libfixedpointnumber

libfixedpointnumber is a Library for fixed point number by C++11.

Summary

Usual progrmas computing fixed point number,

• Bit shift operations are necessary, and they lose readability
• Types which holding fixed point numbers are just standard integral types, it also lose readability
• Not type-safe because types for different-spec fixed point numbers are just standard integral types (they include no Q bit width information anywhere)

If libfixedpointnumber is used in programs computing fixed point number...

• Bit shift operations can be hidden
• Specs of fixed point number (holding standard integral types and how many bits assigned for fractional part) are explicit
• Can not substitute a fixed point number into another spec fixed point number without explicit casting

libfixedpointnumber provides...

• Template class fixed_t to represent fixed point number
• Arithmetic and comparison operators for fixed_t
• Casting as conversion between different spec fixed_t (different bit width or different Q bit width)
• Convert from string like as "3.14" into fixed_t
• Functions for fixed_t which are compatible to part of functions in cmath
• Most members and releated functions of fixed_t are provided as constexpr functions

API Reference

• By Doxygen

Integrate into your project

1. Include header files in libfixedpointnumber/include/
2. Add libfixedpointnumber/include/ into your include path.
3. Compile your codes

It is not necessary to build and link static link library of libfixedpointnumber because all implementations are available in header files.

Usage

Following simple sample code provides to calculate half of input number with converting precision (Latest code of this is in interactive_sample/).

#include <cstdint>
#include <iostream>

#include "fixedpointnumber.h"

using fixed_Q7_t = fixedpointnumber::fixed_t<int16_t, 7>;
using fixed_Q8_t = fixedpointnumber::fixed_t<int16_t, 8>;

int main(int, char**) {
  std::cout << "Input number to divide (less than 128): ";

  fixed_Q7_t num(0);
  std::cin >> num;

  std::cout << "Half of "
            << num
            << " is "
            << fixed_Q8_t(num) / fixed_Q8_t(2)
            << std::endl;

  return 0;
}

You can see following items in this sample

• Input/Output stream operators are supported
• Overloaded basic arithmetic operators are available
• fixed_t can be construct from constant integer

Build samples and unit tests

Makefile of libfixedpointnumber will provide followings on your environments

• Build and run unit tests
• Build and run sample programs
• Some checks
  • Coding rule check by cpplint
  • Static analysis by cppcheck
• Build documents with Doxygen

List of build targets

make target	How it works
all	Same as build-all
build-all	build-sample and build-test
build-sample	Build sample programs
build-test	Build unit tests
check	Process cppcheck and cpplint
clean	Remove generated files
coverage	Create coverage report into out/site/coverage (Must use with BUILD_TYPE=coverage)
cppcheck	Static analytics by cppcheck
cpplint	Lint by cpplint
doc	coverage and doxygen
doxygen	Generate doxygen HTML documents into out/site/Doxygen
latex	Generate doxygen LaTeX documents into out/site/Doxygen
pdf	Generate doxygen PDF documents into out/site/Doxygen
run-all	run-sample and run-test
run-sample	Build (if necessary) and run sample programs
run-test	Build (if necessary) and run unit tests
site	Build tree for project site

Build options

BUILD_TYPE

BUILD_TYPE	How it works
debug	All optimizations are disabled in build
release	Optimizations are enabled in build
coverage	Options for build coverage as make target

• In coverage, optimization are disabled and some additional information to get coverage report will be generated

Build and run unit tests

1. Install Google Test
2. make run-test in this directory

Build coverage report

1. Install gcovr
2. make BUILD_TYPE=coverage coverage, so html report is put into out/coverage_html/

Build and run sample programs

1. make run-sample in this directory

Checks

Some check targets are available in build by Makefile

cpplint

• Install cpplint.py
• Set that directory in your environment variable PATH
• make cpplint

or

• make CPPLINT=your/installed/directory/cpplint.py cpplint

cppcheck

• Install cppechck into your environment
• make cppcheck

Generate Doxygen document files

• Install doxygen into your environment
• make doc

For Windows (Visual Studio)

• The solution file to build samples and tests with Visual Studio is available as build/win_vs/libfixedpointnumber/libfixedpointnumber.sln
• Google Test Framework will be recovered automatically at opening solution file

Environments

As of 2021/02/12, all tests succeed and all samples are working well on following environments

OS	Compiler
FreeBSD 12.2-STABLE	clang++
macOS Catalina 10.15.7	g++ (symlink to clang)
Ubuntu 16.04	g++
Ubuntu 18.04	g++
Windows Server 2016 Datacenter	Visual C++