Hi, Okay, it seems that there is one major issue that might need to be resolved before some effort can be made to structurally start documenting. The issue is where to put the documentation. I said before that it wasn't my call, but since I'm planning to do some documenting, I'd like to add my view to it. My plan is basically to do the following (for R1). I want to ask the developers of the classes to write a bare-bones documentation of their classes. Then I'd like to go over the documentation with a team of documenters, flesh out the descriptions, add some examples, etcetera. This is a good approach because: a) It will create an uniform documentation because it is written by people that work intensively on this, since i) not all developers will have a good talent of writing documentation (no offence) ii) not all developers will have English as a first language b) It will keep the developers 'free' from worrying too much about documentation. I suggest we do the documentations in different files, because: a) It's likely to have many iterations, improvements. Revision control for .cpp files is revision control for the source, and not for every spelling mistake fixed. you don't want to look through a larger number of revisions to see when you might have introduced a bug. b) Some of the documentation might be done by people with less knowledge of C++, or subversion, or diff, or any combination of these. To keep the tree from busting, let's prevent the possibility that it actually can be broken by doing the documentation in different files (yes, I know, there are precautions, but we all have mistakes that happen in oversight). c) Documentation should document header files, which are likely to contain more information than source files. If I add a new inline method, where should I document it? Separate file or in the source file? No matter what you choose, you will always end up having to search for it. A separate file is a separate file. As I said before, Doxygen's parser still checks for missing methods, wrong parameters, invalid function declarations in the .dox files. That won't change. I just want to prevent a mess of different locations of documentation, for both coders and documenters. While the argument for keeping it in the source is compelling, namely that with api changes the coder will most likely update the docs, I think we can put in place a mechanism that will prevent that sort of stuff. Actually ,that's why the header of .dox files should contain the revision numbers of the files it was written against. And post R1, we'll see about that then (something like an obligatory keyword in the log message in case of APi changes, or something like that). Point is, we're not at R1 yet, and I would like to play around with the documentation a bit, without getting into other peoples water. I guess that it might be worth discussing on this single issue, since it affects the process of documenting fundamentally. After that, I hope we can start doing some work. I received emails from two people that are willing to invest time in the API documentation, and I assume it's a waste not to use their offer. Kind regards, Niels