Portrait of Edd Dumbill, taken by Giles Turnbull

Subscribe to updates

Feed icon Atom or RSS

or get email updates

What I make

expectnation
a conference management web application


XTech Conference
a European web technology conference

The other side of the desktop

I've just spent the last two days attending the Desktop Developers Conference, focused on free desktop environments. In its first year, DDC has been a useful event, and contrasts in interesting ways with GUADEC.

Havoc Pennington

Havoc Pennington gave the opening address, a worthwhile ramble through the issues inherent in the task on which we're engaged: creating a desktop platform. Chris Blizzard took some notes on what Havoc said, so I won't regurgitate my notes here in addition.

I was interested in the areas in which Havoc thought we have to improve. An area he mentioned in particular was that in order to better support the user, the desktop needs to work more in terms of "user-interesting objects". Users don't work in terms merely of files, but of emails, people, conversations, documents, appointments, notes and so on. Havoc pointed to Microsoft's WinFS as something that was meant to enable this kind of manipulation of files. It is pretty much what I was trying to get at in the first half of my desktop metadata talk at GUADEC.

Havoc also raised the issue of diverging options for the free desktop world: either to emulate Windows, or build something alternate on its own terms. It was his view that it's easier to win on real user value than to be forever playing catchup with "technical castles of unbounded size." This despite the fact that many open source developers tend to think it's easier to copy features. This being open source, both will probably happen.

There's still room for innovation, said Havoc, in pretty much the same way Nat Friedman did at GUADEC. I was happy to hear him say that.

Havoc talked a lot about freedesktop.org, and it gave me a much better sense of what it is. Freedesktop.org and X.org are building a substrate for desktop platforms. A "backend for toolkits". Rather than aiming for an unattainable Utopia of integration, it's aimed at incremental sharing of code.

Though tactfully not mentioning the Mono/Java debate explicitly, Havoc did point out that it was somewhat of a Holy Grail to be able to share the object, string and container classes among the various desktop systems.

He also had a good point about Linux distributions. It's at the distro currently that the various pieces of integration between the underlying OS and the desktop software is being done. And in each distro different decisions are being made, and different bits of work done to accommodate this. It would be desirable to push this integration work somehow up a level so there is design work being done on a holistic Linux-plus-desktop system at a point before it reaches the vendors. It's good to see some of that is starting to happen at freedesktop.org.

Everyone else

The rest of the talks were very much a grab-bag of everything-but-GNOME in the Linux desktop world. I didn't make as detailed notes as for Havoc's talk, but here's a rough summary of some of them.

X Development at freedesktop.org. Encouraging progress being made at cleaning up X and making it modular. Sounds like real fruit will come from this.

ROX and Zero Install. Neat way of getting round the intrusiveness of installing software on Linux systems. Works well in the controlled setting of ROX, some questions about how it would play in a larger world. Not advisable to just use vanilla upstream tarballs for packaging.

Uniconf. Subsumes all known configuration systems in one API. Seems handy for enterprise management stuff. As a developer of just GNOME apps, it's hard to see why I should use it. Bigger != better.

Multisync. Big take home for me here is that GNOME hackers should give Multisync more love. Often hear UI criticisms of Multisync: the developers seemed very open to receiving help with the UI and interaction model. Exciting development is OpenSync, an attempt to bring together the diverse synching solutions the free desktop has at the moment.

WINE. Basically a bunch of criticism on what GNOME and Linux disitributors need to do better to support ISVs. Valid points, although I was a bit put off by the carping delivery. But there's an important message that came through repeatedly: we need to document, document, document and maintain clear expectations of API stability.

GCJ and the Desktop. Overview of the GCJ work. I had actually thought GCJ more mature than it seemed to be from this presentation.

Eclipse. Making a platform from their toolkit and components. From where I sit, it looks like GCJ+Eclipse is roughly like Mono+Gtk#.

Getting to the Next Web. Mozilla guys on large rant at problems of W3C and this is why they're taking a different route with the WHATWG etc. Would have liked to see more tech and less rant. Also raised the GNOME documentation issue, something close to my heart. GNOME really needs a good developer relations programme. Dangled the prospect of better embedding of Mozilla. How GNOME, and now Mono, needs this. I will dance on the grave of GtkHTML.

Conclusion

Aside from the general wonderfulness of freedesktop.org, the best bit about DDC was the fact that there were a bunch of X developers there, and it seemed like a forum for good communication with them. I think the conference suffered from zero Novell and Sun participation, but so soon after GUADEC (and just before OSCON) I can see why.

blog comments powered by Disqus


You are reading the weblog of Edd Dumbill, writer, programmer, entrepreneur and free software advocate.
Copyright © 2000-2012 Edd Dumbill