I always enjoyed XPath as a lighter-weight mechanism for querying trees. It turns out that there are a set of pretty straightforward changes that can be made to XPath to have it support graphs. To wit: add a notion of a named 'traverser' (depth- vs breadth- search; pre-, in-, or post- order; how nodes are marked; etc.); then, change the notion of 'parent' to 'in' and 'child' to 'out'. XPath can then be implemented as syntactic sugar on top of this lower level "Graph Path" mechanism.
I'd always intended to convince my advisor to have the grad students do a graph-path implementation on top of boost::graph, in his generic programming course; alas, never happened. The 'masters' version of this would be to do a lazily evaluated version of graph path in C++ and Haskell to "teach them the students a lesson".
I'd always intended to convince my advisor to have the grad students do a graph-path implementation on top of boost::graph, in his generic programming course; alas, never happened. The 'masters' version of this would be to do a lazily evaluated version of graph path in C++ and Haskell to "teach them the students a lesson".