On Wed, 2010-05-12 at 22:04 +0100, Leo Butler wrote:
> I've looked at the relevant bit of code and I think it
> is needlessly prevents the user from plotting a parameterised
> curve without a hack like that above.
>
> I'd like to add the patch below. Note that the patch fixes
> a bug in the present code, which swallows something like
>
> plot3d([1,1,1+a],[t,-1,1],[x,-1,1]);
>
> without complaint and without plotting anything.
Sure, you can go ahead and commit that patch. But be aware
that the resulting parametric curves produced by plot3d will be very
ugly and inefficient, since they are being plotted from a mesh of points
as if they were surfaces.
A better approach is make plot3d behave differently when only one
parameter is given. In that case, instead of using the "grid" option to
create a flat mesh that is then deformed, it should use an approach
similar to the one used in plot2d to create a curve. In that way the
helix that Krzysztof wants to plot would look much smoother.
Cheers,
Jaime