, and the very idea of giving any of it away was insane. I later wrote some Java web stuff I thought was very cool, but that, also, was apparently too critical to the business, in spite of the fact that the CTO seemed to absolutely hate Java back then.

At Curl, we had grand plans. MIT Curl was actually released under an MIT license, and we thought we'd be releasing some components of commercial Curl under a spiffy license. But they didn't really need me around at Curl, and I left after only 8 months.

That's when I ended up at ArsDigita, where the founder, Philip Greenspun, was, for all his other faults, a huge advocate of Free Software. Our "framework," the ArsDigita Community System, was released under the General Public License. I contributed a couple of modules (a calendar widget and the bulkmail module.) I'm sure I touched some other pieces, too. In the end, it wasn't our use of the GPL that killed us, but rather the grandiose visions of going public that our founders held.

More recently, I've seen some of the value of open source creep up again. We were having some problems at work, and there was a demand that we investigate better mail systems. Curious, a couple of us started poking around into the problems, and we discovered it was this bug in Thunderbird. With about a day sifting through the Thunderbird message drafting process followed by a few lines of code, I fixed that 4-year-old Thunderbird bug. If you have Thunderbird 2.0, you've got a very small amount of my code in there.

Then, two weeks ago, it became imperative to get the Sympa mailing list manager up and running. I decided I'd use SQLite to do so. Unfortunately, there were some problems, and I spent much of my critical time trying to get things done hunting down an unknown bug in Sympa. Apparently there had been no testing of SQLite in the 5.3 release.

In the end, I'd say at best I've been an active user of open source, contributing little bits and pieces here, but certainly I've never really been an open source developer.