C++ Coding Standards: 101 Rules, Guidelines, and Best Practices, 1st edition

Published by Addison-Wesley Professional (October 25, 2004) © 2005

  • Herb Sutter
  • Andrei Alexandrescu
£44.49

  • A print text (hardcover or paperback) 
  • Free shipping
  • Also available for purchase as an ebook from all major ebook resellers, including InformIT.com

Every software development team should have and follow a coding standard.

It's even better when what the coding standard requires is actually consistent,

reasonable, and correct.

Coding standards have many advantages:

*They improve code quality. This happens automatically when following a

good, simple set of guidelines.

*They improve development speed, because the programmer doesn't need to

always make decisions starting from first principles.

*They enhance teamwork by eliminating needless debates on inconsequential

issues and by making it easy for teammates to read and maintain each other's

code.

The coding standards introduced by this book are a collection of guidelines for

writing high-quality C++ code.

***They are the distilled conclusions of a rich collective experience of the C++

community. Until now, this body of knowledge has been available only as

folklore or spread in bits and pieces throughout books.

Essential guidelines for writing high-quality C++ code-- taught by two foremost masters in the field.

° Provides a collection of programming rules (101) that is terse, easy to understand and follow, and self-sufficient.

° Teaches basic rules and widely-accepted and authoritative practices for writing sound software in C++.

° Herb Sutter and Andrei Alexandrescu are two of the world's foremost experts on software development in general and C++ in particular.

Preface.

1. Organizational and Policy Issues.

    Don't sweat the small stuff. (Or: Know what not to standardize.).

    Compile cleanly at high warning levels.

    Use an automated build system.

    Use a version control system.

    Invest in code reviews.

2. Design Style.

    Give one entity one cohesive responsibility.

    Correctness, simplicity, and clarity come first.

    Know when and how to code for scalability.

    Don't optimize prematurely.

    Don't pessimize prematurely.

    Minimize global and shared data.

    Hide information.

    Know when and how to code for concurrency.

    Ensure resources are owned by objects. Use explicit RAII and smart pointers.

3. Coding Style.

    Prefer compile- and link-time errors to run-time errors.

    Use const proactively.

    Avoid macros.

    Avoid magic numbers.

    Declare variables as locally as possible.

    Always initialize variables.

    Avoid long functions. Avoid deep nesting.

    Avoid initialization dependencies across compilation units.

    Minimize definitional dependencies. Avoid cyclic dependencies.

    Make header files self-sufficient.

    Always write internal #include guards. Never write external #include guards.

4. Functions and Operators.

    Take parameters appropriately by value, (smart) pointer, or reference.

    Preserve natural semantics for overloaded operators.

    Prefer the canonical forms of arithmetic and assignment operators.

    Prefer the canonical form of ++ and --. Prefer calling the prefix forms.

    Consider overloading to avoid implicit type conversions.

    Avoid overloading &&, ||, or , (comma).

    Don't write code that depends on the order of evaluation of functionarguments.

5. Class Design and Inheritance.

    Be clear what kind of class you're writing.

    Prefer minimal classes to monolithic classes.

    Prefer composition to inheritance.

    Avoid inheriting from classes that were not designed to be base classes.

    Prefer providing abstract interfaces.

    Public inheritance is substitutability.

    Inherit, not to reuse, but to be reused.

    Practice safe overriding.

    Consider making virtual functions nonpublic, and public functions nonvirtual.

    Avoid providing implicit conversions.

    Make data members private, except in behaviorless aggregates (C-stylestructs).

    Don't give away your internals.

    Pimpl judiciously.

    Prefer writing nonmember nonfriend functions.

    Always provide new and delete together.

    If you provide any class-specific new, provide all of the standard forms (plain, in-place, and nothrow).

6. Construction, Destruction, and Copying.

    Define and initialize member variables in the same order.

    Prefer initialization to assignment in constructors.

    Avoid calling virtual functions in constructors and destructors.

    Make base class destructors public and virtual, or protected and nonvirtual.

    Destructors, deallocation, and swap never fail.

    Copy and destroy consistently.

    Explicitly enable or disable copying.

    Avoid slicing. Consider Clone instead of copying in base classes.

    Prefer the canonical form of assignment.

    Whenever it makes sense, provide a no-fail swap (and provide it correctly).

7. Namespaces and Modules.

    Keep a type and its nonmember function interface in the same namespace.

    Keep types and functions in separate namespaces unless they're specifically intended to work together.

    Don't write namespace usings in a header file or before an #include.

    Avoid allocating and deallocating memory in different modules.

    Don't define entities with linkage in a header file.

    Don't allow exceptions to propagate across module boundaries.

    Use sufficiently portable types in a module's interface.

8. Templates and Genericity.

    Blend static and dynamic polymorphism judiciously.

    Customize intentionally and explicitly.

    Don't specialize function templates.

    Don't write unintentionally nongeneric code.

9. Error Handling and Exceptions.

    Assert liberally to document internal assumptions and invariants.

    Establish a rational error handling policy, and follow it strictly.

    Distinguish between errors and non-errors.

     Design and write error-safe code.

    Prefer to use exceptions to report errors.

    Throw by value, catch by reference.

    Report, handle, and translate errors appropriately.

    Avoid exception specifications.

10. STL: Containers.

    Use vector by default. Otherwise, choose an appropriate container.

    Use vector and string instead of arrays.

    Use vector (and string::c_str) to exchange data with non-C++ APIs.

    Store only values and smart pointers in containers.

    Prefer push_back to other ways of expanding a sequence.

    Prefer range operations to single-element operations.

    Use the accepted idioms to really shrink capacity and really erase elements.

11. STL: Algorithms.

    Use a checked STL implementation.

    Prefer algorithm calls to handwritten loops.

    Use the right STL search algorithm.

    Use the right STL sort algorithm.

    Make predicates pure functions.

    Prefer function objects over functions as algorithm and comparer arguments.

    Write function objects correctly.

12. Type Safety.

    Avoid type switching; prefer polymorphism.

    Rely on types, not on representations.

    Avoid using reinterpret_cast.

    Avoid using static_cast on pointers.

    Avoid casting away const.

    Don't use C-style casts.

    Don't memcpy or memcmp non-PODs.

    Don't use unions to reinterpret representation.

    Don't use varargs (ellipsis).

    Don't use invalid objects. Don't use unsafe functions.

    Don't treat arrays polymorphically.

Bibliography.

Summary of Summaries.

Index.

Herb Sutter is the author of three highly acclaimed books, Exceptional C++ Style, Exceptional C++, and More Exceptional C++ (Addison-Wesley). He chairs the ISO C++ standards committee, and is contributing editor and columnist for C/C++ Users Journal. As a software architect for Microsoft, Sutter leads the design of C++ language extensions for .NET programming.

Andrei Alexandrescu is the author of the award-winning book Modern C++ Design (Addison-Wesley, 2001) and is a columnist for C/C++ Users Journal.


Need help? Get in touch

Video
Play
Privacy and cookies
By watching, you agree Pearson can share your viewership data for marketing and analytics for one year, revocable by deleting your cookies.

Pearson eTextbook: What’s on the inside just might surprise you

They say you can’t judge a book by its cover. It’s the same with your students. Meet each one right where they are with an engaging, interactive, personalized learning experience that goes beyond the textbook to fit any schedule, any budget, and any lifestyle.