From: Douglas Gregor (dgregor_at_[hidden])
Date: 2005-01-28 01:52:47
On Monday 24 January 2005 11:52 am, David M. Jones wrote:
> I am using the subgraph class template in my work and I have noticed a
> number of issues.
The subgraph adaptor was never really finished; it has quite a few lingering
bugs that need to be addressed, but thus far nobody has taken on that
> (1) I am unable to use boost::listS as the storage type for vertices or
> edges. I see that this is because the association between the local and
> global vertices is stored (in part) in a std::vector. Is there a plan to
> support subgraphs of non-array-indexed graphs?
Oh, yuck. subgraph should really take a VertexIndex parameter to map from
global vertices to indices.
> (2) The remove_edge_if() and clear_vertex() functions have comments stating
> that they are wrong. Are the functions wrong or are the comments wrong?
The functions are definitely wrong.
> the functions are wrong, is there a plan to fix them?
Not currently. They are, unfortunately, hard to implement well.
> (3) The remove_vertex() function states that it is "under construction".
> Who is doing the construction? Is there a time-frame for this?
Nobody is doing the construction now :( remove_vertex is one of the nastier
functions to implement, because it requires traversing both up and down the
subgraph hierarchy removing all of the local vertices, which can invalidate
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk