Capability systems are obviously a good platform for experimentation with GUIs. However, it is necessary for platform success today to have a standard GUI available; for that standard GUI, I think it would be wise to specify something simple, well-understood and, if possible, widely available, so as to minimize the effort involved in porting GUI applications to EROS.
Implementing or porting an X server --- or at least an Xlib-compatible display service --- fulfills these requirements better than any available alternative. GNUStep and Smalltalk fulfill them to some extent, but not as well as X, IMHO. The Mozilla architecture does not fulfill them.
X has the advantage of supporting more video cards, especially current video cards, than any other free software hardware-level pixel-pusher.
Technophilia has been the downfall of Win95, VMS, MVS, and now it will be the downfall of GNOME. Implement the minimal required system to be useful and it has a chance of working correctly.
If we implement something experimental and cool, it may rule or it may suck. If it sucks, people will conclude that EROS sucks, not just that EROS's standard GUI sucks. If it sucks worse than X, Linux people won't want to use it.
A chain of software is as strong as its weakest link. So don't put any more unknown links in there than you have to.
--
<kragen@pobox.com> Kragen Sitaker <http://www.pobox.com/~kragen/>
The Internet stock bubble didn't burst on 1999-11-08. Hurrah!
<URL:http://www.pobox.com/~kragen/bubble.html>
The power didn't go out on 2000-01-01 either. :)