Done:
  1. Implemented comment generation for enums in java (for all of 4 types of them), added test cases to check it. Right now they fail because of the incorrect bison parsing code, I'm working on this.
  2. Improved integration of comments with autodoc and docstring features in python, now autodoc test does not fail and doxygen comments work also for biultin cases. Also all python tests pass now (there was a problem with autodoc test).
  3. Added comment parsing switching on or off, with the help of '-doxygen' and '-nodoxygen' flags in both python and java modules. It is on by default, switching it off also now strips all the comments before the bison parsing code gets to them, so that unsupported comments placements will no longer crash swig.
Commits:
  1. http://swig.svn.sourceforge.net/viewvc/swig?view=revision&revision=13182
  2. http://swig.svn.sourceforge.net/viewvc/swig?view=revision&revision=13183
  3. http://swig.svn.sourceforge.net/viewvc/swig?view=revision&revision=13184
  4. http://swig.svn.sourceforge.net/viewvc/swig?view=revision&revision=13185
Next week:
  1. Deal with errors in parsing code, it still needs much work. The structure of the code right now makes it impossible to properly assign comments to complex structures, like enum elements. And as of now there is some amount of copy-pasted identical code that needs to be refactored. I've posted a question about it in the list, but nobody have answered so far.
  2. Refactor comment generation code, JavaDocConverter and PythonDocConverter classes. They need to be totally rewritten.

--
Regards,
Dmitry Kabak