Is this a plot3d -S+b0 bug?

While working into fixing another bug I’m finding this buggy? behavior.

julia> G = GMT.mat2grid(randn(5, 6));
julia> cmap = grd2cpt(G);
julia> bar3(G, lw=:thinnest, color=cmap, base=0.0, figname="lixo.png", show=1, p="200/10", Vd=1)
        psxyz  -So0.85u/0.85u+b0 -R0.5/6.5/0.5/5.5/-2.200175127803197/1.4241171951096083 -JX14c/9.5c -Baf -Bza -p200/10 -JZ6c -C -i0-2,2 -Wthinnest -P -K > C:\TEMP\GMTjl_tmp.ps

with +b0 we certainly are not expecting to see those negative bars

But on the other hand the man says column (3-D) extending from *base* to *z*. which is what we see, so no bug, just unexpected on first sight.