Rudder

Maybe the best solution :

The content is a collection of .txt files, ordered into directories.

During the build of the documentation, a script is generating the actuel collection of .txt file to be included in the documentation, that is: which files, and in what order.

It will be really easy to tag some of these files so that the script will ignore them, or add them only under certain circumstances.

I see several ways to achieve this :

- symbolic links : files are created in some "fake" hierarchy, and then included into the actual hierarchy (same as sites-enabled for apache2 in Debian). Advantages : as many "actual hierachy" as needed. Drawbacks : maintenance issues expected.

- tags in name : add a "WIP" suffix to the files which are not ready. Advantages: keep it simple. Drawbacks : too much tags will kill the tags...

This ticket is meaningless.
We have PR for unfinished documentation.
We have merge for finished documentation.