Docbook comes with stylesheets to transform it to xhtml, pdf, etc., which can be applied with a post-commit hook. It also knows about XInclude, so you can have separate files for chapters, but still process the document as a whole, which automatically takes care of infra-document links, such as cross-references, citations and indexes. I’ve been using it for lecture notes for a few years now, replacing a LaTeX setup.

As an aside, I’m also involved with a project to digitise a series of medieval French texts into another XML format (TEI) and again, svn has allowed very fine control of collaborative edits to the texts, without risk of unrecoverable overwriting. Is that the most obscure use for svn you’ve come across?

Gary (University of Exeter, UK)

Obviously you can separate out chapters in to separate files and link them to a TOC-like index page, but this in itself can introduce further complexities. For example, cross-referencing between chapters, or remembering to add your latest chapter to the TOC.

An alternative would be to use LaTex, which can produce either PDFs or HTML. It also has the advantage of, like HTML, being a mark-up language based on plain text. Therefore you can still use the text-comparison tools you are used to for merges, etc. This does then require that each user wishing to create either the HTML or PDF files must have a suitable LaTeX environment set up. However, there is no reason why these “compiled” document is not available from a separate location which is generated on an SVN submit - provided the SVN-server has LaTeX installed….