The Saga of Sun RPC

Created
Fri, 27/08/2010 - 17:15
Updated
Fri, 27/08/2010 - 17:15

I first became aware of the Sun RPC license in mid-2001, but my email archives from the time indicate the issue predated my involvement with it; it'd been an issue of consideration since 1994. I later had my first large email thread “free-for-all” on the issue in April 2002, which was the first of too many that I'd have before it was all done. In December 2002, the Debian bug was filed, and then it became a very public debate. Late last week, it was finally resolved. It now ranks as the longest standing Free Software licensing problem of my career. A cast of dozens deserve credit for getting it resolved.

Tom “spot” Callaway does a good job summarizing the recent occurrences on this issue (and by recent, I mean since 2005 — it's been going long enough that five years ago is “recent”), and its final resolution. So, I won't cover that recent history, but I encourage people to read Spot's summary. Simon Phipps, who worked on this issue during his time as the Chief Open Source Officer of Sun, also wrote about his work on the issue. For my part, I'll try to cover the “middle” part of the story from 2001-2005.

So, the funny thing about this license is everyone knew it was Sun's intention to make it Free Software. The code is so old, it dates back to a time when the drafting of Free Software licenses weren't well understood (old-schoolers will, for example, remember the annoying advertising clause in early BSD licenses). Thus, by our modern standards, the Sun RPC license does appear on its face as trivially non-Free, but in its historical context, the intent was actually clear, in my opinion.

Nevertheless, by 2002, we knew how to look at licenses objectively and critically, and it was clear to many people that the license had problems. Competing legal theories existed, but the concerns of Debian were enough to get everyone moving toward a solution.

For my part, I checked in regularly during 2002-2004 with Danese Cooper (who was, effectively, Simon Phipps' predecessor at Sun), until I was practically begging her to pay attention to the issue. While I could frequently get verbal assurances from Danese and other Sun officials that it was their clear intention that glibc be permitted to include the code under the LGPL, I could never get something in writing. I had a hundred other things to worry about, and eventually, I stopped worrying about it. I remember thinking at the time: well, I've notes on all these calls and discussions I've had with Sun people about the license. Worst case scenario: I'll have to testify to this when Sun sues some Free Software project, and there will be a good estoppel defense.

Meanwhile, around early 2004, my friend and colleague at FSF, David “Novalis” Turner took up the cause in earnest. I think he spent a year or two as I did: desperately trying to get others to pay attention and solve the problem. Eventually, he left FSF for other work, and others took up the cause, including Brett Smith (who took over Novalis' FSF job), and, by that time, Spot was also paying attention to this. Both Brett and Spot worked hard to get Simon Phipps attention on it, which finally happened. But around then began that long waiting period while Oracle was preparing to buy Sun. It stopped almost anything anyone wanted to get done with Sun, so everyone just waited (again). It was around that time that I decided I was pretty sure I never wanted to hear the phrase: “Sun RPC license” again in my life.

Meanwhile, Richard Fontana had gone to work for Red Hat, and his self-proclaimed pathological obsession with Free Software (which can only be rivaled by my own) led him to begin discussing the Sun RPC issue again. He and Spot were also doing their best negotiating with Oracle to get it fixed. They took us the last miles of this marathon, and now the job is done.

I admit that I feel of some shame that, in recent years, I've had such fatigue about this issue — a simple one that should've been solved a decade and a half ago — that, since 2008, I've done nothing but kibitz about the issue when people complained. I also didn't believe that a company as disturbing and anti-Free-Software as Oracle could ever be convinced to change a license to be more FaiF. Spot and Fontana proved me wrong, and I'm glad.

Thanks to everyone in this great cast of characters that made this ultimately beneficial production of licensing theater possible. I've been honored that I shared the stage in the first few acts, and sorry that I hid backstage for the last few. It was right to keep working on it until the job was done. As Fontana said: Estoppel may be relevant but never enough; software freedom principle[s] should matter as much as legal risk. … [the] standard for FaiF can't simply be ‘good defense to copyright infringement likely’. Thanks to everyone; I'm so glad I no longer have to wait in fear of a subpoena from Oracle in a lawsuit claiming infringement of their Sun RPC copyrights.