Opening Zope: An Interview with Paul Everitt

by David Sims
   Paul Everitt
CEO of Digital Creations

Zope is an open source application server and content management platform, developed by Digital Creations.

• Jim Fulton, Digital Creations' chief technical officer, is the "benevolent dictator" guiding Zope's development.

• For an update on Zope and Python, read Frank Willison's report from the Eighth Annual Python Conference.

What are the business reasons to open the source code of a software product? We recently talked with Paul Everitt, the CEO of Digital Creations in Fredericksburg, Virginia. Everitt is one of the main drivers behind Digital Creations' content management and application server tool that has become known as Zope. He and his team developed the software when they were contractors for InfiNet, a consortium of newspapers, where they used it to put classified ads and other content online. Zope was written with the Python programming language. We asked him why they chose Python.

listen Listen to this interview:

Paul Everitt. Python was in our blood. I had been a long-time Python user. I'm not that much of a programmer, which probably attracted me to an elegant, very clear, very clean language. I'm more of an architect, and so Python is a very well architected scripting language. Our CTO, Jim Fulton, is a long, long, longtime Python ... key person in the Python community. And he came from a Smalltalk background, and he's an object zealot, and he was very attracted to the object model of Python.

Using Python, they built a classified ad engine and a web management framework. That evolved into their commercial project, which they called Principia.
"By 1997, the application server market was full of billion-dollar companies. It's pretty hard to crack into that kind of market."

Everitt. So we think we've got the best ideas in the world and best ideas always win. So we became a product company in 1997. InfiNet spun us off and we got the intellectual property, and we took the intellectual property and tried to go to market with a product. By 1997, the application server market was full of billion-dollar companies. It's pretty hard to crack into that kind of market.

They needed an angle, and they needed money. Both came from a venture capitalist.

Everitt. We were making our money from services, from consulting, and that's when we were approached by Hadar Pedhazure, a venture capitalist, who convinced us to realize we were already making our money from services. Trying to be a product company, going into a multi-billion dollar market, was going to be a tough fight, and he convinced us that we should be a services company.

Sims. It was actually in Linux magazine, in the January 2000 article, there was a quote, I think it was from you, that said when you sign a $100,000 service contract, the $4,000 product suite becomes kind of irrelevant at that point.

Everitt. Yeah, it's just noise, and in fact we were finding that when you compare it to the competition and you're much less expensive than the competition for these huge deal sizes, they get worried. Why doesn't it cost more? is the response. So you might as well just give away the architecture for free as a way to help you get the foot in the door.

Digital Creations got its first round of funding in October 1998. The next month, they decided to open source their product, and their first open release came in December 1998. I asked Everitt how one goes about managing the transition from proprietary to open source code.

Everitt. That's a good question. You make a lot of mistakes, I'll be honest about it. You learn a lot. We had already been a provider of open source software. Bobo had gone out for a long time, a number of things that went into Python came from us and we maintained them. And we had, in '96, we had released a piece of software to turn web servers into CORBA servers. So we had it in our blood and we were very much consumers of open source. We just patterned a lot of our approach after the Python approach. So Python is the model of a benevolent dictator, like many of the successful open source projects such as Perl. And under that model you can have lots of contributions but you want it to be one person, realistically one person's vision and view of how the system should be.

For Zope, that person is Jim Fulton, Digital Creation's chief technology officer, who guides internal and external development. Everitt says he's glad to see important development work coming in from the Zope development community.

Everitt. The core is still very much a Digital Creations focus. On the other hand, we've seen a number of things happen in the last couple months. For instance, the most important thing to happen in the last couple of months came from outside of Digital Creations. It's something called Python Methods that lets you store code in the database in a safe way. That was a problem that we had been trying to stay away from because it was a hard problem. And somebody else picked it up and iterated on it, iterated on it, iterated on it, and got to a point where we felt like -- and his name is Evan Simpson -- and we felt like he had done such a good job on it that it was time for us to go ahead and talk with him about moving it into the core. That to me is a pretty exciting thing. And there are a couple of other things. A gentleman named Ty Sarna is hosting the object database in a Berkeley database, and that's something we'd wanted to do for a long time.

Sims. In driving development, do you meet every other week or every month and say, you know, "This is a person who's come to us with this idea and they look like they have the experience. We'd like to go with it." Or is it -- is it that organized or does the development not work that organized? Do you keep seeing iterations come in and suddenly it's sort of a light bulb goes off and you have what you were hoping for?

"There's a tremendous Zen to running an open source community."

Everitt. It seems a lot more Darwinistic than centrally planned, kind of a free market. Things that work are things that work. You can't really plan how things will work. You're dealing with a lot of people that you've never seen in all these wild time zones, and you push people along, encourage them when they do a good job, steer them in the right direction, and just see how things shape out. And that's really one of the secrets to running a good community. There's a tremendous Zen to running an open source community.

Sims. Really. Say a little more about that.

Everitt. Sure. For one, it takes resources. That's one of the classic myths. I was the moderator of a panel at the O'Reilly Open Source Conference last year and I had Eric Raymond and Guido [van Rossum, Python's orginator and technical director of the Python Consortium] Frank Willison from O'Reilly [editor-in-chief], and a couple other people, and one of the points I wanted to make, and wanted to get them to talk about, is this myth that going open source saves manpower. To me that is a myth because that means you're ripping off the community. You better plan to devote engineering time and resources to engaging the community and assimilating the great ideas and the good work coming out of the community, and providing the information and the resources the community needs. Then you get what we saw, which is something that came out of nowhere and turned into a platform strategy that everybody's excited about.

Sims. Came out of nowhere, came from just being a product to a platform that people are building applications on top of? Is that what you mean by "platform"?

Everitt. Exactly.

Sims. So it's interesting what you said about having to provide resources, because there's this sort of myth arising that the reason to go open source is get these hackers all over the world to do your development for you, right?

Everitt. Yeah, in fact to put a little bit of timeliness to it, Borland/Inprise/Borland is talking about open sourcing Interbase, their database server. And everybody's really worried because, in the last two weeks, three of the top brass for Interbase left -- the project manager left and a couple of other people. Is this just an attempt to get out of the Interbase business and not have to fulfill contracts with customers? If they just release the code, that is a worse-than-useless gesture, because there's no way realistically that somebody out in the world is going to pick up and have the vision, the Zen, of Interbase. They need to be in it for the long haul, continuing to devote the resources they've devoted to date, if Interbase is going to be a successful open source project -- in my opinion.

Sims. So in your opinion, it isn't that opening open source is, it's not an option as a desperation move -- "We can't support this product anymore and we've got to find other people to do it" -- it's just a way to scale, so to speak.

Everitt. Right. If it is a desperation move -- and something that we firmly believe in is when you talk about open source with the community or with customers, you might as well tell the truth, because they're going to figure it out anyway -- if it's a desperation move everybody will realize it. And it will die just as quickly but in a different way than it would have if you didn't open source it. Really for us, going open source wasn't a question about desperation. It was a platform strategy. It was an attempt to get a platform standard out there, get people using it, doing things with it, and increasing the gravity of the company by getting a lot more people in the platform and on top of the platform.

"By the time I get online about 5:30 in the morning, somebody in the Netherlands has already answered the question."

If going open source was a way to build the platform, it's brought with it another benefit. In a tight labor market, Everitt says that developing an open source platform is a great attraction for hiring developers.

Everitt. If you have a hiring problem, open source your software. Everybody will want to come work for you. It's amazing. People, talented developers love open source, and therefore if you're open source you're not going to have a hiring problem. And we find that we can project ourselves pretty well around the world without having to be in the Valley.

Having a 24-hour workday is another plus.

Everitt. By the time, people pose questions and by the time -- I usually get online about 5:30 in the morning -- by the time I get online somebody in the Netherlands has already answered the question. The worst thing is that they always have a better answer than I would have had.

Sims. Another secret of open source.

Everitt. That's right.


David Sims is the editorial director of the O'Reilly Network.