Ximian Gnome — Linux Desktops

I recently tried Ximian Gnome 2. What do I think? Well, it is definitely a good try, but it is very similar to the Gnome desktop I had on my machine more than 18 months ago. I don’t know whether it is me or it is really happening, but I find the Linux Desktop developments slowing down in the past year or so.
I have been privileged with a ‘switch’ to Macintosh computers, thus making my interaction with Linux boxes more infrequent, and even then from a server administration point of view which is not really focused on getting fancy desktop environments, but — really — has there been much happening on the linux front (at least from a Desktop/HCI/User Experience point of view) in the past 18 months? I don’t think so.
Ximian Gnome is very nice, very polished (although I did find several bugs here and there) and definitely a product I would choose over Windows for Enterprise use, considering the costs involved. But where is the Linux Desktop scene going? With both MacOS X and Windows evolving into composite-based desktops, and the industry slowly adding those same human-centric features that they claimed they would add ten or fifteen years ago (think Object orientation, the Cairo OS, the Taligent OS et al), such as file meta-data and associations, increased integration between applications etc., how is linux going to compete?

Certainly not with a few nicely drawn icons and anti-aliasing. Because this is how linux on the desktop evolved in the past year. from what I can see.
Isn’t it time people thought of a NEW and massively rethought of X Server? A new X implementation that would of course keep compatibility with the existing X11R6, but that would add compositing, (true) alpha blending, native OpenGL support (something similar to OS X’s Quartz Extreme) all in a nice and fast package? Some people have designed (in my opinion) awkward 3d desktop prototypes/window managers that do not seem very good at what they do and are merely an application of current technology. I hope it is clear that this is not what I am talking about.
Isn’t it time people started thinking about something different? Something more user friendly that still has room for that ‘Terminal’ icon somewhere in there? The ‘coding-power’ is here — this has been proven countless times. There is no financial obstacle, since most people don’t do this for the money anyway, and I am sure the interest must be there. Consider Eazel: Full of ex-Apple, HCI experts, and the only thing they produced was a very very buggy version of Nautilus, which — besides the performance improvements — is still the backbone of Ximian, one of the most popular environments for Linux (and maybe other UNIX-like* OS’es)
But to do all that you need libraries. Gnome uses GTK, a C-based library — despite the several C++ wrappers, most people use C to develop apps in it. What a shame! I like C, but I would definitely consider something more object-oriented a must for a successful GUI environment. Sure, I’d probably use C if I wanted to code the ‘Hello World’ example and was just learning the thing. And then we come to QT, which I generally like, if it weren’t for its I-wanna-do-everything-instead-of-the-language mentality that their designers must have. At least it is object oriented.
I think Linux and the world needs something new on the toolkit/libraries front too. Something that takes the good parts of all the above, OpenSTEP et al, BeOS APIs and creates something better. That’s what we need. So please no more same old gnome and kde in a new package. It’s good for enterprise use yeah, but bleeding edge techies need the desktops of the future. Now. And for this to happen, two things are needed:
In the short-term companies like Ximian should become reality, both for Gnome and the KDE projects, in order to provide some organisation and professional support in the fields of HCI/Usability and Guidelines (fields that both projects are lacking in)
In the long-term, there should be open source development of strategic software, such as the aforementioned graphics environment/server, as well as crucial support from software leaders such as IBM in this respect so that industry standard Open-Source, leading environments become a reality.