Sometimes (most of the time), error messages in the MinGW octave come out as binary gobbledygook. I'm not sure what triggers it, but I have a hunch that the following factors play a role:
1) Internal function that calls error()
2) Long error message (I think short ones fare better)
You don't have to try Workshop, gnuplot in command-line octave will do it too. For a quick example, try the example from "help delaunay".
x = rand(1,10); y = rand(size(x)); T = delaunay(x,y); X = [ x(T(:,1)); x(T(:,2)); x(T(:,3)); x(T(:,1)) ]; Y = [ y(T(:,1)); y(T(:,2)); y(T(:,3)); y(T(:,1)) ];
axis([0,1,0,1]); plot(X,Y,'b;;',x,y,'r*;;');
Note: the problem isn't that the example doesn't work (although that's a problem too). The problem is that the error message is fubar. Also note that in my octave, I've renamed
pgnuplot.exe to gnuplot.exe (at least I think it was pgnuplot.exe) but this happens almost all the time with the other builtins I've written.
I reported a gnuplot problem earlier, which John said was related to DEFVARS or something like that, but since I can't tell if that's the same problem, well, you tell me.