Safe Namespace Reference

Arithmetic operations with overflow checks. More...

Namespaces
	Internal
	Helper functions for providing integer overflow checks.

Functions
template<typename T >
T	add (T summand_1, T summand_2)
	Safe addition, throws an exception on overflow. More...
template<typename T >
Internal::enable_if< Internal::is_signed< T >::VALUE, T >::type	abs (T num) throw ()
	Calculates the absolute value of a number without producing negative values. More...

Detailed Description

Arithmetic operations with overflow checks.

Function Documentation

abs()

template<typename T >

Internal::enable_if<Internal::is_signed<T>::VALUE, T>::type Safe::abs	(	T	num	)
throw	(
	)

Calculates the absolute value of a number without producing negative values.

The "standard" implementation of abs(num) (num < 0 ? -num : num) produces negative values when num is the smallest negative number. This is caused by -1 * INTMAX = INTMIN + 1, i.e. the real result of abs(INTMIN) overflows the integer type and results in INTMIN again (this is not guaranteed as it invokes undefined behavior).

This function does not exhibit this behavior, it returns std::numeric_limits<T>::max() when the input is std::numeric_limits<T>::min(). The downside of this is that two negative values produce the same absolute value: std::numeric_limits<T>::min() and std::numeric_limits<T>::min() + 1.

Template Parameters

T	a signed integer type

Parameters

[in]

num

The number which absolute value should be computed.

Exceptions

Never

throws an exception.

Returns

The absolute value of num or std::numeric_limits<T>::max() when num == std::numeric_limits<T>::min().

Examples:

geotag.cpp.

Referenced by Exiv2::TimeValue::copy(), Exiv2::floatToRationalCast(), Exiv2::Internal::print0x9204(), Exiv2::Internal::Nikon3MakerNote::printTimeZone(), and Exiv2::TimeValue::write().

add()

template<typename T >

T Safe::add	(	T	summand_1,
		T	summand_2
	)

Safe addition, throws an exception on overflow.

This function returns the result of summand_1 and summand_2 only when the operation would not overflow, otherwise an exception of type std::overflow_error is thrown.

Parameters

[in]

summand_1,summand_2

summands to be summed up

Returns

the sum of summand_1 and summand_2

Exceptions

std::overflow_error

if the addition would overflow

This function utilizes compiler builtins when available and should have a very small performance hit then. When builtins are unavailable, a more extensive check is required.

Builtins are available for the following configurations:

• GCC/Clang for signed and unsigned int, long and long long (not char & short)
• MSVC for unsigned int, long and long long

Referenced by Exiv2::Image::printIFDStructure(), Exiv2::WebPImage::readMetadata(), Exiv2::Jp2Image::readMetadata(), Exiv2::RafImage::readMetadata(), and Exiv2::JpegBase::readMetadata().