Modeling Algorithms - Compute global properties of tessellated shape

This patch is to replace the partial patch implemented in scope of #29731.

Branch [archived branch] has been created by Participant.

[revision removed]


Detailed log of new commits:

Author: ifv
Date: Thu May 17 15:38:17 2018 +0300

    0029734: Modeling Algorithms - Compute global properties of tessellated shape
    
    New algorithms using mesh data for calculation are added: BRepGProp/BRepGProp_MeshCinert, BRepGProp/BRepGProp_MeshSinert, BRepGProp/BRepGProp_MeshVinert.
    
    In API special flag, which defines preferable source for calculations is added
    Test case is added
    Documentation is updated

Branch CR29734 is ready for review

src/BRepGProp/BRepGProp_MeshCinert.hxx
- 1-4: incorrect information
- 30-35: not relevant description

src/BRepGProp/BRepGProp_MeshCinert.cxx
- 1-2: incorrect information
- Please use English for comments.
- 80-81: simplify as Lower is always == 0.
- 82: protection against division by zero is needed.

src/BRepGProp/BRepGProp_MeshSinert.hxx
- 1-3: incorrect information
- 34: Comment about closed region of space is not relevant to surface properties.
- 43: missing end of sentense point.
- Wrap too long lines.

src/BRepGProp/BRepGProp_MeshSinert.cxx
- 1-3: incorrect information
- 50-55: too heavy computation of triangle area and normal. It is better to avoid creation of an additional array of points.
- 58: The caller has no means to know if the computation is failed.
- 155, 159: skip triangle if CalculateProps is failure.

src/BRepGProp/BRepGProp_MeshVinert.hxx
- 1-3: incorrect information
- 44: missing end of sentense point.
- Wrap too long lines.

src/BRepGProp/BRepGProp_MeshVinert.cxx
- 1-3: incorrect information

The files BRepGProp_MeshSinert.cxx and BRepGProp_MeshVinert.cxx are almost the same except a small portion of the method CalculateProps. Please think about avoiding of code duplication.

src/BRepGProp/BRepGProp.hxx
- 220: misprint "defained"
- Wrap too long lines where they are changed.

src/BRepGProp/BRepGProp.cxx
- 60: use reference.
- 160-188: too heavy calculation of center and normal. Precision depends on the distance of polygon from the origin. I propose to calculate area and normals of triangles with origin in the first point.
- 212: Why management of Mesh inertia is not made like curve inertia, declaring the instance of BRepGProp_MeshCinert where it is needed?
- 293-309: it is better to move the logic into BRepGProp_MeshSinert, passing there triangulation and location.
- 423-439: the same as above, but for BRepGProp_MeshVinert.

src/BRepTest/BRepTest_GPropCommands.cxx
- 45: misprint "geomery"

dox/dev_guides/upgrade/upgrade.md
- Upgrade guide is not needed to be modified, as the changes do not impact the port to new version.

dox/user_guides/draw_test_harness/draw_test_harness.md
- 7375: "forces computation of only closed shells"
- 7377: "will be taken"
- 7378: "are output"

tests/bugs/modalg_7/bug29734
- The same checker code is copied three times. Please make a tcl proc and call it three times.

Please take from the branch CR29731_1 the changes in files

src/BRepTools/BRepTools.cxx
src/BRepTools/BRepTools.hxx
src/DBRep/DBRep.cxx
tests/bugs/modalg_7/bug29731

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch CR29734 is ready for review

The classes BRepGProp_MeshSinert, BRepGProp_MeshVinert are not used any more. Can they be removed then?

src/BRepGProp/BRepGProp_MeshCinert.hxx
src/BRepGProp/BRepGProp_MeshCinert.cxx
src/BRepGProp/BRepGProp_MeshSinert.hxx
src/BRepGProp/BRepGProp_MeshSinert.cxx
src/BRepGProp/BRepGProp_MeshVinert.hxx
src/BRepGProp/BRepGProp_MeshVinert.cxx
src/BRepGProp/BRepGProp_MeshProps.hxx
src/BRepGProp/BRepGProp_MeshProps.cxx
- Please add copyright line in the file beginning:
// Copyright (c) 2018 OPEN CASCADE SAS

src/BRepGProp/BRepGProp_MeshSinert.hxx
- 42: missing end of sentense point.
- 46: Wrap too long line.

src/BRepGProp/BRepGProp_MeshVinert.hxx
- 47: Wrap too long line.

src/BRepGProp/BRepGProp_MeshProps.hxx
- 30: It looks the same as of BRepGProp_MeshSinert. It is needed to tell that this class is capable of computing surface or volume properties.

src/BRepGProp/BRepGProp.hxx
- 96, 135, 148, 186, 200, 215, 217: please wrap long lines.

src/DBRep/DBRep.cxx
- Please update the help of the command tclean with new option "-geom".

tests/bugs/modalg_7/bug29734
- 9-10, 21-22: upvar is not needed for strings.

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been updated forcibly by Author.

[revision removed]

Some code formatting and comments updating have been done.

Reviewed.

+void BRepGProp_MeshProps::Perform(const Handle(Poly_Triangulation)& theMesh,
+                                  const TopLoc_Location& theLoc,
+                                  const TopAbs_Orientation theOri)
+{
+  if (theLoc.IsIdentity())
+  {
+    Perform(theMesh->Nodes(), theMesh->Triangles(), theOri);
+  }
+  else
+  {
+    Standard_Integer aNbNodes = theMesh->NbNodes();
+    TColgp_Array1OfPnt aTrNodes(1, aNbNodes);
+    const gp_Trsf& aTr = theLoc.Transformation();
+    const TColgp_Array1OfPnt& aNodes = theMesh->Nodes();
+    Standard_Integer i;
+    for (i = 1; i <= aNbNodes; ++i)
+    {
+      aTrNodes.SetValue(i, aNodes.Value(i).Transformed(aTr));
+    }
+    Perform(aTrNodes, theMesh->Triangles(), theOri);
+  }

Is it possible avoiding copying of entire triangulation when it is passed with Location?

+private: //! @name private fields
+
+  void CalculateProps(const gp_Pnt& p1, const gp_Pnt& p2, const gp_Pnt& p3,
+                      const gp_Pnt& Apex,
+                      Standard_Real GProps[10],
+                      const Standard_Integer NbGaussPoints,
+                      const Standard_Real* GaussPnts);
+
+  BRepGProp_MeshObjType myType; //!< Type of geometric object

Could you please also make this method protected (or public static)?
It can be reused for implementing the same tool processing triangulation not in Poly_Triangulation form.

Dear Igor, please consider KGV's remarks.

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

CR29734 is updated according to KGV remarks

Reviewed.

Combination -
OCCT branch : [archived branch] SHA - [revision removed]
Products branch : [archived branch] SHA - [revision removed]
was compiled on Linux, MacOS and Windows platforms and tested in optimize mode.

Number of compiler warnings:
No new/fixed warnings

Regressions/Differences/Improvements:
No regressions/differences

CPU differences:
Debian70-64:
OCCT
Total CPU difference: 17086.430000000008 / 17011.039999999866 [+0.44%]
Products
Total CPU difference: 7346.630000000022 / 7436.080000000027 [-1.20%]
Windows-64-VC10:
OCCT
Total CPU difference: 16889.198663398634 / 16821.494229398526 [+0.40%]
Products
Total CPU difference: 8048.8871950998755 / 8185.31006959986 [-1.67%]


Image differences :
No differences that require special attention

Memory differences :
No differences that require special attention

roughBaryCenter() in BRepGProp.cxx uses TopoDS_Vertex in the Shape,
which will be unavailable in case of triangulation-only Faces.
Can this lead to incorrect result?

+  dim = aGProps[0];
+  g.SetX(aGProps[1] / dim);
+  g.SetY(aGProps[2] / dim);
+  g.SetZ(aGProps[3] / dim);

I think that dim should be checked for 0.0 to avoid possible division by zero FPE in case of degenerated triangulation.

+  static Standard_Real GPtsWg[] =

Seems to be static const.

+private: //! @name private fields
+  BRepGProp_MeshObjType myType; //!< Type of geometric object

This field is still private.

Dear Igor, please consider new remarks.

In roughBaryCenter(), you can consider triangulation nodes if i==0 after iteration by vertices.

To get read access to the field "BRepGProp_MeshObjType myType", please add a public method.

> In roughBaryCenter(), you can consider triangulation nodes if i==0 after iteration by vertices.
Andrey suggests that for algorithm it is sufficient providing a point near enough to the shape itself.

E.g., if shape contains at list one TopoDS_Vertex - it would be sufficient taking it's coordinates (without averaging all Vertices, which could be removed)
and if not - take a first Node from Triangulation from any Face.

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

CR29734 is updated

Reviewed.

Branch [archived branch] has been created by Commenter 1.

[revision removed]


Detailed log of new commits:

Author: ifv
Date: Thu May 17 15:38:17 2018 +0300

    0029734: Modeling Algorithms - Compute global properties of tessellated shape
    
    New algorithms using mesh data for calculation have been added: BRepGProp_MeshCinert, BRepGProp_MeshProps.
    In API a special flag, which defines preferable source for calculations, has been added.

Last fix caused regression on windows and Linux on the same case
emesh bugs bug29774


http://occt-tests/CR29734-master-KGV-Products/Debian70-64/summary.html
http://occt-tests/CR29734-master-KGV-Products/Windows-64-VC10/summary.html

Fix has been rebased to current master.

http://jenkins-test-11.nnov.opencascade.com/view/CR29734_1-master-KGV/

Combination -
OCCT branch : [archived branch] SHA - [revision removed]
Products branch : [archived branch] SHA - [revision removed]
was compiled on Linux, MacOS and Windows platforms and tested in optimize mode.

Number of compiler warnings:
No new/fixed warnings

Regressions/Differences/Improvements:
No regressions/differences

CPU differences:
Debian70-64:
OCCT
Total CPU difference: 17038.400000000005 / 17008.500000000095 [+0.18%]
Products
Total CPU difference: 7457.790000000032 / 7460.130000000034 [-0.03%]
Windows-64-VC10:
OCCT
Total CPU difference: 16875.704576898515 / 16835.17551709858 [+0.24%]
Products
Total CPU difference: 8272.811030499875 / 8259.114142699882 [+0.17%]


Image differences :
No differences that require special attention

Memory differences :
No differences that require special attention

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been deleted by Commenter 1.

[revision removed]

Branch [archived branch] has been deleted by Commenter 1.

[revision removed]