Changed the basic reflection mechanism: Instead of having a function templates over all the four containers now we template over Trimesh and we rely on a second function templated on face/vert that wants a vector<face> ; this second function only is eventually overloaded by another function that needs a vector_ocf of faces.
That is Before we had:
- in complex.h
template < class CType0, class CType1, class CType2 , class CType3>
bool HasPerFaceVFAdjacency (const TriMesh < CType0, CType1, CType2, CType3> & /*m*/) {return TriMesh < CType0 , CType1, CType2, CType3>::FaceContainer::value_type::HasVFAdjacency();}
- in the component_ocf.h
template < class VertContainerType, class FaceType, class Container1, class Container2 >
bool HasPerFaceVFAdjacency (const TriMesh < VertContainerType , face::vector_ocf< FaceType >, Container1, Container2 > & m)
{
if(FaceType::HasVFAdjacencyOcf()) return m.face.IsVFAdjacencyEnabled();
else return FaceType::FaceType::HasVFAdjacency();
}
While now we have:
- in complex.h
template < class FaceType > bool FaceVectorHasPerFaceVFAdjacency (const std::vector<FaceType > &) { return FaceType::HasVFAdjacency(); }
template < class TriMeshType> bool HasPerFaceVFAdjacency (const TriMeshType &m) { return tri::FaceVectorHasPerFaceVFAdjacency (m.vert); }
- and in component_ocf.h
template < class FaceType >
bool FaceVectorHasPerFaceVFAdjacency(const face::vector_ocf<FaceType> &fv)
{
if(FaceType::HasVFAdjacencyOcf()) return fv.IsVFAdjacencyEnabled();
else return FaceType::HasVFAdjacency();
}
No modification should be necessary for the existing code.
most relevant changes:
creation of folder:
vcg/connectors
vcg/connectors/hedge.h
vcg/connectors/hedge_component.h
addition to the container of half edges to the trimesh:
HEdgeContainer hedge; // container
int hn; // number of half edges
addition of
vcg/trimesh/update/halfedge_indexed.h
which contains:
- the functions to compute the half edge representation from the indexed and vivecersa
- the functions to add or remove an half edge
Note for the developers: the change to make to existing projects is very little
but strictly necessary to compile. This change IS NOT backward compliant.
==== OLD ==== way to define a TriMesh:
// forward declarations
class MyVertex;
class MyEdge;
class MyFace;
class MyVertex: public VertexSimp2 < MyVertex, MyEdge, MyFace, vertex::Coord3f,...other components>{};
class MyFace: public FaceSimp2 < MyVertex, MyEdge, MyFace, face::VertexRef,...other components>{};
class MyMesh: public TriMesh<vector<MyVertex>,vector<MyFace> >{};
==== NEW ==== way to define a TriMesh:
// forward declarations
class MyVertex;
class MyEdge;
class MyFace;
// declaration of which types is used as VertexType, which type is used as FaceType and so on...
class MyUsedTypes: public vcg::UsedType < vcg::Use<MyVertex>::AsVertexType,
vcg::Use<MyFace>::AsFaceType>{};
class MyVertex: public Vertex < MyUsedTypes, vertex::Coord3f,...other components>{};
class MyFace: public Face < MyUsedTypes, face::VertexRef,...other components>{};
class MyMesh: public TriMesh<vector<MyVertex>,vector<MyFace> >{};
===== classes introduced
[vcg::UsedType] : it is a class containing all the types that must be passed to the definition of Vertex, Face, Edge... This
class replaces the list of typenames to pass as first templates and the need to specify the maximal simplicial. So
<MyVertex, MyEdge, MyFace becomes <MyUsedTypes<
and
VertexSimp2 becomes Vertex
[vcg::Use] : an auxiliary class to give a simple way to specify the role of a type
Note 2: the order of templates parameters to vcg::UsedTypes is unimportant, e.g:
class MyUsedTypes: public vcg::UsedType <vcg::Use<MyVertex>::AsVertexType,
vcg::Use<MyEdge>::AsEdgeType,
vcg::Use<MyFace>::AsFaceType>{};
is the same as:
class MyUsedTypes: public vcg::UsedType <vcg::Use<MyFace>::AsFaceType,
vcg::Use<MyEdge>::AsEdgeType,
vcg::Use<MyVertex>::AsVertexType>{};
Note 3: you only need to specify the type you use. If you do not have edges you do not need
to include vcg::Use<MyEdge>::AsEdgeType in the template list of UsedTypes.
==== the Part II will be a tiny change to the class TriMesh it self.
Paolo Cignoni
ganovelli
Marco Di Benedetto