18 July 2002

It’s 11am the first day of the Boston GNOME Summit and I’m sitting here in room 10-250 at MIT listening to Jeff Waugh’s post mortem on the GNOME 2.0 release process having just given the "Opening Address," which, calling it an "Address" betrays a savage and inappropriate conceit, since essentially it was just a half-hour hung-over extemporaneous rambling about how we as a project need to deglorify architectural work and focus on building a functional & rich & well-integrated desktop for users.

Jeff Waugh in 10-250.

It is a bit funny being back here at MIT. My last clear memory in this room is falling asleep during one of the few early morning classes I attended, a freshman-year solid-state physics course, and then, slumped over in the chair no doubt with mouth open and a silvery trail of saliva peeking out of one mouth corner and making its way down the side of my face, sleeping for what must have been two or three hours, through a biology course and perhaps another and then with a horrifying start waking up confused and foggy-headed in what appeared to be a diff eq class and in the front of the room a German professor with a thick accent was making a joke of some kind no doubt unrelated to my predicament but when the entire room of three hundred some fresh-faced students began all at once to laugh I was gripped with fear and bolted from the room.

At the summit.

(Later in the day.)

One of the things about GNOME is that there exists this whole set of active & vibrant projects which sit on what you might call the periphery of the “GNOME community,” these large satellite projects like Galeon and Gaim and GStreamer, and historically we have done a terrible job of getting all of these projects to work together, to consider themselves a part of the core desktop. I think this is in part a tendency to clique in the GNOME project and in part a desire of those independent projects not to depend on core GNOME libraries, not to be considered part of GNOME so that they do not lose potential users who might choose KDE. There are other reasons too, and other interactions happening here, but I think that these are some of the principal ones.

And the effect of open source cliquishness is that the user loses: the desktop as an umbrella project misses the opportunity to do the kinds of deep integration that make for a rich, fully-integrated user experience. Things like being able to right click in the file manager and send a file to a buddy with your instant messenger or to have a unified set of bookmarks between your browser and your file manager.

Chris Blizzard.

And so in scheduling this conference, in organizing it and assembling it, or, perhaps to be more accurate, "throwing" it, I have made an attempt to bring together people from these outlying projects, and it is interesting to see how that is playing out.

The energy level here is very high (though perhaps that is in part for me a personal high) and this morning I was surprised to see after Jeff Waugh’s talk a rational discussion ensue between the sixty or seventy people assembled there in the room, large groups of smart and opinionated people usually being incapable of engaging in those sorts of activities without undergoing a rapid devolution toward stupid and partisan infighting. But there was none of that this morning and in fact I think there was a sense that we are all in fact on the same page, which to me even after the positive experience of GAUD3C in Sevilla this April, was a pleasing surprise.

Thomas and Andy, two of the GStreamer guys.

Thomas Stichele gave a fantastic talk on the GStreamer architecture this morning. He is an excellent speaker — intelligent, clear and well-prepared — and GStreamer is a rich architecture for manipulating media streams. It allows you to create and string together sets of elements which can source, modify, and sink video and audio streams. You can connect these elements in arbitrary topologies, and I was quite impressed to learn about and see the pipeline editing GUI which you can use to point-and-click to build pipelines, save them as XML, and then programmatically load and use, à la Glade.

Dave Camp held a workshop on extending Nautilus through scripts & views, which I did not attend but which people seemed excited about. Nautilus does have a rather powerful extension architecture, and so hopefully we’ll start to see some of that actually get employed in the next few weeks.