[matplotlib-devel] curvelinear coordinate support in axes_grid and a possible refactoring of mpl.

Hello,

I recently committed a support for a curvelinear grid in the axes_grid
toolkit. What it does for you is to draw a ticks and grid appropriate
for a curvelinear coordinate system, as in the example below.

http://matplotlib.sourceforge.net/_images/demo_floating_axis1.png

The main motivation behind this work was to display astronomical
images with matplotlib. However, I included  this in the axes_grid
toolkit hoping that this could be also useful in other fields.

The reason I'm posting this email is related to a possible refactoring
of the matplolib,  that John once mentioned, to better support the
ticks and spines. To support the curvelinear grid, I had to create a
customized  Axes class, and I believe that some of my effort overlaps
with the refactoring that John mentioned. In the linked document
below, I tried to describe the changes I made for this.

http://matplotlib.sourceforge.net/mpl_toolkits/axes_grid/users/axislines.html

I guess some of my idea or implementation may be considered to be a
part of the refactoring (if it ever happens :) ). Or, at least, I hope
that the issues I had with mpl's current Axes implementation are
reflected in the refactoring efforts.
Well, I just wanted to toss around some ideas and hear what others think.

Regards,

-JJ

 ps. Are we having a sprint during the scipy conference by the way? I
may join remotely.