Open Source Convention

Last Day and Wrap-up

Tim's keynote speech on Thursday highlighted an important new kind of software application and development. (Note: In the interest of full disclosure, this same Tim is my boss.) He asserts that the most important software developments of the last five years have never been downloaded to the desktop. For example, consider Mapquest, Amazon, and Napster. These applications are not distributed in the traditional retail sense; that is, you don't get them on a disk, and you don't download them from the Web. You just go to the Web site and you use them. "Performance" is the mark of use, not ownership.

Tim notes that these applications blur the rules we currently use in the open source movement. It doesn't make sense to get the code from the Amazon site, download it, and create your own Amazon or debug theirs. But it does still make sense to look at these types of applications and figure out how to gain insight and information from them in ways their creators did not intend.

Tim mentioned "amarank," a program we use at O'Reilly to gather and organize book data made freely available at the Amazon site. Tom Christiansen created this Perl script (and Tim Allwine modified it) to use the data that Amazon displays programmatically. Tim suggests that such applications, in the spirit of open source, anticipate this kind of use and provide APIs to make it easier. He noted the reaction of Rael Dornfest, Senior Web Developer at O'Reilly Network and creator of meerkat, when he found out that people were writing programs against meerkat results. Rather than preventing or ignoring their use of his data, he provided an API to let them use it easily and correctly.

This is the spirit of open source; as Tim said, the heart of open source development is Internet-based cooperation.

Tim also noted that these new applications bypass the licensing requirements that many of us are arguing over. Licensing, even the GPL, is all about distribution of the software. If, instead, it is the performance that is distributed, the activities that trigger license requirements go away. Whatever licenses that organizations devise in the future will have to take this change into account.

I thank Tim for mentioning licensing because it gives me an opportunity to bring up the licensing discussions from the Open Documentation Summit on Sunday. The attendees at the Open Doc Summit spent a fair amount of the afternoon discussing the various licenses available as well as their strengths and weaknesses. We looked at the Open Content License, the Open Publications License, and the Free Documentation License.

I was not present for much of this discussion (the steamy, locker-room atmosphere of the late-afternoon conference room drove me repeatedly into the hallway); but Chuck Toporek, able O'Reilly Open Source Editor, told me that there seemed to be a lot of interest in the Free Documentation License. Guylhem Aznar, from the Linux Documentation Project, was the primary advocate of this license at the meeting.

There was some confusion about the new version of the FDL. According to Guylhem, the FDL allows the author to declare parts of the document "non-technical." These "non-technical" sections could contain, apparently, advertisements or attributions which the author could require reusers to include. I was puzzled by the parts of this discussion that I heard, and Chuck admitted having some confusion as well. (I understand some subsequent discussions with Guylhem and some FSF representatives took place later in the conference.) I'd say that the final form of the FDL is still evolving.

There were two other license issues of some importance to the summit attendees. Other than those of us who work for a publisher, attendees didn't like the option provided by the OPL that lets authors limit commercial publication to a designated publisher. Almost nobody felt that it was a big issue because no publisher would be likely (anymore) to take open source material from another publisher and use it. The community would express its disapproval, presumably, in the marketplace.

Nevertheless, I see an advantage to such a clause if it provides motivation for a publisher to take on the financial and resource burden of publishing a book. Editing, printing, storing, and distributing books cost money; and the publisher undertakes the risk at his peril. It is true that community support can add strength to the spirit of a license; but the purpose of a license is to spell out the acceptable uses of the material. If we are to rely on community support, then why do we need a license at all?

We were also concerned that any license could cover material other than text documentation. For example, some attendees were producing training materials and videos. Music was another area of interest. There was some concern that the FDL was not broad enough to cover this kind of material. (The Open Content License seems to do so.)

On another topic: I attended Guido Van Rossum's talk on the State of Python. Although not much has changed since the Python Conference in January (more than adequately covered in some earlier Frankly Speaking columns), this was an important discussion. As you may know, the Python development team has left its previous sponsor, the Center for National Research Initiatives (CNRI), to join a for-profit company, BeOpen. (Also, Guido got married. Nobody better say that Guido is afraid of commitment.)

This change has meant resulted in some contention regarding the licensing of Python. The heads of BeOpen and CNRI have been meeting to discuss this issue and claimed at this talk that they had reached a rough agreement, but they were reluctant to discuss the details of it. There is a meeting today (Friday) of the Python Consortium (companies that contribute financially to the development of Python) where this and other issues related to the move will be discussed.

The Python team will remain in the Reston, VA area. The BeOpen Python team consists of Guido Van Rossum, Barry Warsaw (among other duties, the maintainer of JPython), Jeremy Hylton, Fred Drake, and the legendary Tim Peters who appeared at this conference for the first time (in my experience). For those Python users who rely on Tim's posts but have believed that "Tim Peters" was just an alter ego used by Guido, I can testify that I saw Tim, talked to him, and shook his hand. He's a rather normal human being.

Guido stated that he's looking for some new employees. I suggest you apply after September, when the temperature and the humidity in the Washington area drop below 98.

It now appears that release 1.6 will come from CNRI; release 2.0, which will resemble 1.6 greatly (but include some new features), will be released shortly thereafter by BeOpen. New features, as noted in earlier columns, include UniCode, a new regex engine, better garbage collection, and improved XML support. Release dates are under discussion between the parties. Python code is now managed at Source Forge, and is fully available there.

I missed a lot of the Perl activity, being a Python camp follower, but I understand that the Perl community has decided to name their "best presentation" award after Damien Conway as a way of relieving themselves of the obligation of awarding it to him every year. I saw Damien's talks at YAPC, and they are an experience: erudition, humor, science, and Perl code, all tossed together into a lively talk guaranteed to go over the time limit by at least half an hour. Congratulations to Damien on this recognition of his talents.

I've heard that this conference won't be held in Monterey next year. I think that is a shame because Monterey is a very nice place to spend a week. It does seem to have a strange attitude toward its cannery past: Somehow, I have trouble believing that John Steinbeck would be honored to know that a Plaza named for him contains souvenir shops and video arcades. It made me want to start an even more ironic company: Woody Guthrie Realty. Motto: This land is my land. I even thought of the first development project, a motel attached to the famous, and spectacular, Monterey Bay Aquarium. Its motto: Stay with us and sleep with the fishes.

Let me make one more comment, unrelated to the conference but related to getting there. On the airplane trip out, I began reading Acts of the Apostles, by John F.X. Sundman. It's privately published, but available through a number of retailers including Tech Book Retailer SoftPro (www.softpro.com). (I borrowed my copy from O'Reilly Editor Andy Oram, who worked with John when John was with Sun Microsystems.) I got hooked and found myself sneaking away from BOFs to finish it. It is a techno-thriller, not high literature. The women in it are brilliant, ambitious, mostly amoral, and (surprise!) knock-down gorgeous. John manages to weave a lot of technology into the book at a breathless pace. But I'm not a big fan of technology in fiction, so I wouldn't recommend it for that reason. I am, on the other hand, a Sixties-style paranoid and a conspiracy buff, and I have to say that this book would easily win the World Series of Paranoia. Read it and you'll never think about genomes or nanomachines without a bit of a spinal chill.

Frank Willison

Return to: Frankly Speaking