Table of Contents

Gnuplot.py: Gnuplot  

$Id: TODO.txt,v 2.3 1999/10/14 03:16:14 mhagger Exp $

Some ideas for future work. I don't know whether I'll get around to any of these items; some of them are probably better left undone anyway. Of course if anybody else wants to take one up...

  • Implement missing values. Gnuplot allows you to specify that values are missing from data files via the `set missing' command. I'm thinking about implementing this as an optional mask argument to the Data and GridData PlotItems. (Comments?) The real benefit of missing values is that (I think) gnuplot can still put a mesh on a surface even if the surface has holes.

  • Add a more systematic way of setting arbitrary gnuplot options. One cute possibility that occurred to me (maybe too cute?) would be to implement a magic Gnuplot.set object that could be used like:

    g = Gnuplot.Gnuplot() g.set.data.style(linespoints, 1, 2) g.set.pointsize(2.0)

    which I think could be accomplished with __setattr__ and __getattr__. But this is not a high priority; it's easy enough to use the __call__ method to set arbitrary options.

  • Figure out how to suck gnuplot error messages back into Python and turn them into exceptions. This would be tricky: for one thing, not all gnuplot output is indicative of an error. Moreover, implementation on Windows and Mac would require additional effort to change their GnuplotProcess classes. The output from the show command could be processed in a similar manner.

  • Implement multiple graphs on a single page. I haven't used this feature of gnuplot much myself, so I'm not sure how difficult it would be to build it into the Gnuplot.py framework. Ideally one would like to be able to spawn multiple GnuplotWindow's from a single Gnuplot instance, each one behaving much like a Gnuplot object. Or maybe treat sub-windows as a kind of meta-PlotItem that can itself contain PlotItems.


Table of Contents

This document was automatically generated on Mon Jul 29 17:26:13 2002 by HappyDoc version 2.0.1