Crs 120
Frameworks, Patterns and the C++ Standard Template Library

This course focuses on the approaches and techniques needed in the implementation of large C++ based projects. In particular it concentrates on the use of the Standard Template Library - to obviate the need to re-invent the code for maintaining and manipulating complex data sets, and on the exploitation of patterns - to come up with useful approaches to tackling some of the design and implementation problems that crop up on large projects.

By combining a good coverage of the STL, and a comprehensive overview of Patterns that have been researched over the last 10 year or so the course should provide senior programmers, analysts and technical project managers with the necessary skills and insights to develop large scale applications more rapidly and with fewer bugs in the code.

An important area covered in the course is directed to the art and science of rapidly prototyping applications that will involve the manipulation of rich and complex collections of data (in areas such as data exploration, image processing, operations research and network management).

Course outline

Polishing up C++ knowledge

  • Rapid overview of the basic syntax
  • Namespaces
  • type conversion (casting) operators
  • overloading the new and free operators
  • function objects
  • pointers to member functions of a class
  • Virtual class hierarchies and pure virtual classes
  • static member variables and static member functions
  • templates
  • the templatised function make_pair()
  • the auto_ptr class
  • C++ exception handling - try, catch, throw

Rationale behind the Standard Template Library

  • An introduction and overview of algorithms and their associated data structures
  • specification and description of algorithms
  • Characterisation of the Complexity of algorithms (Big-O notation)
  • verification , validation and testing of algorithms
  • An overview of typical container classes, their underlying abstractions and their associated algorithms

Standard Template Library - Components

  • Sequence containers
  • Associative containers
  • Container adapters
  • Iterators
  • Algorithms
  • Iterator adaptors
  • Exception handling inside STL

Container classes - details and applications

  • Vectors
  • Lists
  • Deques
  • Sets and Multisets
  • Maps and Multimaps
  • Strings as STL containers
  • Arrays as STL containers
  • Hash tables

STL Iterators

  • Input, Output Iterators
  • Forward, Bidirectional and Random Iterators
  • Iterator adaptors - Reverse, Insert and Stream Iterators
  • Auxiliary Iterator Functions - advance(), distance(), iter_swap()
  • Iterator traits

STL Function Objects

  • Used as sorting criteria
  • Used to maintain Internal state
  • Used as a return value of for_each()
  • Function adapters

STL Algorithms

  • for_each()
  • Counting, min/max, searching and range comparison algorithms
  • Algorithms for - copying, transforming and combining, assigning new values, and replacing elements in a container
  • Algorithms for reversing order of, rotating, permuting, shuffling elements and moving elements to the front
  • Sorting algorithms
  • Numeric algorithms

Special containers

  • Stacks
  • Queues
  • Priority Queues
  • Bitsets

Other STL Components

  • Strings
  • IO Stream classes
  • Complex numbers
  • Valarrays
  • Internationalisation
  • Allocators

Patterns

  • Rationale Behind Patterns
  • Patterns - a brief history
  • creational patterns (abstract factory, prototype, singleton)
  • structural patterns (adaptor, composite, facade, flyweight, proxy)
  • behavioural patterns (interpreter, iterator, mediator, memento, observer , visitor)
  • meta-patterns
  • anti-patterns

Strategies and approaches for large projects

  • Object oriented approach to requirements analysis
  • Framework adaptation patterns
  • Design pattern catalogs
  • RAD and JAD approaches
  • Formal contract approaches

Components in Large Projects

  • Identifying components (building blocks)
  • creating hierarchies of components with no physical dependency loops (acyclic physical dependencies)
  • reduction of link-time dependencies (levelisation)
  • reduction of compile-time dependencies (insulation)
  • larger building blocks (packages)

Large projects - logical design, management and implementation issues

  • architecting components
  • designing functional interfaces
  • management and implementation of objects in a large-project environment
  • integration , validation and verification
  • version control and maintenance

Intended Audience

The course is for experienced C++ programmers, analysts and technical project managers who would like to gain a good overview of patterns and their uses, and of the Standard Template Library and its potential.

Attendees are expected to be comfortable working with projects containing many modules, have a good knowledge of debugging techniques, and capable of interpreting sometimes obscure compiler errors and warnings

This course is designed to be operating system, compiler and integrated development environment independent, although, it will by default be taught using workstations running Linux and the GNU C++ compiler

Key Skills

  • Knowledge of the major Patterns and how they can be implemented in C++
  • Use of the Standard Template Library to develop complex data handling processing applications
  • Implementing graph representation and graph traversal algorithms using the STL
  • Realising STL container class persistence with Object Oriented Databases
  • Ability to recognise patterns and situations where certain patterns are applicable
  • Ability to recognise situations where multiple patterns need to be co-ordinated and applied
  • Implementing designs involving Singleton, Observer, Composite, Visitor and Proxy patterns in C++
  • Understanding component and pattern based approaches in the implementation

Practical Work

The course contains a number of challenging exercises. For many of the exercises a lot of the supporting code will be provided, so that students can concentrate on applying one or more key concepts.

Exercises will (typically) consist of some simple warm up mini-problems, followed by a core problem, followed by several challenges of increasing difficulty. The challenges are meant for post course skills development.

The course will use the GNU C++ compiler running under Linux. (It can also be run using Visual Studio under Microsoft Windows). The course hours will be from 9 a.m. to 5.30 p.m. for the first four days, and from 9 a.m. to 4 p.m. on the last day. Students are encouraged to come earlier (8 a.m. or to carry on working after class (till 7.00 p.m.) if they wish to. Where students have their own laptops (running either Linux or Solaris) they can work on the challenge exercises at home, or at their hotel in the evenings.

  • Implementing a graph search algorithm using the STL
  • Realising a simulation of a queueing system using STL
  • Using a variety of patterns and STL containers to implement a distributed command, control and data acquisition system
  • Implementing data persistence using an Object Oriented Database Management System
  • Developing a graphical data display system using a variety of patterns and containers