Adaptor2d_Curve2d, Adaptor3d_Curve and Adaptor3d_Surface now inherit Standard_Transient.
Interfaces Adaptor2d_HCurve2d, Adaptor3d_HCurve, Adaptor3d_HSurface and their subclasses
are now aliases to Adaptor2d_Curve2d, Adaptor3d_Curve and Adaptor3d_Surface.
Removed numerous unsafe reinterpret casts.
Generic classes Adaptor3d_GenHCurve, Adaptor3d_GenHSurface, Adaptor2d_GenHCurve2d have been removed.
Several redundant .lxx files have been merged into .hxx.
Removed obsolete adaptor classes with H suffix.
Import of arbitrary polygon is implemented in VrmlData_IndexedFaceSet.
Class IFCCAFControl_Triangulator has been moved to BRepMesh_Triangulator.
New Draw test /bugs/stlvrml/bug30421 is added.
- Standard_EXPORT which were specified for inline methods were deleted.
- ALL occurrences of DEFINE_STANDARD_RTTI_INLINE were replaced by DEFINE_STANDARD_RTTIEXT in header files and IMPLEMENT_STANDARD_RTTIEXT in source files
- ALL occurrences of "inline" keyword were deleted where it didn't not cause a linkage errors
- Added source files for classes that were without them for IMPLEMENT_STANDARD_RTTIEXT
Added custom meshing core algorithm to generate base mesh using Delabella library,
which can be enabled via IMeshTools_Parameters::MeshAlgo option or CSF_MeshAlgo environment variable.
Do not fill cirles filter upon explicit initialization.
Call base postProcessMesh functionality after initialization of circles in BRepMesh_CustomDelaunayBaseMeshAlgo.
Added Vsprintf() wrapper for vsprintf() preserving C locale.
BRepMesh: Add new mesh parameter *AllowQualityDecrease* which affects the criteria used for checking of the consistency of the existing mesh to new meshing parameters.
So if set to true it will force the meshing of the shape if current deflection strongly vary from the new one, no matter in which side.
BRepTools::Clean: Keep triangulation on non-geometric shapes (faces with no surface or edges with no curves).
Standard_EXPORT is added in declaration of virtual methods where it was missing, to allow creation of sub-classes in other DLLs.
In some classes in OpenGl package Standard_EXPORT is added for all methods to make these classes usable in other DLLs.
BRepMesh_ModelHealer - recognize a small face with 1 wire and 2 small edges as a face for refinement.
BRepMesh_DefaultRangeSplinner - reduce tolerance on tiny faces to improve the meshing results.
Added new test cases: bugs mesh bug25044_*.
Existing test cases updated to reflect improvements.
Fixed license statements within files BRepGProp_MeshProps.cxx, BRepGProp_MeshProps.hxx,
OSD_ThreadPool.cxx and OSD_ThreadPool.hxx.
Removed non-ASCII char from BRepMesh_ModelHealer.hxx.
"endl" manipulator for Message_Messenger is renamed to "Message_EndLine".
The following entities from std namespace are now used
with std:: explicitly specified (from Standard_Stream.hxx):
std::istream,std::ostream,std::ofstream,std::ifstream,std::fstream,
std::filebuf,std::streambuf,std::streampos,std::ios,std::cout,std::cerr,
std::cin,std::endl,std::ends,std::flush,std::setw,std::setprecision,
std::hex,std::dec.
New classes BRepMesh_ConstrainedBaseMeshAlgo, BRepMesh_CustomBaseMeshAlgo and BRepMesh_CustomDelaunayBaseMeshAlgo are added.
These classes allow to add any custom triangulation algorithm to BRepMesh and perform post-processing and optimization of base mesh generated by those algorithms.
BRepMesh_Delaun: added possibility to process constraints when base mesh is generated by different algorithm.
BRepMesh_DelaunayNodeInsertionMeshAlgo: added PreProcessSurfaceNodes flag controlling addition of surface nodes (either before creation of base mesh or after) to gain maximum performance from triangulation algorithms.
Minor changes:
Use simple algorithm for cylinders when internal vertices mode is switched off to speed up computations.
BRepMesh_IncrementalMesh: added Perform method allowing to execute algorithm using manually created Context.
Added new parameter KeepSmallEdges to IMeshTools_Parameters allowing to compute min size for each edge locally, depending on the length of particular edge.
Parameter -smalledges now available for incmesh command.
IMeshTools_ModelAlgo and IMeshTools_ModelBuilder have been changed to provide exception protected interfaces for performing the operations.
Protect single Edge/Face discretization methods from raising exceptions to skip broken Edges/Faces and allow mesh construction on the whole model.
0030551: Foundation Classes - Integer overflow in NCollection_CellFilter HashCode
Signed integers are not used in hash code functions now to prevent undefined behavior on left shift operations with signed integers.
A possibility of negative values of hash codes is eliminated.
INT_MAX → IntegerLast() in hash code functions.
All found hash code functions behaves uniformly now: they return a value in the range [1, theUpperBound]. Relevant comments are added to such functions.
The previous code has a condition to avoid processing the same faces if the face has a location.
The similar condition should be applied during the edges processing.
If not doing this, in the previous implementation, IMeshData_Edge instances are created for all edges(even located), but edges of faces located are not filled with curves.
As a result we had wrong local selection of edges.
Limit addition of edges to data model by ones with unique TShape and location using edges map already available in BRepMesh_ShapeVisitor.
Split seam edge of a cone according to the specified parameters.
Seam edge functor is called before check of triangulation consistency in order to keep face status consistent.
The cause is that split seam edge can set Outdated flag when Reused is set by another functor without reset which can lead to refusement of such face by triangulation procedure.
Fix compilation error on VS2008
Before the fix, angular deflection was scaled to some coefficient in the method BRepMesh_Deflection::ComputeDeflection(...). Now, angular deflection is constant.
1. Check whether the mesh satisfies the required angular deflection has been amended. Namely normals (to the surface) in the ends of any not "frontier" link are made collinear (with the given angular tolerance).
2. New parameters AngleInterior and DeflectionInterior have been added in IMeshTools_Parameters structure.
3. In case of thin long faces with internal edges, add points of internal edges to control parameters using grabParamsOfInternalEdges() in order to avoid aberrations on its ends. Disable addition of parameters from boundary edges in case of BSpline surface. Deviation can be controlled through the deflection parameter.
4. Grab parameters from edges in case if there is just a single interval on BSpline surface along U and V direction.
Removed tight connections between data structures, auxiliary tools and algorithms in order to create extensible solution, easy for maintenance and improvements;
Code is separated on several functional units responsible for specific operation for the sake of simplification of debugging and readability;
Introduced new data structures enabling possibility to manipulate discrete model of particular entity (edge, wire, face) in order to perform computations locally instead of processing an entire model.
The workflow of updated component can be divided on six parts:
* Creation of model data structure: source TopoDS_Shape passed to algorithm is analyzed and exploded on faces and edges. For each topological entity corresponding reflection is created in data model. Note that underlying algorithms use data model as input and access it via common interface which allows user to create custom data model with necessary dependencies between particular entities;
* Discretize edges 3D & 2D curves: 3D curve as well as associated set of 2D curves of each model edge is discretized in order to create coherent skeleton used as a base in faces meshing process. In case if some edge of source shape already contains polygonal data which suites specified parameters, it is extracted from shape and stored to the model as is. Each edge is processed separately, adjacency is not taken into account;
* Heal discrete model: source TopoDS_Shape can contain problems, such as open-wire or self-intersections, introduced during design, exchange or modification of model. In addition, some problems like self-intersections can be introduced by roughly discretized edges. This stage is responsible for analysis of discrete model in order to detect and repair faced problems or refuse model’s part for further processing in case if problem cannot be solved;
* Preprocess discrete model: defines actions specific for implemented approach to be performed before meshing of faces. By default, iterates over model faces and checks consistency of existing triangulations. Cleans topological faces and its adjacent edges from polygonal data in case of inconsistency or marks face of discrete model as not required for computation;
* Discretize faces: represents core part performing mesh generation for particular face based on 2D discrete data related to processing face. Caches polygonal data associated with face’s edges in data model for further processing and stores generated mesh to TopoDS_Face;
* Postprocess discrete model: defines actions specific for implemented approach to be performed after meshing of faces. By default, stores polygonal data obtained on previous stage to TopoDS_Edge objects of source model.
Component is now spread over IMeshData, IMeshTools, BRepMeshData and BRepMesh units.
<!break>
1. Extend "tricheck" DRAW-command in order to find degenerated triangles.
2. Class BRepMesh_FastDiscret::Parameters has been declared as deprecated.
3. NURBS range splitter: do not split intervals without necessity. Intervals are split only in case if it is impossible to compute normals directly on intervals.
4. Default value of IMeshTools_Parameters::MinSize has been changed. New value is equal to 0.1*Deflection.
5. Correction of test scripts:
1) perf mesh bug27119: requested deflection is increased from 1e-6 to 1e-5 to keep reasonable performance (but still reproducing original issue)
2) bugs mesh bug26692_1, 2: make snapshot of triangulation instead of wireframe (irrelevant)
Correction in upgrade guide.
New class OSD_ThreadPool has been introduced to define a Thread Pool for multi-threading algorithm.
Thread Pool assigns a serial number for each thread allowing Multi-Threading algorithm to allocate thread-local storage variables as an array whose size is the same as the number of threads.
OSD_ThreadPool also redirects exceptions to a thread calling parallel execution and consistently initializes FPE exception handling.
New class Standard_Condition provides a platform-independent tool similar to Event in WinAPI.
A new auxiliary function Standard_Atomic_CompareAndSwap() has been introduced
for performing atomic compare and swap of integer number.
Standard_Atomic_Increment/Standard_Atomic_Decrement fallback implementation
using ASM code for x86 processors for GCC has been dropped;
instead, it is expected that GCC should be properly configured targeting modern x86 architectures.
OSD_Signal now declares fFltExceptions as thread_local variable accessible through OSD::ToCatchFloatingSignals() property.
Standard_THREADLOCAL macro (wrapping thread_local attribute) has been moved to public header Standard_Macro.hxx.
OSD_Parallel::ForEach() has been extended with new optional parameter theNbItems and uses OSD_ThreadPool::DefaultPool().
All occurrences of Standard_EXPORT attached to inline methods in OCCT code are eliminated.
Some unused classes and C++ files producing no code are deleted.
Function Strtod() is reimplemented using open source (MIT-style license) code by David M. Gay instead of strtod() provided by standard run-time library. This improves its performance by 3-10 times.
Functions Atof(), Strtod(), Printf(), Sprintf(), Fprintf() are declared as extern "C" to be usable from C programs.
Strtod() is used in Interface_FileReaderData::Fastof() and in RWStl_Reader to accelerate their work.
DRAW command QATestAtof and test perf fclasses strtod are added to check correctness and performance of Strtod().
Test perf draw restore is added to monitor performance of reading BREP files.
Minor off-topic corrections:
- method Standard_GUID::Assign (const Standard_UUID&) is implemented (was empty);
- Precision.hxx is included in BRepMesh_Vertex.hxx that uses it.
Macro NO_CXX_EXCEPTION was removed from code.
Method Raise() was replaced by explicit throw statement.
Method Standard_Failure::Caught() was replaced by normal C++mechanism of exception transfer.
Method Standard_Failure::Caught() is deprecated now.
Eliminated empty constructors.
Updated samples.
Eliminate empty method ChangeValue from NCollection_Map class.
Removed not operable methods from NCollection classes.
Added DRAW command dsetsignal, resetting OSD signal handler with either armed or disabled FPE handler, according to an option.
If called without arguments, it sets FPE handler only if environment variable OSD_FPE is defined (with value different from 0).
On start, DRAW calls dsetsignal to set FPE signal if CSF_FPE is defined.
Test bugs fclasses bug6143 uses dsetsignal to set FPE handler unconditionally before the test command, and resets it to default at the end.
A number of changes in the code have been done in order to fix floating point exceptions that became generated after enabling signals:
- Global functions Sinh() and Cosh() defined in Standard_Real.hxx are improved to raise Standard_NumericError exception if argument is too big (greater than 710.47586), instead of relying on system treatment of floating point overflow. These functions are used instead of sinh and cosh in ElCLib.cxx.
- Maximal value of parameter on hyperbola is restricted by 23 (corresponding to ~1e10 in 3d) in order to avoid FP overflow in Extrema_GenExtCS.cxx, ShapeFix_EdgeProjAux.cxx.
- Interface of the root curve adaptor class Adaptor3d_Curve has been updated to add new virtual methods BasisCurve and OffsetValue. They complement the adaptor for the case of offset curves. These methods are used in Extrema_GenExtCS.cxx to restrict domain search in the case of offset of hyperbola, in order to get rid of floating point overflow. All classes inheriting Adaptor3d_Curve have been changed to implement the new virtual methods.
- Protection against division by zero has been implemented in ApproxInt_KnotTools.cxx, BRepClass3d_SClassifier.cxx, BRepGProp_Face.cxx, BRepMesh_FastDiscretFace.cxx, Geom2dGcc_Circ2d2TanOnIter.cxx, Geom2dInt_Geom2dCurveTool.cxx, IntPolyh_MaillageAffinage.cxx.
- Protection against calling of math functions of infinite arguments has been added in BRepCheck_Edge.cxx, BRepLib.cxx, CSLib_NormalPolyDef.cxx, Extrema_FuncExtPC.gxx, Extrema_GExtPC.gxx, Extrema_GLocateExtPC.gxx, Intf_InterferencePolygonPolyhedron.gxx, ShapeAnalysis_Surface.cxx, ShapeAnalysis_TransferParametersProj.cxx, ShapeAnalysis_Wire.cxx, math_FunctionSetRoot.cxx.
- Proper initialization of local variables is done in BOPAlgo_PaveFiller_6.cxx, XSDRAWSTLVRML.cxx.
- Inconsistent usage of Standard_Boolean* to access integer data in HLR (caused by #27772) is corrected
Some test cases have been updated to actual state.
Code has been updated to remove no-op casts and implicit casts to Standard_Boolean.
Places of inproper use of Standard_Boolean instead of Standard_Integer
have been corrected:
- Bnd_Box, Bnd_Box2d
Bit flags are now defined as private enum
- HLRAlgo_BiPoint, HLRAlgo_EdgesBlock, HLRBRep_EdgeData, HLRBRep_FaceData
Bit flags are now defined as enum
- HLRAlgo_EdgeStatus, HLRBRep_BiPnt2D, HLRBRep_BiPoint
Bit flags are now defined as bool fields
- HLRAlgo_PolyData
Bit flags are now defined as Standard_Integer
- OSD_DirectoryIterator, OSD_FileIterator
Boolean flag is now defined as Standard_Boolean
- ShapeAnalysis_Surface::SurfaceNewton()
now returns Standard_Integer (values 0, 1 or 3)
- ChFi2d_FilletAlgo
now uses TColStd_SequenceOfBoolean instead of TColStd_SequenceOfInteger
for storing boolean flags
Method IFSelect_Dispatch::PacketsCount() has been dropped from interface.
ShapeFix_Solid::Status() has been fixed to decode requested status
instead of returning integer value.
TopOpeBRepBuild_Builder1 now defines map storing Standard_Boolean values
instead of Standard_Integer.
Persistence for Standard_Boolean type has been corrected
to keep backward compatibility:
- BinMDataStd, BinTools, FSD_BinaryFile
Broken Draw Harness commands vdisplaymode and verasemode have been removed.
BRepMesh_FastDiscretFace::initDataStructure() - workaround old gcc limitations
BRepMesh_IncrementalMesh::clear() - avoid ambiguity
1) BRepMesh_FastDiscretFace.cxx:
- exclude planes from procedure of inserting internal points.
- localize declaration of the container aNewVertices in each method where it is needed.
- correct the logic of the method insertInternalVerticesOther, so that to separate the processes of removing extra points and addition of new points in different cycles, thus making the code more clear and in addition stable.
- insert useful output of intermediate mesh to a file in control() method for debug purposes (with definition DEBUG_MESH).
2) Add global functions MeshTest_DrawTriangles and MeshTest_DrawLinks to draw mesh data in debug session.
3) BRepMesh_FastDiscret:
- in the method Add calculations of deflections have been simplified for non-relative mode.
- replace the attribute MinDist with Deflection in EdgeAttributes structure. Correct its computation so that later to store this value as deflection of the polygon.
4) Make protection against exception in the method BRepMesh_Delaun::addTriangle() when an added triangle creates a third connection of a mesh edge.
5) BRepMesh_EdgeTessellator.cxx, BRepMesh_EdgeTessellationExtractor.cxx: use Geom2dAdaptor_Curve in order to use b-spline cache while computing value on a curve.
6) In BndLib_Box2dCurve::PerformBSpline, avoid creating new b-spline in case of requested parameter range differ from natural bounds insignificantly.
7) In GeomAdaptor classes, postpone building of cache till the time of its actual usage. So, creation of an adapter to compute intervals of continuity does not lead to creation of internal cache.
8) In the methods BRepAdaptor_Curve::Bezier and BSpline do not call Transformed() if transformation is identity.
9) In the classes Geom_BSplineCurve, Geom_BSplineSurface, Geom_BezierCurve, Geom_BezierSurface, Geom2d_BSplineCurve, Geom2d_BezierCurve change the method Pole() to return the point by const reference.
10) In CPnts_AbscissaPoint.cxx, compute derivative by D1 instead of DN to make use of b-spline cache.
11) Change test cases to actual state:
- Number of triangles/nodes can grow due to more accurate work with deflection of edges. Now the edge is tessellated using its own tolerance instead of maximal tolerance of all shapes in the face.
- Accept new numbers of mesh errors (free links, free nodes) for really bad shapes.
- Correct the test "bugs/mesh/bug25612" to produce stable result.
- Disable redundant checks in test cases bug25378* (lower limit for computation time).
- Speed up iso-lines computation for offset of bspline surfaces. For that use adaptor instead of original surface in evaluator of approximation.
- Add output of polylines for debug of insertInternalVerticesOther().
Reference data in test case bugs\moddata_2\bug453_3 have been changed to be close to expected theoretical values. This makes the test give stable result on different platforms.
1) Reduce the number of calls to malloc by grouping requests to larger blocks. To achieve this goal, the following ways are used:
- Containers of types sequence, list and map are initialized with an instance of NCollection_IncAllocator, at this taking care of the time of life of allocated objects, so that not to occupy huge amount of memory.
- Allocation of several arrays having the same and short life time is changed so that to allocate a buffer array of necessary size and to place arrays in this buffer.
2) In BRepMesh_FastDiscretFace, optimize the function filterParameters so that to avoid excess memory allocations.
3) In NCollection_CellFilter, change declaration of the method Reset to accept array by reference rather than by value.
4) Add Allocator() method in map, sequence and vector collection classes by analogy with list collection.
5) Correct the size of block for IncAllocator for x64 platform. In order free-ed block to be returned to the system its size should be at least 1024K on x64 and 512K on x86. This allows to retain free virtual space almost to the state before algorithm run.
6) Decrease amount of memory zeroed by calloc. For that, reduce theIncrement parameter of the embedded vectors of the classes NCollection_UBTreeFiller and BRepMesh_VertexInspector to default value 256.
7) Avoid computing bounding box when not necessary (if no relative deflection is used)
8) Cycles by wires of face using TopExp_Explorer are converted to use TopoDS_Iterator instead.
9) BRepMesh_FastDiscret::Add optimized to avoid storing sequences of faces and edges
10) The tests "mesh standard_* W7" are corrected to accept the new behavior. Earlier the following error took place:
Not connected mesh inside face 9
{12 13}
Now this error was replaced with another one:
free nodes (in pairs: face / node):
{9 12}
Actually it is not a regression, rather improvement, if we look at the snapshot.
11) Change other test cases to their actual state.
Do not insert internal nodes for Cylinder in case if it is less than DefFace value or in case of long cylinder with small radius due to protection against overflow during casting to integer.
Small correction of shape name in test case for issue CR27442
