The CHM I was talking about, is proposed as an output format, not the source format (it is compiled anyways...).
As we speak, Ive downloaded an XML/XSLT OS editor: Butterfly. Ive also *finally* understood how docbook works, and Ive downloaded both the XSLTs from the DocBok open repository, as well as Xalan-C, for the XSLT processing.
Since Im interested in helping out in the "Code::Blocks documentation project" as well as for my own project at sourceforge, Ill be giving it a shot, and see what I can come out with in the next couple of days, maybe I shall submit a little "test" to you guys and get some feedback.
Using DocBok is great as It provides the flexibility of XML with the power of XSLT transformations (HTML, PDF, CHM with the help of the freely available Microsoft HTML help toolkit, you name it).
And I agree there should be a downloadable HTML tarball with latest, bleeding edge changes, as well as an online HTML with changes as well, and I would aim to provide the code::blocks release with a CHM file. Why not PDF (well, an available PDF can also be made...) and why CHM, well, CHM I think is better for documentation (more tidy, no acrobat reader needed), and besides, ive looked at the samples in wxwidgets, and In fact, IT DOES support CHM for Linux flawlessly, no need for aditional 3rd party shared objects or anything like that, so Its cross-platform, and we have the advantage C::B is made with that toolkit so...
This is meant for the end user docs, now, probably theres another part of C::B that should be taken in account: the sources (maybe starting with the SDK). For this, I think doxygen does a great job. Ive just seen the CHM bundled with irrlicht latest version, and it rocks. Its probably a "customized" version of doxygen output (with the irrlicht logo and everything - I thought doxygen should only output that horrible logo of itself).
While Im aware probably you need to add "special" comment blocks for the sources (just as javadocs, which Ive used in the past - very easy to setup though), IIRC, it can extract information from undocumented sources (like the C::B ones), so it should be fine at least for a start...
Stay tuned...
Cheers
[edit]
Butterfly plainly sucks. In fact Java based apps suck. Slow, clumsy, crappy GUI. Im hitting myself against my monitor for those 12 Mb, and Im sure my 56k dialup wasnt so happy either.
The journey continues...