I've posted my proposal to google-melange and duplicating here.
http://www.google-melange.com/gsoc/proposal/review/google/gsoc2011/darkproger/1
http://share.darkproger.net/netbsd/sysmodel/proposal.html
Looking for your inquiries!
What's wrong with outputting mdoc instead of XML? It preserves
semantic information and has an existing parser (in base).
One of the goals is to output (or, rather, augment) mdoc documentation
from XML specifications.
What's the point of XML in the first place, then? If the goal's to
generate documentation a la doxygen (question 2: why don't you just use
doxygen, which also has support for extracting code structures?), why
don't you just push it straight out into mdoc? What's the point of the
intermediary? Keep in mind that mdoc can push out XHTML and so on
without any further schemas or transform tools.
> ; By the way, is it possible to dump/pretty-print the
> ; mdoc AST? It will be really nice to use these libraries for this task.
libmandoc.h doesn't allow for the programmatic creation of mdoc manuals,
just parsing them.