Hi, I am running into problems using bundled properties and the newly implemented subgraph for them. In my experience, it seems they are working correctly with the root subgraph, but they don't translate correctly to the children. The code compiles, meaning there is the interface, but I get segfaults at runtime and I think they are related to that problem, for exemple: 7676,0F0F,5050void tree_gibbs_sampler(Graph & g, 7676,0F0F,5050unsigned 7676,0F0F,5050int max_steps, 7676,0F0F,5050unsigned 7676,0F0F,5050int burnoff) { 7676,0F0F,5050typedef 7676,0F0F,5050typename graph_traits<::vertex_iterator VertexIterator; 7676,0F0F,5050typedef 7676,0F0F,5050typename graph_traits<::adjacency_iterator AdjacencyIterator; 7676,0F0F,5050typedef 7676,0F0F,5050typename graph_traits<::vertex_descriptor VertexDescriptor; 7676,0F0F,5050typedef 7676,0F0F,5050typename graph_traits<::edge_descriptor EdgeDescriptor; AdjacencyIterator w,w_end; VertexIterator v, v_end; VertexDescriptor u; subgraph < subgraph_g; copy_graph(g, subgraph_g); copy_graph(g, f); cout <<<< g[0000,0000,FFFF0] <<<< endl; u = vertex(0000,0000,FFFF0,subgraph_g); cout <<<< subgraph_g[u] <<<< endl; // here the last two line produce the same numbers... // here I do something to the root subgraph and produce childrens... 7676,0F0F,5050for ( tie(current_tree, current_tree_end) = subgraph_g.children(); current_tree != current_tree_end; ++current_tree) { 7676,0F0F,5050for (tie(v,v_end) = vertices (*current_tree); v!= v_end; ++v) { cout <<<< (*current_tree)[*v] <<<< endl; } } // The last for loop produces strange results, the bundled properties (here it is just adress to pointers) are not at all consistent anymore... (with the ones in the original graph or even root subgraph) So I think there is a problem with the current implementation. Jean-Noël Rivasseau