cgal/Packages/Tutorial/tutorial/Polyhedron/sgp2004/paper/prereq.tex

66 lines
3.4 KiB
TeX

% Related work and background
% C++ and generic programming
\noindent \textbf{Design patterns and generic programming.}
Before using CGAL, it is mandatory to be familiar with C++,
\emph{design patterns} and the \emph{generic programming paradigm}.
Design patterns \cite{Gamma:1995:DP} describe successful and
\emph{reusable} solutions to standard software problems. The generic
programming paradigm \cite{Alexandrescu:2001:MCD} features the notion
of C++ class templates and function templates, which is at the corner
stone of all features provided by CGAL. Details of the design of the
CGAL polyhedron data structure are given in next section.
An example illustrating design patterns and the generic programming is
the Standard Template Library (STL)~\cite{ms-stl-96}. Generality and
flexibility is achieved with a set of \italic{concepts}, where a
concept is a well defined set of requirements. One of them is the
\italic{iterator} concept, which allows both referring to an item and
traversing a sequence of items. Those items are stored in a data
structure called \italic{container} in STL. Another concept, so-called
\italic{circulator}, allows traversing some circular sequences. They
share most of the requirements with iterators, except the lack of
past-the-end position in the sequence. Since CGAL is strongly inspired
from the generality of STL, it is important to become familiar with
its concepts before starting using it.
%% The generic programming paradigm provides a
%% soft-coupled framework to support data structures and algorithms.
%% It focuses on representing families of the data structures
%% or the algorithms by identifying the \emph{abstract representations}.
%% With the \CC\ template, the standard template library
%% \cite{ms-stl-96} (STL) is the pioneer library based on
%% the generic programming paradigm. The STL provides a set of
%% data structure families, such as vectors, lists and trees,
%% and a set of generic algorithms that can be dynamicly
%% coupled with a data structure. In general,
%% the abstract representation of a data structure family
%% is the \emph{entity type} and of an algorithm family
%% is the \emph{operation type}.
%% %For example are the entity sorting and searching where users
%% %supply the equality predicates to specialize the algorithms.
%% The STL frees software engineers from rebuilding the fundamental
%% data structures and let them focus on the design of the algorithms.
\noindent \textbf{Halfedge data structure.}
\input hds
\noindent \textbf{Geometry data structure and algorithms}
A subdivision surface is the limit surface resulted from the
application of a subdivision algorithm to a control polyhedron.
Subdivision algorithms recursively \emph{refine} (subdivide) the
control polyhedron and \emph{modify} (smooth) the geometry according
to the stencils of the source mesh.
%Subdivisions consist of two
%meta-steps that most geometry processing algorithms have: the
%\emph{connectivity operation} and the
%\emph{geometry operation}. In this paper, we use subdivision
%as the template of the geometry processing algorithms because it
%connectivity operations are more complex than most other
%algorithms. Some other simple algorithms, e.g.\ displacement map, the
%connectivity operation is a null operation. In other words, the source
%and the target mesh have the same connectivity.
Further details on
subdivisions can be found at \cite{Sub:course:2000}.