The `m_eproperty` are not same in subgraph.

[boostorg]

The `m_eproperty` are not same in subgraph.

[anadon]

get two iterators and cycle from one to the other to find an edge with index i

Please clarify

convert it to g1's local edge descriptor

One reason why I want to replace BGL.

So are you saying that the edge between two vertices in a subgraph is not the same edge as between those two vertices in the full graph?

[kenkangxgwe]

get two iterators and cycle from one to the other to find an edge with index i

    typename subgraph_t::edge_iterator e1_i, e1_i_end;
    typename subgraph_t::edge_descriptor ed1;
    for(tie(e1_i, e1_i_end) = edges(g1); e1_i != e1_i_end; ++e1_i)
    {
        if(get(boost::edge_index, g1, /* index */ 1) {
            ed1 = *e1_i;
        }
    }

Try this if you get a segmentation fault:

template<typename subgraph_t>
void global_local_convertion_test(subgraph_t &g)
{
    subgraph_t &r_g = g.root();
    typename subgraph_t::edge_iterator er_i, er_i_end;
    for(tie(er_i, er_i_end) = edges(r_g); er_i != er_i_end; ++er_i)
    {
        if(get(boost::edge_index, r_g, *er_i) == 1)
        {
            g[g.global_to_local(*er_i)];
        }
    }
    typename subgraph_t::const_children_iterator gi, gi_end;
    boost::tie(gi, gi_end) = g.children();
    for(; gi != gi_end; ++gi)
    {
        global_local_convertion_test(*gi);
    }
}

[kenkangxgwe]

So are you saying that the edge between two vertices in a subgraph is not the same edge as between those two vertices in the full graph?

I tried to say the edge between two vertices in a subgraph is not the same edge as the edge between the same two vertices full graph after I convert it from global to local

[anadon]

That global to local might be to have alternate sets of edges in subgraphs. I'll check it out.

Please provide the output from valgrind running your test case in a gist.

[anadon]

    edge_descriptor global_to_local(edge_descriptor e_global) const
    { return is_root() ? e_global : (*m_local_edge.find(get(get(edge_index, root().m_graph), e_global))).second; }

It looks like subgraphs can have alternate sets of edges. I disagree with this design choice and it is opaque, but it is already set.

That moves focus to whether a subgraph's included edges at creation should be left as default, copied from the next level up, or it is obviously incorrect. I'll hit this later. I've been up since 3AM local time, past 4:30 now and I'm dead tired.

[kenkangxgwe]

There you go the log:
https://gist.github.com/kenkangxgwe/af4ccb3cec3ae446dc5d69df2184caba

BTW, it seems that we are in the same Time Zone. Good night, bro.