Question

In the algorithm I’m currently implementing, there is this line (where u is a vertex in a graph, and Pred(u) are all vertices having edges pointing at u):

for all s ∈ Pred(u) ∪ {u}

The Pred(u) part I translate into boost::graph code like this:

boost::graph_traits<Graph>::in_edge_iterator in_begin, in_end;
boost::tie(in_begin, in_end) = boost::in_edges(u, G);
for(boost::graph_traits<Graph>::in_edge_iterator i = in_begin; i != in_end; ++i) {
    // Do stuff
}

For now, I’m doing the Do stuff stuff outside of the loop for u explicitly, but I’d like to do it in the for loop. Is there some trick to create the iterators as if u was returned from boost::in_edges?

Answer 1

I think the solution you are using is alright (as long as the Do stuff code is well factorized).

However, if you often face such issues, you can have a look at Boost.Range, a library for manipulating ranges of values instead of iterators. Here, you could use the join function to obtain the union of your two ranges (the result of boost::in_edges and u).