[GRASS5] 3D Rasters in current CVS

[Benjamin Ducke]

r3.mapcalc seems broken in current CVS (?). Even the most simple
operation like

result = input * 10

gives this error:

FATAl ERROR: G3d_getDoubleRegion: error in G3d_getTilePtr

NVIZ is giving me some trouble, too. I don't understand what
information NVIZ uses to position a 3D raster map in space.
I have tried loading in a plain surface with height range 4 to 9
meters and a 3D raster with Z boundaries set to the same values.
However, the latter is being positioned well below the surface
(the data range of the 3D raster is about 1.0 to 3.0).
The X/Y positioning seems correct.
I can change the Z position manually but am still unable to display anything
using the "slices" method. ISO-surface extraction seems to work.
Any ideas, why this is so?

I would really like to use voxels/3D rasters in GRASS and volunteer to write
a tutorial on it, but I feel currently very vague about some aspects of
implementation.
First, I would like to understand the working of r3.in.ascii a bit better:
am I right in thinking that every raster map in the input file defines the 
lower faces of one level of voxels and that voxels are stacked from bottom to top?
Can 3D rasters also be integers or only fp?


Thanks,

Benjamin.