On 13-Nov-2006, Matthias Brennwald wrote:
| I just went back from Octave 2.9.9 (from HPC) to Octave 2.1.72 (from
| Fink), and the below issue with 3D-plotting vanished.
| > x = [1:10];
| > y = [1:20];
| >
| > [X,Y] = meshgrid(x,y); Z = - X;
| > subplot(3,1,1)
| > mesh(Z)
| > __gnuplot_raw__ ("set logscale x;\n");
| > replot
| >
| > subplot(3,1,2)
| > mesh(x,y,Z)
| > __gnuplot_raw__ ("set logscale x;\n");
| > replot
| >
| > subplot(3,1,3)
| > mesh(X,Y,Z)
| > __gnuplot_raw__ ("set logscale x;\n");
| > replot
| >
| > *****************************
| >
| > I'd expect three identical plots from this, with the x axis scaled

`| > logarithmically. However, the x axis is NOT scaled
``logarithmically.
``| > Even worse, plot 1 looks like what I'd expect (Z values range
``from -1
`| > to -10), whereas the plots 2 and 3 are completely wrong (Z values
| > range from 0 to 20 or so, and there's an oscilation in the y
| > direction). Also, plots 2 and 3 look similar to each other. Still
| > worse is that if I run the code a second time, plot 2 now looks

`| > similar to plot 1 (Z ranging from -1 to -10), but plot 3 still
``shows
`| > Z values from 0 to 20 with the oscillation in the y direction.
I think these problems are fixed in the current CVS sources of Octave.
jwe