On 12/05/2011 05:29 PM, Petr Machata wrote:
Orion Poplawski<orion(a)cora.nwra.com> writes:
> I'm seeing the following boost related build error building paraview
> in rawhide. Do any boost gurus know what the issue might be the
> issue?
Hi there, please do not hesitate to file bugs for such regressions. I
only noticed this message today.
It wasn't obvious to me that it was a regression. New versions of libraries
things often change.
The problem here is in a tree_edge(e,g) event point of boost's
breadth_first_search. In reverse graph, this used to be called so that
"e" was directly the edge type of the edge in graph (vtkEdgeType in this
case). Now it's called with a value of wrapper type
boost::detail::reverse_graph_edge_descriptor<vtkEdgeType>. This wrapper
type has a public member underlying_desc which is the original edge.
Of course, with forward graph, this wrapping doesn't take place, so we
can't just plainly access e.underlying_desc.Id, we must wrap the access
itself. The hack that I'm attaching should take care of it. The build
is progressing now, I'm not sure if I'll hit something more.
I'll open an upstream bug for this.
Do you have a link for this?
Thanks!
--
Orion Poplawski
Technical Manager 303-415-9701 x222
NWRA/CoRA Division FAX: 303-415-9702
3380 Mitchell Lane orion(a)cora.nwra.com
Boulder, CO 80301
http://www.cora.nwra.com