I've done a partial implementation of log plots. The data is transformed and shows up, but I haven't yet implemented smart tick locations or tick labels. (And I haven't checked the auto view setting abilities.)
I don't know how others feel, but I prefer my tick labels on log plots to say "10" and "100" rather than "10^1" and "10^2", so I'd vote for this as the default behavior for matplotlib. (At least for values < 1000 or so.) I mention this because matlab seems to work the other way.
So, here's a patch to the various bits I changed, and a simple example that makes two plots.
John, you're welcome to include this in matplotlib as long as matplotlib remains under the current license.
log_plots_pt_1.patch (9.04 KB)
log_axes.py (223 Bytes)