OpenCascade is a foundation library for geometric modeling. It provides root classes like handles and memory management. It also includes collections, vectors/matrices, primitive geometric types, and math algorithms. Specific features include auto reference counting, an optimized memory manager, runtime type information, and exceptions handling. Geometric entities can be manipulated by value. The library supports units conversion, internationalization, and progress reporting.

1. OpenCascade: (7.1.0)
Foundation Classes
王聖川
2017/8/22

2. Introduction
• Root classes.
• Collections.
• Vectors and matrices.
• Primitive Geometric Types.
• Math algorithms.
• Application services.
• 單位轉換 (UnitsAPI package); Interpreter of expressions
(ExprIntrp package); 多國語言支援(Resource and Message
packages); 運算處理進度回報機制.

3. Root Classes
• Fundmentals: Boolean, character, integer, real…
• Safe handles. Auto reference counting. 物件參考使用
狀態的自動化管理.
• Configurable optimized memory manager. 記憶體最
佳化管理方案.
• Extended runtime type information (RTTI)
mechanism. 執行期間類別資訊判斷機制.
• Exceptions management.
• Strings.
• Mainly implemented in Standard and MMgr
packages.

4. Primitive Types
• Manipulated by value.
C++ Types OCCT Types
int Standard_Integer
double Standard_Real
float Standard_ShortReal
unsigned int Standard_Boolean
char Standard_Character
short Standard_ExtCharacter
char* Standard_CString
void* Standard_Address
short* Standard_ExtString

5. Value or Handle
• Avoid cyclic handles. 避免多個handle形成互相參考的情況
發生, 會導致彼此皆無法解構或被釋放.

6. RTTI
• Standard_Type singleton物件負責類別定義的描述資訊.
• 用法:
• 執行期間型別判斷:

7. Memory Management
• OCCT自己實現的記憶體管理機制, 用在記憶體存取的效
能優化目的.
• 物件被配置和建構, 被釋放以及解構的時機點, 皆透過memory
manager, 避免反覆做真實的記憶體配置.
• Configure: 設定環境變數(?黑人問號).
• MMGT_OPT, MMGT_CELLSIZE, MMGT_THRESHOLD, …
• 可透過改source code. Standard.cxx 中 #define
OCCT_MMGT_OPT_DEFAULT 0  1
• 缺點: 有可能被誤認出現memory leaks, 建議在密集配置物件後,
呼叫 Standard::Purge() .

8. Exception
• OCCT實作自己的例外處理機制, 用法語C++ exception
處理機制類似.
• DomainError::Raise(“error message”); 丟出例外.
• 在try – catch區塊中處理例外.
• Root exception class: Standard_Failure.

9. Collections
• Collection component處理可變數量之集合類型資料.
• Generic (C++ template-like) class implementation.
• Arrays.
• TCollection_Array1(一維陣列), TCollection_Array2(二維陣列), …
• Sequences. 整體集合中只儲存有意義資料, 節省記憶體空間.
• Maps. 透過key-value配對來快速存取資料元素.
• *_DataMap, *_IndexedMap, *_DoubleMap, …
• Lists.
• Iterators. 逐步探索collections的實作類別.
• *_*MapIterator, *_*SetIterator, *_*ListIterator, …

10. NCollections
• NCollections component implements TCollections with
additional features.
• Array1, Array2, List, Sequence, Map, IndexedMap, DataMap, …
• Other types: Vector, UBTree(unbalanced binary tree),
SparseArray, CellFilter(for sorting geometries).
• Features:
• Iterator: Init(), More(), Next(), Value(), ChangeValue().
• BaseCollection class: abstract collection of item with base class. API:
CreateIterator(), Size(), Assign().
• Heterogeneous Assign: accept assignment via other collection
type.
• Allocator: 預設allocator實現Standard::Allocate()與Standard::Free().
NCollection_IncAllocator是另一個allocator實作選項.

11. Vectors and Matrices
• 矩陣運算. Calculations involving vectors and matrices.
• Eigenvalues and Eigenvectors of a squared matrix.
• Solvers for linear algebraic equations.
• Find roots of a set of non-linear equations.
• Find minimum function of independent variables.

12. Primitive Geometric Types
• STEP-compliant implementation:
• Geometric shapes: points, vectors, lines, circles, conics,
planes and elementary surfaces.
• Geometric transformations: translation, rotation, symmetry,
scaling, composed transformations.
• Geometric processor package “gp”.
• Objects are manipulated by value.
• gp curves/surfaces are analytic (no parameterization and
orientation). Use Geom entities for parameterization.
• EICLib package: point  parameter on gp curves.
• EISLib package: point  parameter on gp surfaces.

13. Math Algorithms
• Solvers for linear algebraic (前面提過).
• Ex: Bissec + Newton.
• Find root of F(x)=x^2 – 4,in [1.5, 2.5]

14. Precision
• 每個物件擁有自己個別的精確度資訊.
• 判斷數值相等: if( abs(x2 – x1) < precision ) …
• Parametric space, 精準度常被要求在參數空間與真實座
標空間結果一致.
• Precision package:
• Angular: compare angles.
• Confusion: compare distances.
• Intersection: used by intersection algorithms.
• Approximation: used by approximation algorithms.
• Parametric: derived from 3D space precision.