So, after I have 2.0 built and all the NetBSD sets installed, xf86cfg do
it's work and got startx going, running xclock gets a segfault:

	miyu# xclock -digital
	Warning: Cannot convert string "" to type XftFont
	Segmentation fault (core dumped)
	miyu#
	miyu# gdb `which xclock` xclock.core
	GNU gdb 5.3nb1
	...
	Core was generated by `xclock'.
	Program terminated with signal 11, Segmentation fault.
	...
	(gdb) bt
	#0  0x480689ac in XftCharIndex () from /usr/X11R6/lib/libXft.so.2
	#1  0x48070f14 in XftTextExtents8 () from /usr/X11R6/lib/libXft.so.2
	#2  0x0804a57f in Initialize ()
	#3  0x4819b3a9 in XtInitializeWidgetClass () from
	/usr/X11R6/lib/libXt.so.6
	#4  0x4819b8bb in XtInitializeWidgetClass () from
	/usr/X11R6/lib/libXt.so.6
	#5  0x4819bcbc in _XtCreateWidget () from /usr/X11R6/lib/libXt.so.6
	#6  0x4819bf33 in XtCreateManagedWidget () from /usr/X11R6/lib/libXt.so.6
	#7  0x08049d7e in main ()
	#8  0x0804988a in ___start ()
	(gdb)

Anyone else see this?

Looking around a bit, I wonder why we ship more than one libXft.so (both
2.1 and 1.1 are installed here, and also in the set lists). And ktrace
reveals a file /usr/X11R6/lib/X11/XtErrorDB is not there.

And after running fc-cache, the above problem disappears - how can we make
things work without that, or run it by default, preferably before creating
set or by including the resulting cache in the sets?


 - Hubert

-- 
Hubert Feyrer <hubert@feyrer.de>