The documentation was originally maintained by hand, then we converted to an automated system based on Doxygen. Unfortunately, Doxygen was not designed for working with Lua/C++ hybrid code, so we use a Perl script to rip through our source and create a set of files that are easier for Doxygen to digest. This generally works pretty well, but very little attention has been paid to making the result usable. We would like to use a more robust system, preferably with a documentation generator designed for this task, such as http://stevedonovan.github.io/ldoc.

This would involve converting all existing documentation comments into a format compatible with the new generator.