I will write more about conditions once I am happy with them ;--( At the moment they are processed by a very poor regex-based parser. I need to write a real XPath parser, probably using Parse::Yapp. Then the doc will be written... it will be the XPath spec! Except that this will be "navigational XPath", and that I will also have to properly identify the subset of XPath that can be used for twig_roots (all we know about an element is its ancestors, and maybe their attributes) or twig_handlers (we know everything that happened before the end tag of the element),so I know when to refuse an XPath expression if it is in the wrong context.
As for now, the doc for the development version is now generated through perldoc and is a little better x-linked, so I will add links to the description of the conditions from all navigation functions.