On 7/27/05, Raymond Toy wrote:
> Robert> At this point my recommendation is to turn on the gnuplot
> Robert> option for a logarithmic plot when plot2d is given the log plot
> Robert> option and the plot_format is gnuplot or mgnuplot,
> Robert> and otherwise print a warning message and carry out the
> Robert> log transform in plot.lisp (i.e., plot f(log(x)) instead of f(x)).
>
> I think I agree with what you say.
>
> But the log option really only means sample the x axis in a
> logarithmic fashion. The resulting plot probably won't look too
> different between the log-sampled and non-log-sampled versions,
> because the adaptive plotting will make sure enough samples are taken.
Surely it's not too late for us to assign a different meaning to
the plot2d log option. Wouldn't it be more useful to the end user if
log = true means to make a logarithmic plot, with an understanding
that Maxima will do whatever is necessary to make it look right?
Perhaps I've misunderstood what you're getting at.
Robert Dodier