I say this because I do not see xsltproc, for example, called anywhere during build, and I think if this problem were already solved, I probably would.
While I cannot yet volunteer to work on this problem, a bit of status on where it is would be helpful.
Some things I believe and/or have executed:
- Being new to swig and python and its uses in given situations, I'm still a bit disoriented wrto which gnuradio-core ("core") piece dovetails into what other piece.
I think it goes like this:
In the beginning, there was C++. From that follows python interfaces into C++ via swig. From there we find utility python code in the core written by some human and not a code generating tool. And from there we find python application code here and there.
- True or false (and why so either way): the documentation task above must be solved before we can produce, presumably with epydoc, html docs for *any* core python code.
- Having compiled the core with --enable-doxygen, I notice that there are xml files in the output, presumably the ones mentioned in the mailing list post.
- I notice some %feature("autodoc","1") instances in some .i swig files. It appears we are incrementally solving the problem that is the subject of the mailing list posting above.
- I see many .i files that do not contain "autodoc"/"docstring" directives. Not sure why, or what the implications are. (Notice how
this can mean "I disagree with what I see" or "Someone tell me why I see what I see, because I just don't know". I mean the latter.).
- assuming we get to swig_doc.i files mentioned in the post, would there be one swig_doc.i file for every other .i file in the core?
- a general backing question: a python module can have code that originates in high level python code (a .py file) as well as code that comes to us by way of a swig interface. gnuradio.gr
, e.g., has a flow_graph via 'native' python code as well as a file_sink that comes to us via swig. Presumably we lack input that can generate docs for that swig interface object.
- if this problem were already solved, can I get an example of an html file that would exist as a result? Just the file name, not the contents.