X Developer's Meeting

I agree with your sentiments (I think we all do), but you're mostly talking about things that need to be handled further up or down the stack. Your xmodmap example is perfect; that's a gaping hole in Gnome - not X. (There is some very recent work on the problem in Gnome CVS if you're interested, it's called "gkeycaps." ) Any solution at the X level is going to be suboptimal there.

Similarly, take a look at System Settings -> Display (system-config-display) on a RH system. Desktop level, again - where it integrates best. Better yet, take a look at Robert Love's "Project Utopia" and the components thereof, running the gamut from kernel to desktop. The integration and seamlessness you're looking for ARE being worked on, with a clear goal of increasing the plug-and-play factor and minimizing all the voodoo - including X voodoo. Take a look at some of the use cases:

http://primates.ximian.com/~rml/blog/archives/000395.html

Hell, lots of things are wrong with X, and I'm glad to see the whole messy X.org thing starting to settle down, with hopefully much good work on the horizon... But I think most of the *configuration* issues in Linux are 1) system-wide, requiring system-wide solutions like HAL, D-BUS, intelligent policy, etc., and 2) will have front-end solutions that are desktop-centric - in our case, Gnome desktop-centric (gkeycaps, gnome-volume-manager, system-config-display, blah blah blah).

There's my two cents. Sorry, no refunds.