Last night I read much of Producing Open Source Software [link] by Karl Fogel, and then implemented some of his recommendations. I'd been overlooking the need to make it explicit that the Python Cartographic Library is open source, and also retooled the mission statement.
The goal of PCL is to completely replace MapServer's mapscript and become the way for Python programmers to use the C/C++ open source mapping software stack composed of PROJ.4, GEOS, GDAL/OGR, and MapServer. MapServer's mapscript module, my ward of the past few years, has an awkward, un-Pythonic API that routinely trips up programmers. MapServer is mature software and strongly obliged to be backwards compatible with old scripts, so while I've been able to fix many of its bugs and plug memory leaks, its clunky API is pretty much carved in stone. PCL, on the other hand, behaves like one expect a Python package to behave. It has actual classes, iterators, docstrings, unit tests, and warm fuzzy feelings. It is what Python programmers should be evaluating for use with new projects.
A larger goal of mine is to see something like SciPy for geospatial programming. PCL could be the cartographic component, PySAL might be the spatial analysis component. Hobu, who pointed me at Fogel's book, has been talking about overhauling the old PyOGCLib; that might provide a component for parsing and writing GML, and working with OGC W*S protocols.