Pacific Connection(英語)

A GNOME Update: Miguel de Icaza and the Launch of the GNOME Foundation

When 19-year-old Matthew Tang decided to revamp his home computer, he committed an act that has become the badge of honor in the free software movement: he reformatted the disk, wiping it clean of Microsoft Windows. Tang, a freshman student majoring computer science at the Metropolitan State College of Denver, installed Red Hat Linux 6.1, the GNOME desktop and StarOffice, which had recently been acquired by Sun Microsystems as an open source project. He loaded browsers from Netscape and, and the Pine email program, which was created by the University of Washington. For his programming work, he now uses the Mesa 3-D graphics library, which was originally created by Brian Paul as an alternative to Silicon Graphics' OpenGL graphics program. For code development, he uses the kwrite text editor.

"The first time I installed Linux, it was a pain," he recalls. "It's still not as easy as running Windows. But there's a reward of running software that's available for free, strictly from the Net." As for Bill Gates and company, Tang wants nothing to do with them. "He's a monopoly," he says. "Hackers broke into Microsoft's system. If the source code was open, that wouldn't have been a problem." What about Microsoft's intellectual property rights? "Information should be free," Tang responds. "And source code is information."

As one of the next generation computer users, Tang is still in the minority. Most of his fellow students still use Windows. But Tang is hardly alone in his disdain for commercial" software of all kinds. Indeed, his views are mirrored by the man behind GNOME, the desktop environment Tang runs on his system: Miguel de Icaza who says that in his personal universe, any software package that isn't free doesn't exist. "That's pretty much what my laptop looks like," he said, when I mentioned Tang's setup. "I removed Windows from my hard drive so that I had more room."

The 27-year-old De Icaza's latest creation is the GNOME Foundation, a non-profit organization to promote and coordinate GNOME development. According to the foundation's official FAQ, written by Dan Mueth and Havoc Pennington, the organization was formed in response to the growth of the GNOME Development Project, now encompassing hundreds of developers in a number of companies. "GNOME's previous, completely informal method for making decisions does not scale well to its current size. The GNOME Foundation will provide a forum where all of the members of the GNOME community can have a voice and help decide the direction that GNOME moves in." The foundation will become the mouthpiece for communicating about GNOME to the media, (although an e-mail I sent to the foundation got no reply) and will put "decisions once made behind the scenes by the top GNOME hackers" on the table for all to view.

Membership is open to anyone who has contributed to the project and the primary benefit is the ability to vote for the foundation's board of directors. While corporations are represented on an advisory board, that board has no ability to make decisions. Supporting companies include Compaq, Eazel, Free Software Foundation, Gnumatic, Helix Code, Henzai, Hewlett-Packard, IBM, Object Management Group, Red Hat, Sun Microsystems, TurboLinux, and VA Linux.

"On the corporate side, more companies now have a stake in GNOME," said Bart Decrem: co-founder of Eazel. "So we needed a more structured governance model. The foundation is about finding a way to maintain the control and decision-making power with the hackers, while still giving companies, particularly those which stake their future on GNOME, to keep up to date and have their voices heard. The foundation is not a consortium---that's an important point. This is the case of hackers organizing themselves. While we have an advisory board of influential companies, as well as the Free Software Foundation, decisions will still be made by made an elected board."

Havoc Pennington, a software engineer at the Red Hat Advanced Development Lab and a member of the foundation's 10-person interim steering committee, has a similar view. "Once you have 300 people, you have to have some kind of organization. The foundation is a way for companies to get involved in the project. And it gives us a legal entity to do things on behalf of the GNOME project, like having a bank account and holding a copyright."

One of the biggest names on the GNOME Foundation's corporate advisory board is Sun Microsystems. John Heard, group technologist for Sun's webtop and application software says that the foundation is "a major step forward in the co-evolution of both open source and commercial development. " Heard says that the lines between commercial and open source development are blurring, particularly with network based applications, which are inherently distributed and built in building block fashion. He believes that the emergence of a hybrid development model combining open source and commercial development is inevitable. "The commercial companies bring a robustness and consistency, which we as end users want to see. Open source tends to be more fluid, but that has power in its own right, with the ability to generate a lot of innovation. So bringing those two together is very important."

"There will never be a KDE Foundation"

The spotlight on GNOME has put KDE, the other Linux desktop, in the shade. The GNOME and KDE communities have tried hard to keep relationships cordial between their two competing GUIs. But in truth, a single dominant GUI would be as beneficial to Linux as Windows was to DOS and the MacOS is to the Macintosh platform. The only hope for two co-equal interfaces is if users could freely interchange them without losing overhead, and without adding to IT's support burden. Because GNOME and KDE come from two different arenas, GNOME and KDE, that appears unlikely at best. They may come to support each others' applications, but only through a kludge, not a single API.

In an FAQ, KDE noted that the GNOME Foundation simply changes the way GNOME is governed, controlled, defined and released. "We have a similar mechanism for accepting donations...and we already have a system for releases and standards." The group hinted that, in opening the door to companies, the GNOME camp has compromised its standing, and said that there will never be a KDE Foundation because "KDE has always been and always will be controlled by the developers that work on it and are willing to do the code. We will resist any and all attempts to change this....the KDE project is possibly the only large Open Source project that has neither a 'benevolent dictator' nor an elected governing board (or any voting at all). In a sense, we are the only large 'pure' 'BIZARRE' style project out there....and it works. We have no intention of changing an obviously winning formula."

But if the history of Linux is any indicator, corporate support does matter. GNOME gets a boost when a cross-section of Linux-sympathetic companies lend their support to the newly founded GNOME Foundation and when companies like Sun and HP say their will bundle GNOME on their servers. With the GNOME-KDE face off, we are seeing the same cultural upheaval that Linux faced with the free-wheeling operating system as it began to gain support from the commercial sector. In this love/hate relationship, like it or not, the commercial sector support counts when you are trying to be a de facto standard.

Meanwhile, de Icaza is not looking back. The fast talking, far-traveling, high-energy instigator of the GNOME project and GNOME Foundation keeps his traveling schedule on the Web (along with a link to a Courtney Love byline on music piracy), has been profiled in Time and Newsweek magazines, and has co-founded, with Nat Friedman, the GNOME-focused Helix Code in a Cambridge, Massachusetts. ( I spoke with him by telephone.

What are your hopes for the GNOME Foundation?
The foundation is just a mechanism for getting more people working together to achieve the goals of the free software movement. We want to close the gap between proprietary and free software offerings in all computing scenarios, including desktops, servers, and transactional systems.That's where the GNOME project comes in. We've been working on things like the printing architecture, so that applications can use native printer drivers and spoolers can integrate better with the system. It's not fun to do, but it needs to be done. We've been working on imaging architecture and component architecture, which is a big deal. We've been working on a fast CORBA implementation.None of this stuff is tied to the user interface. When people usually think of GNOME, they are thinking of the thin layer on top, which is the engine's applications. But we've been mostly focusing on the infrastructure underneath and building applications on top of that infrastructure. I don't think that just having a nice interface is going to cut it.
Some would argue that these are core operating system issues. How does your work differ from that Linus Torvolds has done on Linux?
Here's the confusion. The only thing that Linus does is the kernel. What people refer to as "Linux" is actually the Linux kernel, a bunch of utilities, pieces of GNOME, pieces of X-windows, etc. Linus is just working on the core of this system and has practically no input or impact outside of this.
How much is now GNOME tied strictly with Linux versus those other OSs?
Very little. We have developers working on BSD. A few weeks ago we had people working on NetBSD, on FreeBSD, on Solaris, on HP-UX, on IRIX, as well as on Linux. It's pretty much the same GNOME environment and the same GNOME architecture available across all of these UNIX systems. So the effort is pretty much independent of Linux.
But people have tended to view them in the same box. Will that ever change?
If you are shopping for a completely free solution for your desktop, I think Linux and GNOME are a very good combo, as are FreeBSD and GNOME, or NetBSD or OpenBSD. They are all pretty good combinations of software.
Let's talk about your company, Helix Code. You are planning to derive your revenues from a subscription model. What does a subscriber get?
That's precisely the part that we're working on. Right now we have the Updater service-a free service. But we're going to expand the number of services.
What does Updater do?
Updater provides an easy way for you to update your system to any new versions of the software. Here's the issue: although you keep hearing about how open source is wonderful and how all the bugs get fixed so quickly, and how the improvements are deployed quickly--- all that comes in the form of patches. And there's no easy way for people to install these patches. That's what Updater does. You can configure it to work automatically, although, for privacy reasons, many people may not do so. Or you can double click on an icon, which tells you what new updates are available.
Other companies, like Eazel, are also looking at the subscriptions. Is this going to become a viable model for raising revenue?
Yes, I think so. With Open Source software, you cannot sell software licenses because anybody who receives a license can redistribute it. So selling permission to use software as a business model doesn't work. Eazel is working on a set of services and we're working on a different set of services. But that's what we're both moving towards.
What about hosted applications?
I'm not sure exactly what a hosted application is. I keep reading about that, but every time somebody explains it to me, it sounds like a stupid idea. Let me see if I've got it right. It's an application running remotely over the Web and you pay to use this remote application. It sounds stupid to me. I see the Web as this tremendously slow, somewhat unreliable device. It's usually pretty good, but last night I was trying to buy milk from when my IP decided not to work. So I wouldn't want my spreadsheet to be running on a remote server I don't have access to.The other point is that I refuse to use an application that's not going to let me make improvements or changes to the application, which is the whole problem I have with proprietary software.
Windows still has an enormous library of applications. How is the gap closing?
There's a lot of people who are porting their applications to Linux. But speaking personally, any software which is not free doesn't exist for me. If somebody ports a proprietary application to Linux, it doesn't have any effect on my life. As far as I know, Oracle for Linux doesn't exist, and the only databases we should be pushing, from the free software perspective, is MySQL and Progress. My approach is that if it's not free, it doesn't exist and if people need a product, we need to write a replacement.
Some old geezers in this business would say that's a radical concept, that people are entitled to the financial benefits of intellectual property.
That's fine.
But you're saying that in your world that's not...
It's not an option for me. I'm not going to use any proprietary software. If there is no solution for me, then I'm going to write one. That's one option. If not, then I'm not going to use that piece of software. I've been beaten too many times in the past by companies that decided to stop supporting a project, by companies that decided not to add a feature I needed because it wasn't in their interest, companies that didn't release bug fixes that were affecting me, by release schedules that are off by two years---with no way of obtaining a fix in the meantime. The problems are too many. If the application is not released from an open source license, then at some point, an open source version is going to get written.
From your perspective, how is the free software movement playing in Japan?
I know there is a lot of work in Japan. But it's very strange. We get patches from the Japanese people in bursts, but they tend not to participate on our mailing list. Maybe they participate more on their own mailing list, but they tend to be very quiet, at least on ours.
So you are you seeing the development work, but little interaction.
Right. They develop a lot, but they don't participate on many of the decisions.
Do you think it's a language gap?
Maybe. There's a lot of European contributions, a lot of American contributions, and surprisingly, a lot of Mexican contributions for the GNOME project, as well as from Russia and Australia. I don't know what's going on with the others. Maybe it's a language gap, or maybe they just prefer to work as a team---producing a patch and sending it.

With StarOffice, Sun Enters the Free Software Movement

Sun Microsystems has given a big boost to GNOME by integrating StarOffice suite into the GNOME Office Suite, thereby giving GNOME users a spreadsheet, word processor, presentation and graphics packages. The result is another step forward for Linux on the client, giving the OS more equal footing with Windows and its enormous application library.

In a move similar to the one Netscape took in transforming its browser into an open source project, Sun has put the source code of StarOffice 6.0 online and established an open source development project called Developers will have their hands full. StarOffice consists of some nine million lines of code, licensed under the GNU public license as well as the Sun Industry Standard Source License (SISSL), which requires parties to adhere to Sun's APIs and compatibility test. The project was announced last July, and is coordinated through CollabNet, a company that specializes in coordinating open source projects. StarOffice consists of 75 modules grouped into 18 projects.

While nobody is claiming that StarOffice yet has the functionality of Microsoft Office, Sun's John Heard argues that the package is a superior alternative to Microsoft Works, Microsoft's low end office suite. Not only does StarOffice work better, Heard says, but the price is right: free. StarOffice's user of XML as the default documents format will help ensure universal exchangeability between documents from different applications---a problem that has plagued users in the past. Already, StarOffice can exchange documents with Microsoft Office. Given Microsoft's entanglements over whether the company is an illegal monopoly, this interchangeability is likely to last.

Sun regards StarOffice as a multi-platform office suite solution, with versions available for Linux, Windows, and, in the next release, Mac OS. That move has given StarOffice the room it needs to take off. "We've distributed over 15 million copies since we made the acquisition," Heard says. "Just go down to Fry's [a chain of Silicon Valley computer stores] and you'll see it already packaged into many of the PCs that you buy off the shelf. It's true that a large number of enterprise customers have heavy investment in Microsoft files and macros, making it a major undertaking to shift to an open source alternative." And so the groundswell is happening at home, where, Heard believes, it will trickle up to the commercial world. "My kids at home use StarOffice now. The kids at school are using it, as are the universities---not only because of the price but also because of the functionality. The functionality is adequate. It meets their needs. It reads and writes files that have come to you over e-mail from a Microsoft Office product."

The freeware version StarOffice is certainly a boost to Linux. So why does Sun, a company that makes its living selling servers built around Solaris, care about popularizing a competing OS? Heard says that, in general, Linux is good for the UNIX community, raising the spotlight on UNIX APIs. Sun also considers Linux an important kernel platform for Java and has released a version of the Java 2.1.3 virtual PC (VPC) on Linux. And, as Sun sees it, GNOME and Linux both have a future on Internet appliances. "Approximately 60 percent of the effort going on in Internet appliances is based on Linux," Heard says, including work by Transmeta and Red Hat. He believes that StarOffice also has a place here, perhaps as a way to reply to wireless e-mail.

Is Heard really suggesting that a full-blown office application can reside on an appliance? "Moore's Law changes economics, not just performance," he says. "Because you can get more power and memory into a small device, you now have the capacity to shrink-wrap code into the read only memory of that device. And when you want to migrate to a newer version, you do to a major software upgrade. You literally discard the device. That's exactly what appliances are about."

But what about Microsoft, which has struggled for a market in shrinking its office application suite into the handheld Windows CE environment. Heard argues that Microsoft applications are inherently difficult to shrink. "When you write a Microsoft foundation class-based application, it expects to have the desktop there. Appliances, especially handhelds, work very differently. They generally don't have keyboards, substituting buttons or touchpads. So the applications aren't well-matched. The other thing is that Windows CE hasn't provided the natural compatibility between the Windows desktop platform and the Windows CE devices. So there's value in having consistency of the technology. The model we see is Linux on the desktop, UNIX on the server, and a Linux subset kernel embedded in devices. That model is very attractive to technology people."

And that's where GNOME comes in. Heard argues that GNOME can provide the foundation technologies for designing appliance user interfaces. "It is well designed for that. The traditional way of building desktop user interfaces is to build the basic interface first, and bolt the rest of the technology on. GNOME comes from the opposite direction. They built a whole suite of libraries---a foundation technology---and then put the GNOME User Environment on top. That environment is highly and dynamically customizable. One of the ways GNOME does this is through a very powerful theme engine, so that the desktop can look like Windows if you like, or a Macintosh. You can render the desktop in many different looks and behaviors because the GNOME foundation structure was designed that way."