For my in-progress NetBSD/sparc stuff, I've done mostly obvious changes. 

(1)	anything which assumed __NetBSD__ == i386 now
	has to do an appropriate switcheroo.  Of course in X-land
	defined(anything) usually fails because they deliuberately
	undef it. Once you create the appropriate AnythingArchitecture
	symbol they *want* its not too bad..

	This means the .tmpl, a .cf and imakemdep.c needed some frobbing.
	I actually made a NetBSD.cf to avoid pain, and shifted some
	'vendor-common' stuff from x386.cf into it, removed 99% of the
	rest, added some garlic and nuked it in the oven.

	then SparcArchitecture and SparcNetBSD become viable
	#if instances inside X-land to match i386NetBSd.

	A grep check down an XFree86-2.1 munged X11R5p26 tree reveals
	very little that is even remotely a candidate for problems
	apart from server/ddx/... Most of the code was #if BSD > 1993
	stuff which works fine all underlying architectures of NetBSD.

(2)	BUILD_SERVER = NO

	because (a) I'm still getting client side right
		(b) there is some stuff missing from /usr/include
			and I don't want to pollute the tree
			with StunOS code

so far it make Worlds into a complete build of Makefiles and is
making lib as I type. This is like meaningless. not until I see
the xterm up on the screen do I believe its workable.

I suspect it'll be alright. If I can get a view of the .cf .tmpl and
imakemdep.c/h I'd be more confident, but I really shouldn't have been
so rude about Xfree86 to Theo. Really guys, its not that much to change.

-George

------------------------------------------------------------------------------