Merge branch 'main' into dlemire/adding_bloat_analysis

.github/workflows/alpine.yml

@@ -1,27 +1,47 @@
 name: Alpine Linux
-'on':
+on:
   - push
   - pull_request
+
 jobs:
-  ubuntu-build:
+  build:
+    name: Build on Alpine ${{ matrix.arch }}
     runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        arch:
+          - x86_64
+          - x86
+          - aarch64
+          - armv7
+          - ppc64le
+          - riscv64
     steps:
-      - uses: actions/checkout@v2
+      - name: Checkout repository
-      - name: start docker
+        uses: actions/checkout@v1
+
+      - name: Install latest Alpine Linux for ${{ matrix.arch }}
+        uses: jirutka/setup-alpine@v1
+        with:
+          arch: ${{ matrix.arch }}
+          branch: ${{ matrix.arch == 'riscv64' && 'edge' || 'latest-stable' }}
+          packages: >
+            build-base
+            cmake
+            g++
+            linux-headers
+            git
+            bash
+            build-base
+      - name: Prepare
         run: |
-          docker run -w /src -dit --name alpine -v $PWD:/src alpine:latest
+          cmake -DFASTFLOAT_TEST=ON -B build
-          echo 'docker exec alpine "$@";' > ./alpine.sh
+        shell: alpine.sh {0}
-          chmod +x ./alpine.sh
+      - name: Build
-      - name: install packages
         run: |
-          ./alpine.sh apk update
+          cmake --build build
-          ./alpine.sh apk add build-base cmake g++ linux-headers git bash
+        shell: alpine.sh {0}
-      - name: cmake
+      - name: Test
         run: |
-          ./alpine.sh cmake -DFASTFLOAT_TEST=ON  -B build_for_alpine
+          ctest --test-dir build -R basictest
-      - name: build
+        shell: alpine.sh {0}
-        run: |
-          ./alpine.sh cmake --build build_for_alpine
-      - name: test
-        run: |
-          ./alpine.sh bash -c "cd build_for_alpine && ctest  -R basictest"

.github/workflows/amalgamate-ubuntu20.yml

@@ -1,20 +1,12 @@
-name: Amalgamate Ubuntu 20.04 CI (GCC 9, 8)
+name: Amalgamate Ubuntu 20.04 CI (GCC 9)
 
 on: [push, pull_request]
 
 jobs:
   ubuntu-build:
     runs-on: ubuntu-20.04
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          # Legacy/x86 compilers cause CI failures.
-          #- {cxx: -DCMAKE_CXX_COMPILER=g++-8, arch:                         }
-          - {cxx:                           , arch:                         } # default=gcc9
-          #- {cxx:                           , arch: -DCMAKE_CXX_FLAGS="-m32"} # default=gcc9
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
       - name: Compile with amalgamation
         run: |
           mkdir build &&

.github/workflows/msys2-clang.yml

@@ -23,7 +23,7 @@ jobs:
       CMAKE_GENERATOR: Ninja
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
       - uses: msys2/setup-msys2@v2
         with:
           update: true

.github/workflows/msys2.yml

@@ -29,7 +29,7 @@ jobs:
       CMAKE_GENERATOR: Ninja
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
       - uses: msys2/setup-msys2@v2
         with:
           update: true

.github/workflows/s390x.yml

@@ -0,0 +1,30 @@
+name: Ubuntu s390x (GCC 11)
+
+on:
+  push:
+    branches:
+      - master
+  pull_request:
+    branches:
+      - master
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - uses: uraimo/run-on-arch-action@v2
+        name: Test
+        id: runcmd
+        with:
+          arch: s390x
+          githubToken: ${{ github.token }}
+          distro: ubuntu_latest
+          install: |
+            apt-get update -q -y
+            apt-get install -y cmake make g++
+          run: |
+            cmake -DCMAKE_BUILD_TYPE=Release -B build
+            cmake --build build -j=2
+            ctest --output-on-failure --test-dir build
+

.github/workflows/ubuntu18.yml

@@ -1,21 +1,12 @@
-name: Ubuntu 18.04 CI (GCC 7, 6, 5)
+name: Ubuntu 18.04 CI (GCC 7)
 
 on: [push, pull_request]
 
 jobs:
   ubuntu-build:
     runs-on: ubuntu-18.04
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          # Legacy/x86 compilers cause CI failures.
-          #- {cxx: -DCMAKE_CXX_COMPILER=g++-5, arch:                         }
-          #- {cxx: -DCMAKE_CXX_COMPILER=g++-6, arch:                         }
-          - {cxx:                           , arch:                         } # default=gcc7
-          #- {cxx:                           , arch: -DCMAKE_CXX_FLAGS="-m32"} # default=gcc7
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
       - name: Setup cmake
         uses: jwlawson/actions-setup-cmake@v1.4
         with:

.github/workflows/ubuntu20-cxx20.yml

@@ -8,7 +8,7 @@ jobs:
     strategy:
       fail-fast: false
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
       - name: Use cmake
         run: |
           mkdir build &&

.github/workflows/ubuntu20-fastmath.yml

@@ -0,0 +1,16 @@
+name: Ubuntu 20.04 CI (GCC 9, fast-math)
+
+on: [push, pull_request]
+
+jobs:
+  ubuntu-build:
+    runs-on: ubuntu-20.04
+    steps:
+      - uses: actions/checkout@v3
+      - name: Use cmake
+        run: |
+          mkdir build &&
+          cd build &&
+          cmake -DCMAKE_CXX_FLAGS="-ffast-math" -DFASTFLOAT_TEST=ON ..  &&
+          cmake --build .   &&
+          ctest --output-on-failure

.github/workflows/ubuntu20.yml

@@ -5,16 +5,8 @@ on: [push, pull_request]
 jobs:
   ubuntu-build:
     runs-on: ubuntu-20.04
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          # Legacy/x86 compilers cause CI failures.
-          #- {cxx: -DCMAKE_CXX_COMPILER=g++-8, arch:                         }
-          - {cxx:                           , arch:                         } # default=gcc9
-          #- {cxx:                           , arch: -DCMAKE_CXX_FLAGS="-m32"} # default=gcc9
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
       - name: Use cmake
         run: |
           mkdir build &&

.github/workflows/ubuntu22-clang.yml

@@ -0,0 +1,18 @@
+name: Ubuntu 22.04 CI (clang 14)
+
+on: [push, pull_request]
+
+jobs:
+  ubuntu-build:
+    runs-on: ubuntu-22.04
+    steps:
+      - uses: actions/checkout@v3
+      - name: Install clang++-14
+        run: sudo apt-get install -y clang++-14
+      - name: Use cmake
+        run: |
+          mkdir build &&
+          cd build &&
+          CXX=clang++-14 cmake -DFASTFLOAT_TEST=ON  ..  &&
+          cmake --build .   &&
+          ctest --output-on-failure 

.github/workflows/ubuntu22-gcc12.yml

@@ -0,0 +1,16 @@
+name: Ubuntu 22.04 CI (GCC 12)
+
+on: [push, pull_request]
+
+jobs:
+  ubuntu-build:
+    runs-on: ubuntu-22.04
+    steps:
+      - uses: actions/checkout@v3
+      - name: Use cmake
+        run: |
+          mkdir build &&
+          cd build &&
+          CXX=g++-12 CXXFLAGS=-Werror cmake -DFASTFLOAT_TEST=ON  ..  &&
+          cmake --build .   &&
+          ctest --output-on-failure 

.github/workflows/ubuntu22.yml

@@ -0,0 +1,16 @@
+name: Ubuntu 22.04 CI (GCC 11)
+
+on: [push, pull_request]
+
+jobs:
+  ubuntu-build:
+    runs-on: ubuntu-22.04
+    steps:
+      - uses: actions/checkout@v3
+      - name: Use cmake
+        run: |
+          mkdir build &&
+          cd build &&
+          cmake -DFASTFLOAT_TEST=ON  ..  &&
+          cmake --build .   &&
+          ctest --output-on-failure 

.github/workflows/vs15-ci.yml

@@ -1,29 +0,0 @@
-name: VS15-CI
-
-on: [push, pull_request]
-
-jobs:
-  ci:
-    if: >-
-      ! contains(toJSON(github.event.commits.*.message), '[skip ci]') &&
-      ! contains(toJSON(github.event.commits.*.message), '[skip github]')
-    name: windows-vs15
-    runs-on: windows-2016
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - {gen: Visual Studio 15 2017, arch: Win32}
-          - {gen: Visual Studio 15 2017, arch: x64}
-    steps:
-    - uses: actions/checkout@v2
-    - name: Configure
-      run: |
-          mkdir build
-          cd build && cmake  -G "${{matrix.gen}}" -A ${{matrix.arch}} -DFASTFLOAT_TEST=ON ..
-    - name: Build
-      run: cmake --build build --verbose --config Release --parallel
-    - name: 'Run CTest'
-      run: |
-          cd build
-          ctest -C Release --output-on-failure -R basictest

.github/workflows/vs16-arm-ci.yml

@@ -1,21 +0,0 @@
-name: VS16-ARM-CI
-
-on: [push, pull_request]
-
-jobs:
-  ci:
-    name: windows-vs16
-    runs-on: windows-latest
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - {arch: ARM}
-          - {arch: ARM64}
-    steps:
-      - name: checkout
-        uses: actions/checkout@v2
-      - name: Use cmake
-        run: |
-          cmake -A ${{ matrix.arch }} -DCMAKE_CROSSCOMPILING=1 -DFASTFLOAT_TEST=ON -B build  &&
-          cmake --build build --verbose

.github/workflows/vs16-ci.yml

@@ -1,29 +0,0 @@
-name: VS16-CI
-
-on: [push, pull_request]
-
-jobs:
-  ci:
-    name: windows-vs16
-    runs-on: windows-latest
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - {gen: Visual Studio 16 2019, arch: Win32}
-          - {gen: Visual Studio 16 2019, arch: x64}
-    steps:
-      - name: checkout
-        uses: actions/checkout@v2
-      - name: Use cmake
-        run: |
-          mkdir build &&
-          cd build &&
-          cmake  ${{matrix.cxx}} ${{matrix.arch}} -DFASTFLOAT_TEST=ON -DCMAKE_INSTALL_PREFIX:PATH=destination ..  &&
-          cmake --build .  --verbose &&
-          ctest --output-on-failure  &&
-          cmake --install . &&
-          cd ../tests/installation_tests/find  &&
-          mkdir build && cd build && cmake -DCMAKE_INSTALL_PREFIX:PATH=../../../build/destination .. &&  cmake --build .   --verbose
-          cd ../../issue72_installation  &&
-          mkdir build && cd build && cmake -DCMAKE_INSTALL_PREFIX:PATH=../../../build/destination .. &&  cmake --build .    --verbose

.github/workflows/vs16-clang-ci.yml

@@ -1,27 +0,0 @@
-name: VS16-CLANG-CI
-
-on: [push, pull_request]
-
-jobs:
-  ci:
-    name: windows-vs16
-    runs-on: windows-latest
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - {gen: Visual Studio 16 2019, arch: Win32}
-          - {gen: Visual Studio 16 2019, arch: x64}
-    steps:
-      - name: checkout
-        uses: actions/checkout@v2
-      - name: Configure
-        run: |
-          mkdir build
-          cd build && cmake -G "${{matrix.gen}}" -A ${{matrix.arch}} -T ClangCL -DFASTFLOAT_TEST=ON ..
-      - name: Build
-        run: cmake --build build --config Release --parallel
-      - name: Run basic tests
-        run: |
-          cd build
-          ctest -C Release --output-on-failure -R basictest

.github/workflows/vs16-cxx20.yml

@@ -1,24 +0,0 @@
-name: VS16-CI C++20
-
-on: [push, pull_request]
-
-jobs:
-  ci:
-    name: windows-vs16
-    runs-on: windows-latest
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - {gen: Visual Studio 16 2019, arch: Win32}
-          - {gen: Visual Studio 16 2019, arch: x64}
-    steps:
-      - name: checkout
-        uses: actions/checkout@v2
-      - name: Use cmake
-        run: |
-          mkdir build &&
-          cd build &&
-          cmake  ${{matrix.cxx}} ${{matrix.arch}}  -DCMAKE_CXX_STANDARD=20 -DFASTFLOAT_TEST=ON -DCMAKE_INSTALL_PREFIX:PATH=destination ..  &&
-          cmake --build .  --verbose &&
-          ctest --output-on-failure

.github/workflows/vs17-arm-ci.yml

@@ -0,0 +1,30 @@
+name: VS17-ARM-CI
+
+on: [push, pull_request]
+
+jobs:
+  ci:
+    name: vs17/${{matrix.arch}}/${{matrix.cfg}}
+    runs-on: windows-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - {gen: Visual Studio 17 2022, arch: ARM, cfg: Release}
+          - {gen: Visual Studio 17 2022, arch: ARM, cfg: Debug}
+          - {gen: Visual Studio 17 2022, arch: ARM64, cfg: Release}
+          - {gen: Visual Studio 17 2022, arch: ARM64, cfg: Debug}
+    steps:
+      - name: checkout
+        uses: actions/checkout@v3
+      - name: configure
+        run: |
+          cmake -S . -B build -G "${{matrix.gen}}" -A ${{matrix.arch}} -DCMAKE_CROSSCOMPILING=1 -DFASTFLOAT_TEST=ON
+      - name: build
+        run: |
+          cmake --build build --verbose --config ${{matrix.cfg}} --parallel
+      # disabled because it requires a toolchain
+      #- name: test
+      #  run: |
+      #    cd build &&
+      #    ctest --output-on-failure -C ${{matrix.cfg}}

.github/workflows/vs17-ci.yml

@@ -0,0 +1,43 @@
+name: VS17-CI
+
+on: [push, pull_request]
+
+jobs:
+  ci:
+    name: vs17/${{matrix.arch}}/${{matrix.cfg}}
+    runs-on: windows-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - {gen: Visual Studio 17 2022, arch: Win32, cfg: Release}
+          - {gen: Visual Studio 17 2022, arch: Win32, cfg: Debug}
+          - {gen: Visual Studio 17 2022, arch: x64, cfg: Release}
+          - {gen: Visual Studio 17 2022, arch: x64, cfg: Debug}
+    steps:
+      - name: checkout
+        uses: actions/checkout@v3
+      - name: configure
+        run: |
+          cmake -S . -B build -G "${{matrix.gen}}" -A ${{matrix.arch}} -DFASTFLOAT_TEST=ON -DCMAKE_INSTALL_PREFIX:PATH=destination
+      - name: build
+        run: |
+          cmake --build build --verbose --config ${{matrix.cfg}} --parallel
+      - name: test
+        run: |
+          cd build &&
+          ctest --output-on-failure -C ${{matrix.cfg}}
+      - name: install
+        run: |
+          cd build &&
+          cmake --install .
+      - name: test install (find)
+        run: |
+          cd tests/installation_tests/find  &&
+          cmake -S . -B build -G "${{matrix.gen}}" -A ${{matrix.arch}} -DCMAKE_INSTALL_PREFIX:PATH=../../../build/destination &&
+          cmake --build build --verbose --config ${{matrix.cfg}} --parallel
+      - name: test install (issue 72)
+        run: |
+          cd tests/installation_tests/issue72_installation  &&
+          cmake -S . -B build -G "${{matrix.gen}}" -A ${{matrix.arch}} -DCMAKE_INSTALL_PREFIX:PATH=../../../build/destination &&
+          cmake --build build --verbose --config ${{matrix.cfg}} --parallel

.github/workflows/vs17-clang-ci.yml

@@ -0,0 +1,28 @@
+name: VS17-CLANG-CI
+
+on: [push, pull_request]
+
+jobs:
+  ci:
+    name: vs17/${{matrix.arch}}/${{matrix.cfg}}
+    runs-on: windows-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - {gen: Visual Studio 17 2022, arch: Win32, cfg: Release}
+          - {gen: Visual Studio 17 2022, arch: Win32, cfg: Debug}
+          - {gen: Visual Studio 17 2022, arch: x64, cfg: Release}
+          - {gen: Visual Studio 17 2022, arch: x64, cfg: Debug}
+    steps:
+      - name: checkout
+        uses: actions/checkout@v3
+      - name: Configure
+        run: |
+          cmake -S . -B build -G "${{matrix.gen}}" -A ${{matrix.arch}} -T ClangCL -DFASTFLOAT_TEST=ON
+      - name: Build
+        run: cmake --build build --config ${{matrix.cfg}} --parallel --verbose
+      - name: Run basic tests
+        run: |
+          cd build
+          ctest -C ${{matrix.cfg}} --output-on-failure -R basictest

.github/workflows/vs17-cxx20.yml

@@ -0,0 +1,29 @@
+name: VS17-CI C++20
+
+on: [push, pull_request]
+
+jobs:
+  ci:
+    name: vs17/${{matrix.arch}}/${{matrix.cfg}}
+    runs-on: windows-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - {gen: Visual Studio 17 2022, arch: Win32, cfg: Release}
+          - {gen: Visual Studio 17 2022, arch: Win32, cfg: Debug}
+          - {gen: Visual Studio 17 2022, arch: x64, cfg: Release}
+          - {gen: Visual Studio 17 2022, arch: x64, cfg: Debug}
+    steps:
+      - name: checkout
+        uses: actions/checkout@v3
+      - name: configure
+        run: |
+          cmake -S . -B build -G "${{matrix.gen}}" -A ${{matrix.arch}}  -DCMAKE_CXX_STANDARD=20 -DFASTFLOAT_TEST=ON -DCMAKE_INSTALL_PREFIX:PATH=destination ..
+      - name: build
+        run: |
+          cmake --build build --verbose --config ${{matrix.cfg}} --parallel
+      - name: test
+        run: |
+          cd build &&
+          ctest --output-on-failure -C ${{matrix.cfg}}

CMakeLists.txt

@@ -1,9 +1,7 @@
 cmake_minimum_required(VERSION 3.9)
 
-project(fast_float VERSION 3.2.0 LANGUAGES CXX)
+project(fast_float VERSION 3.10.0 LANGUAGES CXX)
 option(FASTFLOAT_TEST "Enable tests" OFF)
-set(CMAKE_CXX_STANDARD 11 CACHE STRING "C++ standard to be used")
-set(CMAKE_CXX_STANDARD_REQUIRED ON)
 if(FASTFLOAT_TEST)
   enable_testing()
   add_subdirectory(tests)
@@ -22,14 +20,21 @@ if (NOT CMAKE_BUILD_TYPE)
   endif()
 endif()
 
+option(FASTFLOAT_INSTALL "Enable install" ON)
+
+if(FASTFLOAT_INSTALL)
+  include(GNUInstallDirs)
+endif()
 
 add_library(fast_float INTERFACE)
+add_library(FastFloat::fast_float ALIAS fast_float)
 target_include_directories(
   fast_float
   INTERFACE
     $<BUILD_INTERFACE:${PROJECT_SOURCE_DIR}/include>
-    $<INSTALL_INTERFACE:include>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}>
 )
+target_compile_features(fast_float INTERFACE cxx_std_11)
 if(FASTFLOAT_SANITIZE)
   target_compile_options(fast_float INTERFACE -fsanitize=address -fno-omit-frame-pointer -fsanitize=undefined -fno-sanitize-recover=all)
   target_link_libraries(fast_float INTERFACE -fsanitize=address -fno-omit-frame-pointer -fsanitize=undefined -fno-sanitize-recover=all)
@@ -42,24 +47,30 @@ if(MSVC_VERSION GREATER 1910)
 endif()
 
 
-include(CMakePackageConfigHelpers)
+if(FASTFLOAT_INSTALL)
+  include(CMakePackageConfigHelpers)
 
-set(FASTFLOAT_VERSION_CONFIG "${CMAKE_CURRENT_BINARY_DIR}/module/FastFloatConfigVersion.cmake")
+  set(FASTFLOAT_VERSION_CONFIG "${CMAKE_CURRENT_BINARY_DIR}/module/FastFloatConfigVersion.cmake")
-set(FASTFLOAT_PROJECT_CONFIG "${CMAKE_CURRENT_BINARY_DIR}/module/FastFloatConfig.cmake")
+  set(FASTFLOAT_PROJECT_CONFIG "${CMAKE_CURRENT_BINARY_DIR}/module/FastFloatConfig.cmake")
-set(FASTFLOAT_INSTALL_DIR "share/FastFloat")
+  set(FASTFLOAT_CONFIG_INSTALL_DIR "${CMAKE_INSTALL_DATAROOTDIR}/cmake/FastFloat")
 
-write_basic_package_version_file("${FASTFLOAT_VERSION_CONFIG}" VERSION ${PROJECT_VERSION} COMPATIBILITY SameMajorVersion)
+  if(${CMAKE_VERSION} VERSION_LESS "3.14")
-configure_package_config_file("cmake/config.cmake.in"
+    write_basic_package_version_file("${FASTFLOAT_VERSION_CONFIG}" VERSION ${PROJECT_VERSION} COMPATIBILITY SameMajorVersion)
-                              "${FASTFLOAT_PROJECT_CONFIG}"
+  else()
-                              INSTALL_DESTINATION "${FASTFLOAT_INSTALL_DIR}")
+    write_basic_package_version_file("${FASTFLOAT_VERSION_CONFIG}" VERSION ${PROJECT_VERSION} COMPATIBILITY SameMajorVersion ARCH_INDEPENDENT)
+  endif()
+  configure_package_config_file("cmake/config.cmake.in"
+                                "${FASTFLOAT_PROJECT_CONFIG}"
+                                INSTALL_DESTINATION "${FASTFLOAT_CONFIG_INSTALL_DIR}")
 
-install(DIRECTORY "${PROJECT_SOURCE_DIR}/include/fast_float" DESTINATION "include")
+  install(DIRECTORY "${PROJECT_SOURCE_DIR}/include/fast_float" DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}")
-install(FILES "${FASTFLOAT_PROJECT_CONFIG}" "${FASTFLOAT_VERSION_CONFIG}" DESTINATION "${FASTFLOAT_INSTALL_DIR}")
+  install(FILES "${FASTFLOAT_PROJECT_CONFIG}" "${FASTFLOAT_VERSION_CONFIG}" DESTINATION "${FASTFLOAT_CONFIG_INSTALL_DIR}")
-install(EXPORT ${PROJECT_NAME}-targets NAMESPACE FastFloat:: DESTINATION "${FASTFLOAT_INSTALL_DIR}")
+  install(EXPORT ${PROJECT_NAME}-targets NAMESPACE FastFloat:: DESTINATION "${FASTFLOAT_CONFIG_INSTALL_DIR}")
 
-install(TARGETS fast_float 
+  install(TARGETS fast_float
-        EXPORT ${PROJECT_NAME}-targets
+          EXPORT ${PROJECT_NAME}-targets
-        RUNTIME DESTINATION bin
+          RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
-        ARCHIVE DESTINATION lib
+          ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
-        LIBRARY DESTINATION lib
+          LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
-)
+  )
+endif()

LICENSE-APACHE

@@ -175,18 +175,7 @@
 
    END OF TERMS AND CONDITIONS
 
-   APPENDIX: How to apply the Apache License to your work.
+   Copyright 2021 The fast_float authors
-
-      To apply the Apache License to your work, attach the following
-      boilerplate notice, with the fields enclosed by brackets "{}"
-      replaced with your own identifying information. (Don't include
-      the brackets!)  The text should be enclosed in the appropriate
-      comment syntax for the file format. We also recommend that a
-      file or class name and description of purpose be included on the
-      same "printed page" as the copyright notice for easier
-      identification within third-party archives.
-
-   Copyright 2020 The fast_float authors 
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

LICENSE-MIT

@@ -1,3 +1,7 @@
+MIT License
+
+Copyright (c) 2021 The fast_float authors
+
 Permission is hereby granted, free of charge, to any
 person obtaining a copy of this software and associated
 documentation files (the "Software"), to deal in the

README.md

@@ -1,12 +1,5 @@
 ## fast_float number parsing library: 4x faster than strtod
 
-![Ubuntu 20.04 CI (GCC 9)](https://github.com/lemire/fast_float/workflows/Ubuntu%2020.04%20CI%20(GCC%209)/badge.svg)
-![Ubuntu 18.04 CI (GCC 7)](https://github.com/lemire/fast_float/workflows/Ubuntu%2018.04%20CI%20(GCC%207)/badge.svg)
-![Alpine Linux](https://github.com/lemire/fast_float/workflows/Alpine%20Linux/badge.svg)
-![MSYS2-CI](https://github.com/lemire/fast_float/workflows/MSYS2-CI/badge.svg)
-![VS16-CLANG-CI](https://github.com/lemire/fast_float/workflows/VS16-CLANG-CI/badge.svg)
-[![VS16-CI](https://github.com/fastfloat/fast_float/actions/workflows/vs16-ci.yml/badge.svg)](https://github.com/fastfloat/fast_float/actions/workflows/vs16-ci.yml)
-
 The fast_float library provides fast header-only implementations for the C++ from_chars
 functions for `float` and `double` types.  These functions convert ASCII strings representing
 decimal values (e.g., `1.3e10`) into binary types. We provide exact rounding (including
@@ -28,8 +21,8 @@ struct from_chars_result {
 ```
 
 It parses the character sequence [first,last) for a number. It parses floating-point numbers expecting
-a locale-independent format equivalent to the C++17 from_chars function. 
+a locale-independent format equivalent to the C++17 from_chars function.
-The resulting floating-point value is the closest floating-point values (using either float or double), 
+The resulting floating-point value is the closest floating-point values (using either float or double),
 using the "round to even" convention for values that would otherwise fall right in-between two values.
 That is, we provide exact parsing according to the IEEE standard.
 
@@ -47,7 +40,7 @@ Example:
 ``` C++
 #include "fast_float/fast_float.h"
 #include <iostream>
- 
+
 int main() {
     const std::string input =  "3.1416 xyz ";
     double result;
@@ -60,15 +53,15 @@ int main() {
 
 
 Like the C++17 standard, the `fast_float::from_chars` functions take an optional last argument of
-the type `fast_float::chars_format`. It is a bitset value: we check whether 
+the type `fast_float::chars_format`. It is a bitset value: we check whether
 `fmt & fast_float::chars_format::fixed` and `fmt & fast_float::chars_format::scientific` are set
 to determine whether we allow the fixed point and scientific notation respectively.
 The default is  `fast_float::chars_format::general` which allows both `fixed` and `scientific`.
 
-The library seeks to follow the C++17 (see [20.19.3](http://eel.is/c++draft/charconv.from.chars).(7.1))  specification. 
+The library seeks to follow the C++17 (see [20.19.3](http://eel.is/c++draft/charconv.from.chars).(7.1))  specification.
 * The `from_chars` function does not skip leading white-space characters.
 * [A leading `+` sign](https://en.cppreference.com/w/cpp/utility/from_chars) is forbidden.
-* It is generally impossible to represent a decimal value exactly as binary floating-point number (`float` and `double` types). We seek the nearest value. We round to an even mantissa when we are in-between two binary floating-point numbers. 
+* It is generally impossible to represent a decimal value exactly as binary floating-point number (`float` and `double` types). We seek the nearest value. We round to an even mantissa when we are in-between two binary floating-point numbers.
 
 Furthermore, we have the following restrictions:
 * We only support `float` and `double` types at this time.
@@ -77,22 +70,22 @@ Furthermore, we have the following restrictions:
 
 We support Visual Studio, macOS, Linux, freeBSD. We support big and little endian. We support 32-bit and 64-bit systems.
 
-
+We assume that the rounding mode is set to nearest (`std::fegetround() == FE_TONEAREST`).
 
 ## Using commas as decimal separator
 
 
 The C++ standard stipulate that `from_chars` has to be locale-independent. In
-particular, the decimal separator has to be the period (`.`). However, 
+particular, the decimal separator has to be the period (`.`). However,
-some users still want to use the `fast_float` library with in a locale-dependent 
+some users still want to use the `fast_float` library with in a locale-dependent
 manner. Using a separate function called `from_chars_advanced`, we allow the users
-to pass a `parse_options` instance which contains a custom decimal separator (e.g., 
+to pass a `parse_options` instance which contains a custom decimal separator (e.g.,
 the comma). You may use it as follows.
 
 ```C++
 #include "fast_float/fast_float.h"
 #include <iostream>
- 
+
 int main() {
     const std::string input =  "3,1416 xyz ";
     double result;
@@ -104,23 +97,62 @@ int main() {
 }
 ```
 
+You can parse delimited numbers:
+```C++
+  const std::string input =   "234532.3426362,7869234.9823,324562.645";
+  double result;
+  auto answer = fast_float::from_chars(input.data(), input.data()+input.size(), result);
+  if(answer.ec != std::errc()) {
+    // check error
+  }
+  // we have result == 234532.3426362.
+  if(answer.ptr[0] != ',') {
+    // unexpected delimiter
+  }
+  answer = fast_float::from_chars(answer.ptr + 1, input.data()+input.size(), result);
+  if(answer.ec != std::errc()) {
+    // check error
+  }
+  // we have result == 7869234.9823.
+  if(answer.ptr[0] != ',') {
+    // unexpected delimiter
+  }
+  answer = fast_float::from_chars(answer.ptr + 1, input.data()+input.size(), result);
+  if(answer.ec != std::errc()) {
+    // check error
+  }
+  // we have result == 324562.645.
+```
 
-## Reference
 
-- Daniel Lemire, [Number Parsing at a Gigabyte per Second](https://arxiv.org/abs/2101.11408), Software: Pratice and Experience 51 (8), 2021.
+## Relation With Other Work
+
+The fast_float library is part of:
+
+- GCC (as of version 12): the `from_chars` function in GCC relies on fast_float.
+- [WebKit](https://github.com/WebKit/WebKit), the engine behind Safari (Apple's web browser)
+
+
+The fastfloat algorithm is part of the [LLVM standard libraries](https://github.com/llvm/llvm-project/commit/87c016078ad72c46505461e4ff8bfa04819fe7ba).
+
+There is a [derived implementation part of AdaCore](https://github.com/AdaCore/VSS).
+
+
+The fast_float library provides a performance similar to that of the [fast_double_parser](https://github.com/lemire/fast_double_parser) library but using an updated algorithm reworked from the ground up, and while offering an API more in line with the expectations of C++ programmers. The fast_double_parser library is part of the [Microsoft LightGBM machine-learning framework](https://github.com/microsoft/LightGBM).
+
+## References
+
+- Daniel Lemire, [Number Parsing at a Gigabyte per Second](https://arxiv.org/abs/2101.11408), Software: Practice and Experience 51 (8), 2021.
+- Noble Mushtak, Daniel Lemire, [Fast Number Parsing Without Fallback](https://arxiv.org/abs/2212.06644), Software: Practice and Experience (to appear)
 
 ## Other programming languages
 
 - [There is an R binding](https://github.com/eddelbuettel/rcppfastfloat) called `rcppfastfloat`.
 - [There is a Rust port of the fast_float library](https://github.com/aldanor/fast-float-rust/) called `fast-float-rust`.
-- [There is a Java port of the fast_float library](https://github.com/wrandelshofer/FastDoubleParser) called `FastDoubleParser`.
+- [There is a Java port of the fast_float library](https://github.com/wrandelshofer/FastDoubleParser) called `FastDoubleParser`. It used for important systems such as [Jackson](https://github.com/FasterXML/jackson-core).
 - [There is a C# port of the fast_float library](https://github.com/CarlVerret/csFastFloat) called `csFastFloat`.
 
 
-## Relation With Other Work
-
-The fast_float library provides a performance similar to that of the [fast_double_parser](https://github.com/lemire/fast_double_parser) library but using an updated algorithm reworked from the ground up, and while offering an API more in line with the expectations of C++ programmers. The fast_double_parser library is part of the [Microsoft LightGBM machine-learning framework](https://github.com/microsoft/LightGBM).
-
 ## Users
 
 The fast_float library is used by [Apache Arrow](https://github.com/apache/arrow/pull/8494) where it multiplied the number parsing speed by two or three times. It is also used by [Yandex ClickHouse](https://github.com/ClickHouse/ClickHouse) and by [Google Jsonnet](https://github.com/google/jsonnet).
@@ -133,14 +165,14 @@ It can parse random floating-point numbers at a speed of 1 GB/s on some systems.
 <img src="http://lemire.me/blog/wp-content/uploads/2020/11/fastfloat_speed.png" width="400">
 
 ```
-$ ./build/benchmarks/benchmark 
+$ ./build/benchmarks/benchmark
 # parsing random integers in the range [0,1)
-volume = 2.09808 MB 
+volume = 2.09808 MB
-netlib                                  :   271.18 MB/s (+/- 1.2 %)    12.93 Mfloat/s  
+netlib                                  :   271.18 MB/s (+/- 1.2 %)    12.93 Mfloat/s 
-doubleconversion                        :   225.35 MB/s (+/- 1.2 %)    10.74 Mfloat/s  
+doubleconversion                        :   225.35 MB/s (+/- 1.2 %)    10.74 Mfloat/s 
-strtod                                  :   190.94 MB/s (+/- 1.6 %)     9.10 Mfloat/s  
+strtod                                  :   190.94 MB/s (+/- 1.6 %)     9.10 Mfloat/s 
-abseil                                  :   430.45 MB/s (+/- 2.2 %)    20.52 Mfloat/s  
+abseil                                  :   430.45 MB/s (+/- 2.2 %)    20.52 Mfloat/s 
-fastfloat                               :  1042.38 MB/s (+/- 9.9 %)    49.68 Mfloat/s  
+fastfloat                               :  1042.38 MB/s (+/- 9.9 %)    49.68 Mfloat/s 
 ```
 
 See https://github.com/lemire/simple_fastfloat_benchmark for our benchmarking code.
@@ -181,23 +213,23 @@ You should change the `GIT_TAG` line so that you recover the version you wish to
 
 ## Using as single header
 
-The script `script/amalgamate.py` may be used to generate a single header 
+The script `script/amalgamate.py` may be used to generate a single header
 version of the library if so desired.
-Just run the script from the root directory of this repository. 
+Just run the script from the root directory of this repository.
 You can customize the license type and output file if desired as described in
 the command line help.
 
 You may directly download automatically generated single-header files:
 
-https://github.com/fastfloat/fast_float/releases/download/v1.1.2/fast_float.h
+https://github.com/fastfloat/fast_float/releases/download/v3.4.0/fast_float.h
 
 ## Credit
 
-Though this work is inspired by many different people, this work benefited especially from exchanges with 
+Though this work is inspired by many different people, this work benefited especially from exchanges with
-Michael Eisel, who motivated the original research with his key insights, and with Nigel Tao who provided 
+Michael Eisel, who motivated the original research with his key insights, and with Nigel Tao who provided
 invaluable feedback. Rémy Oudompheng first implemented a fast path we use in the case of long digits.
 
-The library includes code adapted from Google Wuffs (written by Nigel Tao) which was originally published 
+The library includes code adapted from Google Wuffs (written by Nigel Tao) which was originally published
 under the Apache 2.0 license.
 
 ## License

include/fast_float/ascii_number.h

@@ -12,9 +12,11 @@ namespace fast_float {
 
 // Next function can be micro-optimized, but compilers are entirely
 // able to optimize it well.
-fastfloat_really_inline bool is_integer(char c)  noexcept  { return c >= '0' && c <= '9'; }
+fastfloat_really_inline constexpr bool is_integer(char c) noexcept {
+  return c >= '0' && c <= '9';
+}
 
-fastfloat_really_inline uint64_t byteswap(uint64_t val) {
+fastfloat_really_inline constexpr uint64_t byteswap(uint64_t val) {
   return (val & 0xFF00000000000000) >> 56
     | (val & 0x00FF000000000000) >> 40
     | (val & 0x0000FF0000000000) >> 24
@@ -44,7 +46,8 @@ fastfloat_really_inline void write_u64(uint8_t *chars, uint64_t val) {
 }
 
 // credit  @aqrit
-fastfloat_really_inline uint32_t  parse_eight_digits_unrolled(uint64_t val) {
+fastfloat_really_inline FASTFLOAT_CONSTEXPR14
+uint32_t parse_eight_digits_unrolled(uint64_t val) {
   const uint64_t mask = 0x000000FF000000FF;
   const uint64_t mul1 = 0x000F424000000064; // 100 + (1000000ULL << 32)
   const uint64_t mul2 = 0x0000271000000001; // 1 + (10000ULL << 32)
@@ -59,7 +62,7 @@ fastfloat_really_inline uint32_t parse_eight_digits_unrolled(const char *chars)
 }
 
 // credit @aqrit
-fastfloat_really_inline bool is_made_of_eight_digits_fast(uint64_t val)  noexcept  {
+fastfloat_really_inline constexpr bool is_made_of_eight_digits_fast(uint64_t val)  noexcept  {
   return !((((val + 0x4646464646464646) | (val - 0x3030303030303030)) &
      0x8080808080808080));
 }
@@ -93,7 +96,11 @@ parsed_number_string parse_number_string(const char *p, const char *pend, parse_
   answer.valid = false;
   answer.too_many_digits = false;
   answer.negative = (*p == '-');
+#if FASTFLOAT_ALLOWS_LEADING_PLUS // disabled by default
+  if ((*p == '-') || (*p == '+')) {
+#else
   if (*p == '-') { // C++17 20.19.3.(7.1) explicitly forbids '+' sign here
+#endif
     ++p;
     if (p == pend) {
       return answer;
@@ -106,10 +113,6 @@ parsed_number_string parse_number_string(const char *p, const char *pend, parse_
 
   uint64_t i = 0; // an unsigned int avoids signed overflows (which are bad)
 
-  while ((std::distance(p, pend) >= 8) && is_made_of_eight_digits_fast(p)) {
-    i = i * 100000000 + parse_eight_digits_unrolled(p); // in rare cases, this will overflow, but that's ok
-    p += 8;
-  }
   while ((p != pend) && is_integer(*p)) {
     // a multiplication by 10 is cheaper than an arbitrary integer
     // multiplication

include/fast_float/bigint.h

@@ -17,7 +17,7 @@ namespace fast_float {
 // we might have platforms where `CHAR_BIT` is not 8, so let's avoid
 // doing `8 * sizeof(limb)`.
 #if defined(FASTFLOAT_64BIT) && !defined(__sparc)
-#define FASTFLOAT_64BIT_LIMB
+#define FASTFLOAT_64BIT_LIMB 1
 typedef uint64_t limb;
 constexpr size_t limb_bits = 64;
 #else
@@ -54,23 +54,23 @@ struct stackvec {
     FASTFLOAT_ASSERT(try_extend(s));
   }
 
-  limb& operator[](size_t index) noexcept {
+  FASTFLOAT_CONSTEXPR14 limb& operator[](size_t index) noexcept {
     FASTFLOAT_DEBUG_ASSERT(index < length);
     return data[index];
   }
-  const limb& operator[](size_t index) const noexcept {
+  FASTFLOAT_CONSTEXPR14 const limb& operator[](size_t index) const noexcept {
     FASTFLOAT_DEBUG_ASSERT(index < length);
     return data[index];
   }
   // index from the end of the container
-  const limb& rindex(size_t index) const noexcept {
+  FASTFLOAT_CONSTEXPR14 const limb& rindex(size_t index) const noexcept {
     FASTFLOAT_DEBUG_ASSERT(index < length);
     size_t rindex = length - index - 1;
     return data[rindex];
   }
 
   // set the length, without bounds checking.
-  void set_len(size_t len) noexcept {
+  FASTFLOAT_CONSTEXPR14 void set_len(size_t len) noexcept {
     length = uint16_t(len);
   }
   constexpr size_t len() const noexcept {
@@ -83,12 +83,12 @@ struct stackvec {
     return size;
   }
   // append item to vector, without bounds checking
-  void push_unchecked(limb value) noexcept {
+  FASTFLOAT_CONSTEXPR14 void push_unchecked(limb value) noexcept {
     data[length] = value;
     length++;
   }
   // append item to vector, returning if item was added
-  bool try_push(limb value) noexcept {
+  FASTFLOAT_CONSTEXPR14 bool try_push(limb value) noexcept {
     if (len() < capacity()) {
       push_unchecked(value);
       return true;
@@ -137,7 +137,7 @@ struct stackvec {
   // check if any limbs are non-zero after the given index.
   // this needs to be done in reverse order, since the index
   // is relative to the most significant limbs.
-  bool nonzero(size_t index) const noexcept {
+  FASTFLOAT_CONSTEXPR14 bool nonzero(size_t index) const noexcept {
     while (index < len()) {
       if (rindex(index) != 0) {
         return true;
@@ -147,14 +147,14 @@ struct stackvec {
     return false;
   }
   // normalize the big integer, so most-significant zero limbs are removed.
-  void normalize() noexcept {
+  FASTFLOAT_CONSTEXPR14 void normalize() noexcept {
     while (len() > 0 && rindex(0) == 0) {
       length--;
     }
   }
 };
 
-fastfloat_really_inline
+fastfloat_really_inline FASTFLOAT_CONSTEXPR14
 uint64_t empty_hi64(bool& truncated) noexcept {
   truncated = false;
   return 0;

include/fast_float/decimal_to_binary.h

@@ -63,7 +63,7 @@ namespace detail {
 // create an adjusted mantissa, biased by the invalid power2
 // for significant digits already multiplied by 10 ** q.
 template <typename binary>
-fastfloat_really_inline
+fastfloat_really_inline FASTFLOAT_CONSTEXPR14
 adjusted_mantissa compute_error_scaled(int64_t q, uint64_t w, int lz) noexcept  {
   int hilz = int(w >> 63) ^ 1;
   adjusted_mantissa answer;

include/fast_float/digit_comparison.h

@@ -23,7 +23,8 @@ constexpr static uint64_t powers_of_ten_uint64[] = {
 // this algorithm is not even close to optimized, but it has no practical
 // effect on performance: in order to have a faster algorithm, we'd need
 // to slow down performance for faster algorithms, and this is still fast.
-fastfloat_really_inline int32_t scientific_exponent(parsed_number_string& num) noexcept {
+fastfloat_really_inline FASTFLOAT_CONSTEXPR14
+int32_t scientific_exponent(parsed_number_string& num) noexcept {
   uint64_t mantissa = num.mantissa;
   int32_t exponent = int32_t(num.exponent);
   while (mantissa >= 10000) {
@@ -44,40 +45,24 @@ fastfloat_really_inline int32_t scientific_exponent(parsed_number_string& num) n
 // this converts a native floating-point number to an extended-precision float.
 template <typename T>
 fastfloat_really_inline adjusted_mantissa to_extended(T value) noexcept {
+  using equiv_uint = typename binary_format<T>::equiv_uint;
+  constexpr equiv_uint exponent_mask = binary_format<T>::exponent_mask();
+  constexpr equiv_uint mantissa_mask = binary_format<T>::mantissa_mask();
+  constexpr equiv_uint hidden_bit_mask = binary_format<T>::hidden_bit_mask();
+
   adjusted_mantissa am;
   int32_t bias = binary_format<T>::mantissa_explicit_bits() - binary_format<T>::minimum_exponent();
-  if (std::is_same<T, float>::value) {
+  equiv_uint bits;
-    constexpr uint32_t exponent_mask = 0x7F800000;
+  ::memcpy(&bits, &value, sizeof(T));
-    constexpr uint32_t mantissa_mask = 0x007FFFFF;
+  if ((bits & exponent_mask) == 0) {
-    constexpr uint64_t hidden_bit_mask = 0x00800000;
+    // denormal
-    uint32_t bits;
+    am.power2 = 1 - bias;
-    ::memcpy(&bits, &value, sizeof(T));
+    am.mantissa = bits & mantissa_mask;
-    if ((bits & exponent_mask) == 0) {
-      // denormal
-      am.power2 = 1 - bias;
-      am.mantissa = bits & mantissa_mask;
-    } else {
-      // normal
-      am.power2 = int32_t((bits & exponent_mask) >> binary_format<T>::mantissa_explicit_bits());
-      am.power2 -= bias;
-      am.mantissa = (bits & mantissa_mask) | hidden_bit_mask;
-    }
   } else {
-    constexpr uint64_t exponent_mask = 0x7FF0000000000000;
+    // normal
-    constexpr uint64_t mantissa_mask = 0x000FFFFFFFFFFFFF;
+    am.power2 = int32_t((bits & exponent_mask) >> binary_format<T>::mantissa_explicit_bits());
-    constexpr uint64_t hidden_bit_mask = 0x0010000000000000;
+    am.power2 -= bias;
-    uint64_t bits;
+    am.mantissa = (bits & mantissa_mask) | hidden_bit_mask;
-    ::memcpy(&bits, &value, sizeof(T));
-    if ((bits & exponent_mask) == 0) {
-      // denormal
-      am.power2 = 1 - bias;
-      am.mantissa = bits & mantissa_mask;
-    } else {
-      // normal
-      am.power2 = int32_t((bits & exponent_mask) >> binary_format<T>::mantissa_explicit_bits());
-      am.power2 -= bias;
-      am.mantissa = (bits & mantissa_mask) | hidden_bit_mask;
-    }
   }
 
   return am;
@@ -97,12 +82,13 @@ fastfloat_really_inline adjusted_mantissa to_extended_halfway(T value) noexcept
 
 // round an extended-precision float to the nearest machine float.
 template <typename T, typename callback>
-fastfloat_really_inline void round(adjusted_mantissa& am, callback cb) noexcept {
+fastfloat_really_inline FASTFLOAT_CONSTEXPR14
+void round(adjusted_mantissa& am, callback cb) noexcept {
   int32_t mantissa_shift = 64 - binary_format<T>::mantissa_explicit_bits() - 1;
   if (-am.power2 >= mantissa_shift) {
     // have a denormal float
     int32_t shift = -am.power2 + 1;
-    cb(am, std::min(shift, 64));
+    cb(am, std::min<int32_t>(shift, 64));
     // check for round-up: if rounding-nearest carried us to the hidden bit.
     am.power2 = (am.mantissa < (uint64_t(1) << binary_format<T>::mantissa_explicit_bits())) ? 0 : 1;
     return;
@@ -126,23 +112,19 @@ fastfloat_really_inline void round(adjusted_mantissa& am, callback cb) noexcept
 }
 
 template <typename callback>
-fastfloat_really_inline
+fastfloat_really_inline FASTFLOAT_CONSTEXPR14
 void round_nearest_tie_even(adjusted_mantissa& am, int32_t shift, callback cb) noexcept {
-  uint64_t mask;
+  const uint64_t mask
-  uint64_t halfway;
+  = (shift == 64)
-  if (shift == 64) {
+    ? UINT64_MAX
-    mask = UINT64_MAX;
+    : (uint64_t(1) << shift) - 1;
-  } else {
+  const uint64_t halfway
-    mask = (uint64_t(1) << shift) - 1;
+  = (shift == 0)
-  }
+    ? 0
-  if (shift == 0) {
+    : uint64_t(1) << (shift - 1);
-    halfway = 0;
-  } else {
-    halfway = uint64_t(1) << (shift - 1);
-  }
   uint64_t truncated_bits = am.mantissa & mask;
-  uint64_t is_above = truncated_bits > halfway;
+  bool is_above = truncated_bits > halfway;
-  uint64_t is_halfway = truncated_bits == halfway;
+  bool is_halfway = truncated_bits == halfway;
 
   // shift digits into position
   if (shift == 64) {
@@ -156,7 +138,8 @@ void round_nearest_tie_even(adjusted_mantissa& am, int32_t shift, callback cb) n
   am.mantissa += uint64_t(cb(is_odd, is_halfway, is_above));
 }
 
-fastfloat_really_inline void round_down(adjusted_mantissa& am, int32_t shift) noexcept {
+fastfloat_really_inline FASTFLOAT_CONSTEXPR14
+void round_down(adjusted_mantissa& am, int32_t shift) noexcept {
   if (shift == 64) {
     am.mantissa = 0;
   } else {
@@ -215,7 +198,7 @@ void parse_eight_digits(const char*& p, limb& value, size_t& counter, size_t& co
   count += 8;
 }
 
-fastfloat_really_inline
+fastfloat_really_inline FASTFLOAT_CONSTEXPR14
 void parse_one_digit(const char*& p, limb& value, size_t& counter, size_t& count) noexcept {
   value = value * 10 + limb(*p - '0');
   p++;

include/fast_float/fast_float.h

@@ -44,7 +44,7 @@ struct parse_options {
  * Like the C++17 standard, the `fast_float::from_chars` functions take an optional last argument of
  * the type `fast_float::chars_format`. It is a bitset value: we check whether
  * `fmt & fast_float::chars_format::fixed` and `fmt & fast_float::chars_format::scientific` are set
- * to determine whether we allowe the fixed point and scientific notation respectively.
+ * to determine whether we allow the fixed point and scientific notation respectively.
  * The default is  `fast_float::chars_format::general` which allows both `fixed` and `scientific`.
  */
 template<typename T>
@@ -58,6 +58,6 @@ template<typename T>
 from_chars_result from_chars_advanced(const char *first, const char *last,
                                       T &value, parse_options options)  noexcept;
 
-}
+} // namespace fast_float
 #include "parse_number.h"
 #endif // FASTFLOAT_FAST_FLOAT_H

include/fast_float/float_common.h

@@ -5,18 +5,18 @@
 #include <cstdint>
 #include <cassert>
 #include <cstring>
+#include <type_traits>
 
 #if (defined(__x86_64) || defined(__x86_64__) || defined(_M_X64)   \
        || defined(__amd64) || defined(__aarch64__) || defined(_M_ARM64) \
        || defined(__MINGW64__)                                          \
        || defined(__s390x__)                                            \
-       || (defined(__ppc64__) || defined(__PPC64__) || defined(__ppc64le__) || defined(__PPC64LE__)) \
+       || (defined(__ppc64__) || defined(__PPC64__) || defined(__ppc64le__) || defined(__PPC64LE__)) )
-       || defined(__EMSCRIPTEN__))
+#define FASTFLOAT_64BIT 1
-#define FASTFLOAT_64BIT
 #elif (defined(__i386) || defined(__i386__) || defined(_M_IX86)   \
-     || defined(__arm__) || defined(_M_ARM)                   \
+     || defined(__arm__) || defined(_M_ARM) || defined(__ppc__)   \
-     || defined(__MINGW32__))
+     || defined(__MINGW32__) || defined(__EMSCRIPTEN__))
-#define FASTFLOAT_32BIT
+#define FASTFLOAT_32BIT 1
 #else
   // Need to check incrementally, since SIZE_MAX is a size_t, avoid overflow.
   // We can never tell the register width, but the SIZE_MAX is a good approximation.
@@ -24,9 +24,9 @@
   #if SIZE_MAX == 0xffff
     #error Unknown platform (16-bit, unsupported)
   #elif SIZE_MAX == 0xffffffff
-    #define FASTFLOAT_32BIT
+    #define FASTFLOAT_32BIT 1
   #elif SIZE_MAX == 0xffffffffffffffff
-    #define FASTFLOAT_64BIT
+    #define FASTFLOAT_64BIT 1
   #else
     #error Unknown platform (not 32-bit, not 64-bit?)
   #endif
@@ -40,7 +40,9 @@
 #define FASTFLOAT_VISUAL_STUDIO 1
 #endif
 
-#ifdef _WIN32
+#if defined __BYTE_ORDER__ && defined __ORDER_BIG_ENDIAN__
+#define FASTFLOAT_IS_BIG_ENDIAN (__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)
+#elif defined _WIN32
 #define FASTFLOAT_IS_BIG_ENDIAN 0
 #else
 #if defined(__APPLE__) || defined(__FreeBSD__)
@@ -48,7 +50,11 @@
 #elif defined(sun) || defined(__sun)
 #include <sys/byteorder.h>
 #else
+#ifdef __has_include
+#if __has_include(<endian.h>)
 #include <endian.h>
+#endif //__has_include(<endian.h>)
+#endif //__has_include
 #endif
 #
 #ifndef __BYTE_ORDER__
@@ -75,22 +81,28 @@
 #endif
 
 #ifndef FASTFLOAT_ASSERT
-#define FASTFLOAT_ASSERT(x)  { if (!(x)) abort(); }
+#define FASTFLOAT_ASSERT(x)  { ((void)(x)); }
 #endif
 
 #ifndef FASTFLOAT_DEBUG_ASSERT
-#include <cassert>
+#define FASTFLOAT_DEBUG_ASSERT(x) { ((void)(x)); }
-#define FASTFLOAT_DEBUG_ASSERT(x) assert(x)
 #endif
 
 // rust style `try!()` macro, or `?` operator
 #define FASTFLOAT_TRY(x) { if (!(x)) return false; }
 
+// Testing for https://wg21.link/N3652, adopted in C++14
+#if __cpp_constexpr >= 201304
+#define FASTFLOAT_CONSTEXPR14 constexpr
+#else
+#define FASTFLOAT_CONSTEXPR14
+#endif
+
 namespace fast_float {
 
 // Compares two ASCII strings in a case insensitive manner.
-inline bool fastfloat_strncasecmp(const char *input1, const char *input2,
+inline FASTFLOAT_CONSTEXPR14 bool
-                                  size_t length) {
+fastfloat_strncasecmp(const char *input1, const char *input2, size_t length) {
   char running_diff{0};
   for (size_t i = 0; i < length; i++) {
     running_diff |= (input1[i] ^ input2[i]);
@@ -107,14 +119,14 @@ template <typename T>
 struct span {
   const T* ptr;
   size_t length;
-  span(const T* _ptr, size_t _length) : ptr(_ptr), length(_length) {}
+  constexpr span(const T* _ptr, size_t _length) : ptr(_ptr), length(_length) {}
-  span() : ptr(nullptr), length(0) {}
+  constexpr span() : ptr(nullptr), length(0) {}
 
   constexpr size_t len() const noexcept {
     return length;
   }
 
-  const T& operator[](size_t index) const noexcept {
+  FASTFLOAT_CONSTEXPR14 const T& operator[](size_t index) const noexcept {
     FASTFLOAT_DEBUG_ASSERT(index < length);
     return ptr[index];
   }
@@ -123,8 +135,8 @@ struct span {
 struct value128 {
   uint64_t low;
   uint64_t high;
-  value128(uint64_t _low, uint64_t _high) : low(_low), high(_high) {}
+  constexpr value128(uint64_t _low, uint64_t _high) : low(_low), high(_high) {}
-  value128() : low(0), high(0) {}
+  constexpr value128() : low(0), high(0) {}
 };
 
 /* result might be undefined when input_num is zero */
@@ -155,14 +167,14 @@ fastfloat_really_inline int leading_zeroes(uint64_t input_num) {
 #ifdef FASTFLOAT_32BIT
 
 // slow emulation routine for 32-bit
-fastfloat_really_inline uint64_t emulu(uint32_t x, uint32_t y) {
+fastfloat_really_inline constexpr uint64_t emulu(uint32_t x, uint32_t y) {
     return x * (uint64_t)y;
 }
 
 // slow emulation routine for 32-bit
 #if !defined(__MINGW64__)
-fastfloat_really_inline uint64_t _umul128(uint64_t ab, uint64_t cd,
+fastfloat_really_inline constexpr uint64_t _umul128(uint64_t ab, uint64_t cd,
-                                          uint64_t *hi) {
+                                                    uint64_t *hi) {
   uint64_t ad = emulu((uint32_t)(ab >> 32), (uint32_t)cd);
   uint64_t bd = emulu((uint32_t)ab, (uint32_t)cd);
   uint64_t adbc = ad + emulu((uint32_t)ab, (uint32_t)(cd >> 32));
@@ -181,8 +193,9 @@ fastfloat_really_inline uint64_t _umul128(uint64_t ab, uint64_t cd,
 fastfloat_really_inline value128 full_multiplication(uint64_t a,
                                                      uint64_t b) {
   value128 answer;
-#ifdef _M_ARM64
+#if defined(_M_ARM64) && !defined(__MINGW32__)
   // ARM64 has native support for 64-bit multiplications, no need to emulate
+  // But MinGW on ARM64 doesn't have native support for 64-bit multiplications
   answer.high = __umulh(a, b);
   answer.low = a * b;
 #elif defined(FASTFLOAT_32BIT) || (defined(_WIN64) && !defined(__clang__))
@@ -201,10 +214,10 @@ struct adjusted_mantissa {
   uint64_t mantissa{0};
   int32_t power2{0}; // a negative value indicates an invalid result
   adjusted_mantissa() = default;
-  bool operator==(const adjusted_mantissa &o) const {
+  constexpr bool operator==(const adjusted_mantissa &o) const {
     return mantissa == o.mantissa && power2 == o.power2;
   }
-  bool operator!=(const adjusted_mantissa &o) const {
+  constexpr bool operator!=(const adjusted_mantissa &o) const {
     return mantissa != o.mantissa || power2 != o.power2;
   }
 };
@@ -215,25 +228,91 @@ constexpr static int32_t invalid_am_bias = -0x8000;
 constexpr static double powers_of_ten_double[] = {
     1e0,  1e1,  1e2,  1e3,  1e4,  1e5,  1e6,  1e7,  1e8,  1e9,  1e10, 1e11,
     1e12, 1e13, 1e14, 1e15, 1e16, 1e17, 1e18, 1e19, 1e20, 1e21, 1e22};
-constexpr static float powers_of_ten_float[] = {1e0, 1e1, 1e2, 1e3, 1e4, 1e5,
+constexpr static float powers_of_ten_float[] = {1e0f, 1e1f, 1e2f, 1e3f, 1e4f, 1e5f,
-                                                1e6, 1e7, 1e8, 1e9, 1e10};
+                                                1e6f, 1e7f, 1e8f, 1e9f, 1e10f};
+// used for max_mantissa_double and max_mantissa_float
+constexpr uint64_t constant_55555 = 5 * 5 * 5 * 5 * 5;
+// Largest integer value v so that (5**index * v) <= 1<<53.
+// 0x10000000000000 == 1 << 53
+constexpr static uint64_t max_mantissa_double[] = {
+      0x10000000000000,
+      0x10000000000000 / 5,
+      0x10000000000000 / (5 * 5),
+      0x10000000000000 / (5 * 5 * 5),
+      0x10000000000000 / (5 * 5 * 5 * 5),
+      0x10000000000000 / (constant_55555),
+      0x10000000000000 / (constant_55555 * 5),
+      0x10000000000000 / (constant_55555 * 5 * 5),
+      0x10000000000000 / (constant_55555 * 5 * 5 * 5),
+      0x10000000000000 / (constant_55555 * 5 * 5 * 5 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555),
+      0x10000000000000 / (constant_55555 * constant_55555 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * 5 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * 5 * 5 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * 5 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * 5 * 5 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * 5 * 5 * 5 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * constant_55555),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * constant_55555 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * constant_55555 * 5 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * constant_55555 * 5 * 5 * 5),
+      0x10000000000000 / (constant_55555 * constant_55555 * constant_55555 * constant_55555 * 5 * 5 * 5 * 5)};
+  // Largest integer value v so that (5**index * v) <= 1<<24.
+  // 0x1000000 == 1<<24
+  constexpr static uint64_t max_mantissa_float[] = {
+      0x1000000,
+      0x1000000 / 5,
+      0x1000000 / (5 * 5),
+      0x1000000 / (5 * 5 * 5),
+      0x1000000 / (5 * 5 * 5 * 5),
+      0x1000000 / (constant_55555),
+      0x1000000 / (constant_55555 * 5),
+      0x1000000 / (constant_55555 * 5 * 5),
+      0x1000000 / (constant_55555 * 5 * 5 * 5),
+      0x1000000 / (constant_55555 * 5 * 5 * 5 * 5),
+      0x1000000 / (constant_55555 * constant_55555),
+      0x1000000 / (constant_55555 * constant_55555 * 5)};
 
 template <typename T> struct binary_format {
+  using equiv_uint = typename std::conditional<sizeof(T) == 4, uint32_t, uint64_t>::type;
+
   static inline constexpr int mantissa_explicit_bits();
   static inline constexpr int minimum_exponent();
   static inline constexpr int infinite_power();
   static inline constexpr int sign_index();
-  static inline constexpr int min_exponent_fast_path();
+  static inline constexpr int min_exponent_fast_path(); // used when fegetround() == FE_TONEAREST
   static inline constexpr int max_exponent_fast_path();
   static inline constexpr int max_exponent_round_to_even();
   static inline constexpr int min_exponent_round_to_even();
-  static inline constexpr uint64_t max_mantissa_fast_path();
+  static inline constexpr uint64_t max_mantissa_fast_path(int64_t power);
+  static inline constexpr uint64_t max_mantissa_fast_path(); // used when fegetround() == FE_TONEAREST
   static inline constexpr int largest_power_of_ten();
   static inline constexpr int smallest_power_of_ten();
   static inline constexpr T exact_power_of_ten(int64_t power);
   static inline constexpr size_t max_digits();
+  static inline constexpr equiv_uint exponent_mask();
+  static inline constexpr equiv_uint mantissa_mask();
+  static inline constexpr equiv_uint hidden_bit_mask();
 };
 
+template <> inline constexpr int binary_format<double>::min_exponent_fast_path() {
+#if (FLT_EVAL_METHOD != 1) && (FLT_EVAL_METHOD != 0)
+  return 0;
+#else
+  return -22;
+#endif
+}
+
+template <> inline constexpr int binary_format<float>::min_exponent_fast_path() {
+#if (FLT_EVAL_METHOD != 1) && (FLT_EVAL_METHOD != 0)
+  return 0;
+#else
+  return -10;
+#endif
+}
+
 template <> inline constexpr int binary_format<double>::mantissa_explicit_bits() {
   return 52;
 }
@@ -274,34 +353,30 @@ template <> inline constexpr int binary_format<float>::infinite_power() {
 template <> inline constexpr int binary_format<double>::sign_index() { return 63; }
 template <> inline constexpr int binary_format<float>::sign_index() { return 31; }
 
-template <> inline constexpr int binary_format<double>::min_exponent_fast_path() {
-#if (FLT_EVAL_METHOD != 1) && (FLT_EVAL_METHOD != 0)
-  return 0;
-#else
-  return -22;
-#endif
-}
-template <> inline constexpr int binary_format<float>::min_exponent_fast_path() {
-#if (FLT_EVAL_METHOD != 1) && (FLT_EVAL_METHOD != 0)
-  return 0;
-#else
-  return -10;
-#endif
-}
-
 template <> inline constexpr int binary_format<double>::max_exponent_fast_path() {
   return 22;
 }
 template <> inline constexpr int binary_format<float>::max_exponent_fast_path() {
   return 10;
 }
-
 template <> inline constexpr uint64_t binary_format<double>::max_mantissa_fast_path() {
   return uint64_t(2) << mantissa_explicit_bits();
 }
+template <> inline constexpr uint64_t binary_format<double>::max_mantissa_fast_path(int64_t power) {
+  // caller is responsible to ensure that
+  // power >= 0 && power <= 22
+  //
+  return max_mantissa_double[power];
+}
 template <> inline constexpr uint64_t binary_format<float>::max_mantissa_fast_path() {
   return uint64_t(2) << mantissa_explicit_bits();
 }
+template <> inline constexpr uint64_t binary_format<float>::max_mantissa_fast_path(int64_t power) {
+  // caller is responsible to ensure that
+  // power >= 0 && power <= 10
+  //
+  return max_mantissa_float[power];
+}
 
 template <>
 inline constexpr double binary_format<double>::exact_power_of_ten(int64_t power) {
@@ -339,6 +414,33 @@ template <> inline constexpr size_t binary_format<float>::max_digits() {
   return 114;
 }
 
+template <> inline constexpr binary_format<float>::equiv_uint
+    binary_format<float>::exponent_mask() {
+  return 0x7F800000;
+}
+template <> inline constexpr binary_format<double>::equiv_uint
+    binary_format<double>::exponent_mask() {
+  return 0x7FF0000000000000;
+}
+
+template <> inline constexpr binary_format<float>::equiv_uint
+    binary_format<float>::mantissa_mask() {
+  return 0x007FFFFF;
+}
+template <> inline constexpr binary_format<double>::equiv_uint
+    binary_format<double>::mantissa_mask() {
+  return 0x000FFFFFFFFFFFFF;
+}
+
+template <> inline constexpr binary_format<float>::equiv_uint
+    binary_format<float>::hidden_bit_mask() {
+  return 0x00800000;
+}
+template <> inline constexpr binary_format<double>::equiv_uint
+    binary_format<double>::hidden_bit_mask() {
+  return 0x0010000000000000;
+}
+
 template<typename T>
 fastfloat_really_inline void to_float(bool negative, adjusted_mantissa am, T &value) {
   uint64_t word = am.mantissa;
@@ -357,6 +459,28 @@ fastfloat_really_inline void to_float(bool negative, adjusted_mantissa am, T &va
 #endif
 }
 
+#if FASTFLOAT_SKIP_WHITE_SPACE // disabled by default
+template <typename = void>
+struct space_lut {
+  static constexpr bool value[] = {
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+};
+
+template <typename T>
+constexpr bool space_lut<T>::value[];
+
+inline constexpr bool is_space(uint8_t c) { return space_lut<>::value[c]; }
+#endif
 } // namespace fast_float
 
 #endif

include/fast_float/parse_number.h

@@ -60,6 +60,70 @@ from_chars_result parse_infnan(const char *first, const char *last, T &value)  n
   return answer;
 }
 
+/**
+ * Returns true if the floating-pointing rounding mode is to 'nearest'.
+ * It is the default on most system. This function is meant to be inexpensive.
+ * Credit : @mwalcott3
+ */
+fastfloat_really_inline bool rounds_to_nearest() noexcept {
+  // https://lemire.me/blog/2020/06/26/gcc-not-nearest/
+#if (FLT_EVAL_METHOD != 1) && (FLT_EVAL_METHOD != 0)
+  return false;
+#endif
+  // See
+  // A fast function to check your floating-point rounding mode
+  // https://lemire.me/blog/2022/11/16/a-fast-function-to-check-your-floating-point-rounding-mode/
+  //
+  // This function is meant to be equivalent to :
+  // prior: #include <cfenv>
+  //  return fegetround() == FE_TONEAREST;
+  // However, it is expected to be much faster than the fegetround()
+  // function call.
+  //
+  // The volatile keywoard prevents the compiler from computing the function
+  // at compile-time.
+  // There might be other ways to prevent compile-time optimizations (e.g., asm).
+  // The value does not need to be std::numeric_limits<float>::min(), any small
+  // value so that 1 + x should round to 1 would do (after accounting for excess
+  // precision, as in 387 instructions).
+  static volatile float fmin = std::numeric_limits<float>::min();
+  float fmini = fmin; // we copy it so that it gets loaded at most once.
+  //
+  // Explanation:
+  // Only when fegetround() == FE_TONEAREST do we have that
+  // fmin + 1.0f == 1.0f - fmin.
+  //
+  // FE_UPWARD:
+  //  fmin + 1.0f > 1
+  //  1.0f - fmin == 1
+  //
+  // FE_DOWNWARD or  FE_TOWARDZERO:
+  //  fmin + 1.0f == 1
+  //  1.0f - fmin < 1
+  //
+  // Note: This may fail to be accurate if fast-math has been
+  // enabled, as rounding conventions may not apply.
+  #if FASTFLOAT_VISUAL_STUDIO
+  #   pragma warning(push)
+  //  todo: is there a VS warning?
+  //  see https://stackoverflow.com/questions/46079446/is-there-a-warning-for-floating-point-equality-checking-in-visual-studio-2013
+  #elif defined(__clang__)
+  #   pragma clang diagnostic push
+  #   pragma clang diagnostic ignored "-Wfloat-equal"
+  #elif defined(__GNUC__)
+  #   pragma GCC diagnostic push
+  #   pragma GCC diagnostic ignored "-Wfloat-equal"
+  #endif
+  return (fmini + 1.0f == 1.0f - fmini);
+  #if FASTFLOAT_VISUAL_STUDIO
+  #   pragma warning(pop)
+  #elif defined(__clang__)
+  #   pragma clang diagnostic pop
+  #elif defined(__GNUC__)
+  #   pragma GCC diagnostic pop
+  #endif
+}
+
 } // namespace detail
 
 template<typename T>
@@ -76,6 +140,11 @@ from_chars_result from_chars_advanced(const char *first, const char *last,
 
 
   from_chars_result answer;
+#if FASTFLOAT_SKIP_WHITE_SPACE  // disabled by default
+  while ((first != last) && fast_float::is_space(uint8_t(*first))) {
+    first++;
+  }
+#endif
   if (first == last) {
     answer.ec = std::errc::invalid_argument;
     answer.ptr = first;
@@ -87,13 +156,45 @@ from_chars_result from_chars_advanced(const char *first, const char *last,
   }
   answer.ec = std::errc(); // be optimistic
   answer.ptr = pns.lastmatch;
-  // Next is Clinger's fast path.
+  // The implementation of the Clinger's fast path is convoluted because
-  if (binary_format<T>::min_exponent_fast_path() <= pns.exponent && pns.exponent <= binary_format<T>::max_exponent_fast_path() && pns.mantissa <=binary_format<T>::max_mantissa_fast_path() && !pns.too_many_digits) {
+  // we want round-to-nearest in all cases, irrespective of the rounding mode
-    value = T(pns.mantissa);
+  // selected on the thread.
-    if (pns.exponent < 0) { value = value / binary_format<T>::exact_power_of_ten(-pns.exponent); }
+  // We proceed optimistically, assuming that detail::rounds_to_nearest() returns
-    else { value = value * binary_format<T>::exact_power_of_ten(pns.exponent); }
+  // true.
-    if (pns.negative) { value = -value; }
+  if (binary_format<T>::min_exponent_fast_path() <= pns.exponent && pns.exponent <= binary_format<T>::max_exponent_fast_path() && !pns.too_many_digits) {
-    return answer;
+    // Unfortunately, the conventional Clinger's fast path is only possible
+    // when the system rounds to the nearest float.
+    //
+    // We expect the next branch to almost always be selected.
+    // We could check it first (before the previous branch), but
+    // there might be performance advantages at having the check
+    // be last.
+    if(detail::rounds_to_nearest())  {
+      // We have that fegetround() == FE_TONEAREST.
+      // Next is Clinger's fast path.
+      if (pns.mantissa <=binary_format<T>::max_mantissa_fast_path()) {
+        value = T(pns.mantissa);
+        if (pns.exponent < 0) { value = value / binary_format<T>::exact_power_of_ten(-pns.exponent); }
+        else { value = value * binary_format<T>::exact_power_of_ten(pns.exponent); }
+        if (pns.negative) { value = -value; }
+        return answer;
+      }
+    } else {
+      // We do not have that fegetround() == FE_TONEAREST.
+      // Next is a modified Clinger's fast path, inspired by Jakub Jelínek's proposal
+      if (pns.exponent >= 0 && pns.mantissa <=binary_format<T>::max_mantissa_fast_path(pns.exponent)) {
+#if defined(__clang__)
+        // Clang may map 0 to -0.0 when fegetround() == FE_DOWNWARD
+        if(pns.mantissa == 0) {
+          value = 0;
+          return answer;
+        }
+#endif
+        value = T(pns.mantissa) * binary_format<T>::exact_power_of_ten(pns.exponent);
+        if (pns.negative) { value = -value; }
+        return answer;
+      }
+    }
   }
   adjusted_mantissa am = compute_float<binary_format<T>>(pns.exponent, pns.mantissa);
   if(pns.too_many_digits && am.power2 >= 0) {

include/fast_float/simple_decimal_conversion.h

@@ -137,7 +137,7 @@ inline uint64_t round(decimal &h) {
   }
   bool round_up = false;
   if (dp < h.num_digits) {
-    round_up = h.digits[dp] >= 5; // normally, we round up  
+    round_up = h.digits[dp] >= 5; // normally, we round up
     // but we may need to round to even!
     if ((h.digits[dp] == 5) && (dp + 1 == h.num_digits)) {
       round_up = h.truncated || ((dp > 0) && (1 & h.digits[dp - 1]));
@@ -266,7 +266,7 @@ adjusted_mantissa compute_float(decimal &d) {
     return answer;
   } else if(d.decimal_point >= 310) {
     // We have something at least as large as 0.1e310 which is
-    // always infinite.  
+    // always infinite.
     answer.power2 = binary::infinite_power();
     answer.mantissa = 0;
     return answer;

script/amalgamate.py

@@ -3,46 +3,78 @@ processed_files = { }
 
 # authors
 for filename in ['AUTHORS', 'CONTRIBUTORS']:
-  with open(filename) as f:
+  with open(filename, encoding='utf8') as f:
     text = ''
     for line in f:
       if filename == 'AUTHORS':
         text += '// fast_float by ' + line
       if filename == 'CONTRIBUTORS':
         text += '// with contributions from ' + line
-    processed_files[filename] = text
+    processed_files[filename] = text + '//\n'
 
 # licenses
 for filename in ['LICENSE-MIT', 'LICENSE-APACHE']:
-  with open(filename) as f:
+  lines = []
-    text = ''
+  with open(filename, encoding='utf8') as f:
-    for line in f:
+    lines = f.readlines()
-      text += '// ' + line
+
-    processed_files[filename] = text
+  # Retrieve subset required for inclusion in source
+  if filename == 'LICENSE-APACHE':
+    lines = [
+      '   Copyright 2021 The fast_float authors\n',
+      *lines[179:-1]
+    ]
+
+  text = ''
+  for line in lines:
+    text += '//    ' + line.strip() + '\n'
+  processed_files[filename] = text
 
 # code
 for filename in [ 'fast_float.h', 'float_common.h', 'ascii_number.h', 
                   'fast_table.h', 'decimal_to_binary.h', 'bigint.h',
                   'ascii_number.h', 'digit_comparison.h', 'parse_number.h']:
-  with open('include/fast_float/' + filename) as f:
+  with open('include/fast_float/' + filename, encoding='utf8') as f:
     text = ''
     for line in f:
       if line.startswith('#include "'): continue
       text += line
-    processed_files[filename] = text
+    processed_files[filename] = '\n' + text
 
 # command line
 import argparse
 
 parser = argparse.ArgumentParser(description='Amalgamate fast_float.')
-parser.add_argument('--license', default='MIT', help='choose license')
+parser.add_argument('--license', default='DUAL', choices=['DUAL', 'MIT', 'APACHE'], help='choose license')
 parser.add_argument('--output', default='', help='output file (stdout if none')
 
 args = parser.parse_args()
 
-text = '\n\n'.join([
+def license_content(license_arg):
+  result = []
+
+  if license_arg == 'DUAL':
+    result += [
+      '// Licensed under the Apache License, Version 2.0, or the\n',
+      '// MIT License at your option. This file may not be copied,\n',
+      '// modified, or distributed except according to those terms.\n',
+      '//\n'
+    ]
+
+  if license_arg in ('DUAL', 'MIT'):
+    result.append('// MIT License Notice\n//\n')
+    result.append(processed_files['LICENSE-MIT'])
+    result.append('//\n')
+  if license_arg in ('DUAL', 'APACHE'):
+    result.append('// Apache License (Version 2.0) Notice\n//\n')
+    result.append(processed_files['LICENSE-APACHE'])
+    result.append('//\n')
+
+  return result
+
+text = ''.join([
   processed_files['AUTHORS'], processed_files['CONTRIBUTORS'], 
-  processed_files['LICENSE-' + args.license],
+  *license_content(args.license),
   processed_files['fast_float.h'], processed_files['float_common.h'], 
   processed_files['ascii_number.h'], processed_files['fast_table.h'],
   processed_files['decimal_to_binary.h'], processed_files['bigint.h'],
@@ -50,7 +82,7 @@ text = '\n\n'.join([
   processed_files['parse_number.h']])
 
 if args.output:
-  with open(args.output, 'wt') as f:
+  with open(args.output, 'wt', encoding='utf8') as f:
     f.write(text)
 else:
   print(text)

tests/CMakeLists.txt

@@ -38,7 +38,11 @@ add_library(supplemental-data INTERFACE)
 target_compile_definitions(supplemental-data INTERFACE SUPPLEMENTAL_TEST_DATA_DIR="${supplemental_test_files_BINARY_DIR}/data")
 function(fast_float_add_cpp_test TEST_NAME)
     add_executable(${TEST_NAME} ${TEST_NAME}.cpp)
-    add_test(${TEST_NAME} ${TEST_NAME})
+    if(CMAKE_CROSSCOMPILING)
+        set(ccemulator ${CMAKE_CROSSCOMPILING_EMULATOR})
+    endif()
+    add_test(NAME ${TEST_NAME}
+        COMMAND ${ccemulator} $<TARGET_FILE:${TEST_NAME}>)
     if(CMAKE_CXX_COMPILER_ID MATCHES "MSVC")
       target_compile_options(${TEST_NAME} PUBLIC /EHsc)
     endif()
@@ -52,10 +56,12 @@ function(fast_float_add_cpp_test TEST_NAME)
     endif()
 endfunction(fast_float_add_cpp_test)
 
-
+fast_float_add_cpp_test(rcppfastfloat_test)
 fast_float_add_cpp_test(example_test)
 fast_float_add_cpp_test(example_comma_test)
 fast_float_add_cpp_test(basictest)
+target_compile_features(basictest PRIVATE cxx_std_17)
+
 fast_float_add_cpp_test(long_test)
 fast_float_add_cpp_test(powersoffive_hardround)
 fast_float_add_cpp_test(string_test)

tests/basictest.cpp

@@ -10,6 +10,7 @@
 #include <limits>
 #include <string>
 #include <system_error>
+#include <cfenv>
 
 #ifndef SUPPLEMENTAL_TEST_DATA_DIR
 #define SUPPLEMENTAL_TEST_DATA_DIR "data/"
@@ -42,6 +43,147 @@
 #define FASTFLOAT_ODDPLATFORM 1
 #endif
 
+
+#define iHexAndDec(v) std::hex << "0x" << (v) << " (" << std::dec << (v) << ")"
+#define fHexAndDec(v) std::hexfloat << (v) << " (" << std::defaultfloat << (v) << ")"
+
+
+const char * round_name(int d) {
+  switch(d) {
+    case FE_UPWARD:
+      return "FE_UPWARD";
+    case FE_DOWNWARD:
+      return "FE_DOWNWARD";
+    case FE_TOWARDZERO:
+      return "FE_TOWARDZERO";
+    case FE_TONEAREST:
+      return "FE_TONEAREST";
+    default:
+      return "UNKNOWN";
+  }
+}
+
+
+#define FASTFLOAT_STR(x)   #x
+#define SHOW_DEFINE(x) printf("%s='%s'\n", #x, FASTFLOAT_STR(x))
+
+TEST_CASE("system_info") {
+    std::cout << "system info:" << std::endl;
+#ifdef _MSC_VER
+    SHOW_DEFINE(_MSC_VER);
+#endif
+#ifdef FASTFLOAT_64BIT_LIMB
+    SHOW_DEFINE(FASTFLOAT_64BIT_LIMB);
+#endif
+#ifdef __clang__
+    SHOW_DEFINE(__clang__);
+#endif
+#ifdef FASTFLOAT_VISUAL_STUDIO
+    SHOW_DEFINE(FASTFLOAT_VISUAL_STUDIO);
+#endif
+#ifdef FASTFLOAT_IS_BIG_ENDIAN
+    #if FASTFLOAT_IS_BIG_ENDIAN
+      printf("big endian\n");
+    #else
+      printf("little endian\n");
+    #endif
+#endif
+#ifdef FASTFLOAT_32BIT
+    SHOW_DEFINE(FASTFLOAT_32BIT);
+#endif
+#ifdef FASTFLOAT_64BIT
+    SHOW_DEFINE(FASTFLOAT_64BIT);
+#endif
+#ifdef FLT_EVAL_METHOD
+    SHOW_DEFINE(FLT_EVAL_METHOD);
+#endif
+#ifdef _WIN32
+    SHOW_DEFINE(_WIN32);
+#endif
+#ifdef _WIN64
+    SHOW_DEFINE(_WIN64);
+#endif
+    std::cout << "fegetround() = " << round_name(fegetround()) << std::endl;
+    std::cout << std::endl;
+
+}
+
+
+TEST_CASE("rounds_to_nearest") {
+  //
+  // If this function fails, we may be left in a non-standard rounding state.
+  //
+  static volatile float fmin = std::numeric_limits<float>::min();
+  fesetround(FE_UPWARD);
+  std::cout << "FE_UPWARD: fmin + 1.0f = " << iHexAndDec(fmin + 1.0f) << " 1.0f - fmin = " << iHexAndDec(1.0f - fmin) << std::endl;
+  CHECK(fegetround() == FE_UPWARD);
+  CHECK(fast_float::detail::rounds_to_nearest() == false);
+
+  fesetround(FE_DOWNWARD);
+  std::cout << "FE_DOWNWARD: fmin + 1.0f = " << iHexAndDec(fmin + 1.0f) << " 1.0f - fmin = " << iHexAndDec(1.0f - fmin) << std::endl;
+  CHECK(fegetround() == FE_DOWNWARD);
+  CHECK(fast_float::detail::rounds_to_nearest() == false);
+
+  fesetround(FE_TOWARDZERO);
+  std::cout << "FE_TOWARDZERO: fmin + 1.0f = " << iHexAndDec(fmin + 1.0f) << " 1.0f - fmin = " << iHexAndDec(1.0f - fmin) << std::endl;
+  CHECK(fegetround() == FE_TOWARDZERO);
+  CHECK(fast_float::detail::rounds_to_nearest() == false);
+
+  fesetround(FE_TONEAREST);
+  std::cout << "FE_TONEAREST: fmin + 1.0f = " << iHexAndDec(fmin + 1.0f) << " 1.0f - fmin = " << iHexAndDec(1.0f - fmin) << std::endl;
+  CHECK(fegetround() == FE_TONEAREST);
+#if (FLT_EVAL_METHOD == 1) || (FLT_EVAL_METHOD == 0)
+  CHECK(fast_float::detail::rounds_to_nearest() == true);
+#endif
+}
+
+TEST_CASE("parse_zero") {
+  //
+  // If this function fails, we may be left in a non-standard rounding state.
+  //
+  const char * zero = "0";
+  uint64_t float64_parsed;
+  double f = 0;
+  ::memcpy(&float64_parsed, &f, sizeof(f));
+  CHECK(float64_parsed == 0);
+
+  fesetround(FE_UPWARD);
+  auto r1 = fast_float::from_chars(zero, zero + 1, f);
+  CHECK(r1.ec == std::errc());
+  std::cout << "FE_UPWARD parsed zero as " << iHexAndDec(f) << std::endl;
+  CHECK(f == 0);
+  ::memcpy(&float64_parsed, &f, sizeof(f));
+  std::cout << "double as uint64_t is " << float64_parsed << std::endl;
+  CHECK(float64_parsed == 0);
+
+  fesetround(FE_TOWARDZERO);
+  auto r2 = fast_float::from_chars(zero, zero + 1, f);
+  CHECK(r2.ec == std::errc());
+  std::cout << "FE_TOWARDZERO parsed zero as " << iHexAndDec(f) << std::endl;
+  CHECK(f == 0);
+  ::memcpy(&float64_parsed, &f, sizeof(f));
+  std::cout << "double as uint64_t is " << float64_parsed << std::endl;
+  CHECK(float64_parsed == 0);
+
+  fesetround(FE_DOWNWARD);
+  auto r3 = fast_float::from_chars(zero, zero + 1, f);
+  CHECK(r3.ec == std::errc());
+  std::cout << "FE_DOWNWARD parsed zero as " << iHexAndDec(f) << std::endl;
+  CHECK(f == 0);
+  ::memcpy(&float64_parsed, &f, sizeof(f));
+  std::cout << "double as uint64_t is " << float64_parsed << std::endl;
+  CHECK(float64_parsed == 0);
+
+  fesetround(FE_TONEAREST);
+  auto r4 = fast_float::from_chars(zero, zero + 1, f);
+  CHECK(r4.ec == std::errc());
+  std::cout << "FE_TONEAREST parsed zero as " << iHexAndDec(f) << std::endl;
+  CHECK(f == 0);
+  ::memcpy(&float64_parsed, &f, sizeof(f));
+  std::cout << "double as uint64_t is " << float64_parsed << std::endl;
+  CHECK(float64_parsed == 0);
+}
+
 // C++ 17 because it is otherwise annoying to browse all files in a directory.
 // We also only run these tests on little endian systems.
 #if (FASTFLOAT_CPLUSPLUS >= 201703L) && (FASTFLOAT_IS_BIG_ENDIAN == 0) && !defined(FASTFLOAT_ODDPLATFORM)
@@ -50,59 +192,77 @@
 #include <filesystem>
 #include <charconv>
 
-// return true on succcess
+
+
+// return true on success
 bool check_file(std::string file_name) {
   std::cout << "Checking " << file_name << std::endl;
-  size_t number{0};
+  // We check all rounding directions, for each file.
-  std::fstream newfile(file_name, std::ios::in);
+  std::vector<int> directions = {FE_UPWARD, FE_DOWNWARD, FE_TOWARDZERO, FE_TONEAREST};
-  if (newfile.is_open()) {
+  for (int d : directions) {
-    std::string str;
+    std::cout << "fesetround to " << round_name(d) << std::endl;
-    while (std::getline(newfile, str)) {
+    fesetround(d);
-      if (str.size() > 0) {
+    size_t number{0};
-        // Read 32-bit hex
+    std::fstream newfile(file_name, std::ios::in);
-        uint32_t float32;
+    if (newfile.is_open()) {
-        auto r32 = std::from_chars(str.data() + 5, str.data() + str.size(),
+      std::string str;
+      while (std::getline(newfile, str)) {
+        if (str.size() > 0) {
+          // Read 32-bit hex
+          uint32_t float32;
+          auto r32 = std::from_chars(str.data() + 5, str.data() + str.size(),
                                    float32, 16);
-        if(r32.ec != std::errc()) { std::cerr << "32-bit parsing failure\n"; return false; }
+          if(r32.ec != std::errc()) { std::cerr << "32-bit parsing failure\n"; return false; }
-        // Read 64-bit hex
+          // Read 64-bit hex
-        uint64_t float64;
+          uint64_t float64;
-        auto r64 = std::from_chars(str.data() + 14, str.data() + str.size(),
+          auto r64 = std::from_chars(str.data() + 14, str.data() + str.size(),
                                    float64, 16);
-        if(r64.ec != std::errc()) { std::cerr << "64-bit parsing failure\n"; return false; }
+          if(r64.ec != std::errc()) { std::cerr << "64-bit parsing failure\n"; return false; }
-        // The string to parse:
+          // The string to parse:
-        const char *number_string = str.data() + 31;
+          const char *number_string = str.data() + 31;
-        const char *end_of_string = str.data() + str.size();
+          const char *end_of_string = str.data() + str.size();
-        // Parse as 32-bit float
+          // Parse as 32-bit float
-        float parsed_32;
+          float parsed_32;
-        auto fast_float_r32 = fast_float::from_chars(number_string, end_of_string, parsed_32);
+          auto fast_float_r32 = fast_float::from_chars(number_string, end_of_string, parsed_32);
-        if(fast_float_r32.ec != std::errc()) { std::cerr << "parsing failure\n"; return false; }
+          if(fast_float_r32.ec != std::errc()) { std::cerr << "parsing failure\n"; return false; }
-        // Parse as 64-bit float
+          // Parse as 64-bit float
-        double parsed_64;
+          double parsed_64;
-        auto fast_float_r64 = fast_float::from_chars(number_string, end_of_string, parsed_64);
+          auto fast_float_r64 = fast_float::from_chars(number_string, end_of_string, parsed_64);
-        if(fast_float_r64.ec != std::errc()) { std::cerr << "parsing failure\n"; return false; }
+          if(fast_float_r64.ec != std::errc()) { std::cerr << "parsing failure\n"; return false; }
-        // Convert the floats to unsigned ints.
+          // Convert the floats to unsigned ints.
-        uint32_t float32_parsed;
+          uint32_t float32_parsed;
-        uint64_t float64_parsed;
+          uint64_t float64_parsed;
-        ::memcpy(&float32_parsed, &parsed_32, sizeof(parsed_32));
+          ::memcpy(&float32_parsed, &parsed_32, sizeof(parsed_32));
-        ::memcpy(&float64_parsed, &parsed_64, sizeof(parsed_64));
+          ::memcpy(&float64_parsed, &parsed_64, sizeof(parsed_64));
-        // Compare with expected results
+          // Compare with expected results
-        if (float32_parsed != float32) {
+          if (float32_parsed != float32) {
-          std::cout << "bad 32 " << str << std::endl;
+            std::cout << "bad 32 " << str << std::endl;
-          return false;
+            std::cout << "parsed as " << iHexAndDec(parsed_32) << std::endl;
+            std::cout << "as  raw uint32_t, parsed = " << float32_parsed << ", expected = " << float32 << std::endl;
+            std::cout << "fesetround: " << round_name(d) << std::endl;
+            fesetround(FE_TONEAREST);
+            return false;
+          }
+          if (float64_parsed != float64) {
+            std::cout << "bad 64 " << str << std::endl;
+            std::cout << "parsed as " << iHexAndDec(parsed_64) << std::endl;
+            std::cout << "as raw uint64_t, parsed = " << float64_parsed << ", expected = " << float64 << std::endl;
+            std::cout << "fesetround: " << round_name(d) << std::endl;
+            fesetround(FE_TONEAREST);
+            return false;
+          }
+          number++;
         }
-        if (float64_parsed != float64) {
-          std::cout << "bad 64 " << str << std::endl;
-          return false;
-        }
-        number++;
       }
+      std::cout << "checked " << std::defaultfloat << number << " values" << std::endl;
+      newfile.close(); // close the file object
+    } else {
+      std::cout << "Could not read  " << file_name << std::endl;
+      fesetround(FE_TONEAREST);
+      return false;
     }
-    std::cout << "checked " << std::defaultfloat << number << " values" << std::endl;
-    newfile.close(); // close the file object
-  } else {
-    std::cout << "Could not read  " << file_name << std::endl;
-    return false;
   }
+  fesetround(FE_TONEAREST);
   return true;
 }
 
@@ -125,9 +285,6 @@ TEST_CASE("leading_zeroes") {
   CHECK(fast_float::leading_zeroes(bit << 63) ==  0);
 }
 
-#define iHexAndDec(v) std::hex << "0x" << (v) << " (" << std::dec << (v) << ")"
-#define fHexAndDec(v) std::hexfloat << (v) << " (" << std::defaultfloat << (v) << ")"
-
 void test_full_multiplication(uint64_t lhs, uint64_t rhs, uint64_t expected_lo, uint64_t expected_hi) {
   fast_float::value128 v;
   v = fast_float::full_multiplication(lhs, rhs);
@@ -224,6 +381,33 @@ TEST_CASE("decimal_point_parsing") {
   }
 }
 
+TEST_CASE("issue19") {
+  const std::string input =   "234532.3426362,7869234.9823,324562.645";
+  double result;
+  auto answer = fast_float::from_chars(input.data(), input.data()+input.size(), result);
+  CHECK_MESSAGE(answer.ec == std::errc(), "We want to parse up to 234532.3426362\n");
+  CHECK_MESSAGE(answer.ptr == input.data() + 14,
+                "Parsed the number " << result
+                << " and stopped at the wrong character: after " << (answer.ptr - input.data()) << " characters");
+  CHECK_MESSAGE(result == 234532.3426362, "We want to parse234532.3426362\n");
+  CHECK_MESSAGE(answer.ptr[0] == ',', "We want to parse up to the comma\n");
+
+  answer = fast_float::from_chars(answer.ptr + 1, input.data()+input.size(), result);
+  CHECK_MESSAGE(answer.ec == std::errc(), "We want to parse 7869234.9823\n");
+  CHECK_MESSAGE(answer.ptr == input.data() + 27,
+                "Parsed the number " << result
+                << " and stopped at the wrong character " << (answer.ptr - input.data()));
+  CHECK_MESSAGE(answer.ptr[0] == ',', "We want to parse up to the comma\n");
+  CHECK_MESSAGE(result == 7869234.9823, "We want to parse up 7869234.9823\n");
+
+  answer = fast_float::from_chars(answer.ptr + 1, input.data()+input.size(), result);
+  CHECK_MESSAGE(answer.ec == std::errc(), "We want to parse 324562.645\n");
+  CHECK_MESSAGE(answer.ptr == input.data() + 38,
+                "Parsed the number " << result
+                << " and stopped at the wrong character " << (answer.ptr - input.data()));
+  CHECK_MESSAGE(result == 324562.645, "We want to parse up 7869234.9823\n");
+}
+
 TEST_CASE("issue19") {
   const std::string input =  "3.14e";
   double result;
@@ -454,6 +638,9 @@ TEST_CASE("64bit.inf") {
 }
 
 TEST_CASE("64bit.general") {
+  verify("22250738585072012e-324",0x1p-1022); /* limit between normal and subnormal*/
+  verify("-22250738585072012e-324",-0x1p-1022); /* limit between normal and subnormal*/
+  verify("-1e-999",-0.0);
   verify("-2.2222222222223e-322",-0x1.68p-1069);
   verify("9007199254740993.0", 0x1p+53);
   verify("860228122.6654514319E+90", 0x1.92bb20990715fp+328);
@@ -588,6 +775,7 @@ TEST_CASE("32bit.inf") {
 }
 
 TEST_CASE("32bit.general") {
+  verify("-1e-999",-0.0f);
   verify("1.1754941406275178592461758986628081843312458647327962400313859427181746759860647699724722770042717456817626953125", 0x1.2ced3p+0f);
   verify("1.1754941406275178592461758986628081843312458647327962400313859427181746759860647699724722770042717456817626953125e-38", 0x1.fffff8p-127f);
   verify(append_zeros("1.1754941406275178592461758986628081843312458647327962400313859427181746759860647699724722770042717456817626953125",655), 0x1.2ced3p+0f);

tests/example_test.cpp

@@ -3,12 +3,57 @@
 #include <iostream>
 #include <string>
 #include <system_error>
- 
+
+
+bool many() {
+  const std::string input =   "234532.3426362,7869234.9823,324562.645";
+  double result;
+  auto answer = fast_float::from_chars(input.data(), input.data()+input.size(), result);
+  if(answer.ec != std::errc()) { return false; }
+  if(result != 234532.3426362) { return false; }
+  if(answer.ptr[0] != ',') { return false; }
+  answer = fast_float::from_chars(answer.ptr + 1, input.data()+input.size(), result);
+  if(answer.ec != std::errc()) { return false; }
+  if(result != 7869234.9823) { return false; }
+  if(answer.ptr[0] != ',') { return false; }
+  answer = fast_float::from_chars(answer.ptr + 1, input.data()+input.size(), result);
+  if(answer.ec != std::errc()) { return false; }
+  if(result != 324562.645) { return false; }
+  return true;
+}
+
+
+void many_loop() {
+  const std::string input =   "234532.3426362,7869234.9823,324562.645";
+  double result;
+  const char* pointer = input.data();
+  const char* end_pointer = input.data() + input.size();
+
+  while(pointer < end_pointer) {
+    auto answer = fast_float::from_chars(pointer, end_pointer, result);
+    if(answer.ec != std::errc()) {
+        std::cerr << "error while parsing" << std::endl;
+        break;
+    }
+    std::cout << "parsed: " << result << std::endl;
+    pointer = answer.ptr;
+    if ((answer.ptr < end_pointer) && (*pointer == ',')) {
+      pointer++;
+    }
+  }
+}
+
 int main() {
     const std::string input =  "3.1416 xyz ";
     double result;
     auto answer = fast_float::from_chars(input.data(), input.data()+input.size(), result);
     if((answer.ec != std::errc()) || ((result != 3.1416))) { std::cerr << "parsing failure\n"; return EXIT_FAILURE; }
     std::cout << "parsed the number " << result << std::endl;
+
+    if(!many()) {
+        printf("Bug\n");
+        return EXIT_FAILURE;
+    }
+    many_loop();
     return EXIT_SUCCESS;
 }

tests/exhaustive32_midpoint.cpp

@@ -132,7 +132,7 @@ bool allvalues() {
 
 inline void Assert(bool Assertion) {
 #if defined(__CYGWIN__) || defined(__MINGW32__) || defined(__MINGW64__)  || defined(sun) || defined(__sun)
-  if (!Assertion) { std::cerr << "Omitting hard falure on msys/cygwin/sun systems."; }
+  if (!Assertion) { std::cerr << "Omitting hard failure on msys/cygwin/sun systems."; }
 #else 
   if (!Assertion) { throw std::runtime_error("bug"); }
 #endif

tests/installation_tests/find/CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.15)
 
-project(test_simdjson_install VERSION 0.1.0 LANGUAGES CXX)
+project(test_install VERSION 0.1.0 LANGUAGES CXX)
 
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
@@ -27,4 +27,4 @@ int main() {
 
 
 add_executable(repro main.cpp)
-target_link_libraries(repro PUBLIC FastFloat::fast_float)
+target_link_libraries(repro PUBLIC FastFloat::fast_float)

tests/rcppfastfloat_test.cpp

@@ -0,0 +1,123 @@
+/**
+ * See https://github.com/eddelbuettel/rcppfastfloat/issues/4
+ */
+
+#define FASTFLOAT_ALLOWS_LEADING_PLUS 1
+#define FASTFLOAT_SKIP_WHITE_SPACE 1 // important !
+#include "fast_float/fast_float.h"
+#include <iostream>
+#include <string>
+#include <vector>
+
+bool eddelbuettel() {
+  std::vector<std::string> inputs = {"infinity",
+                                     " \r\n\t\f\v3.16227766016838 \r\n\t\f\v",
+                                     " \r\n\t\f\v3 \r\n\t\f\v",
+                                     "  1970-01-01",
+                                     "-NaN",
+                                     "-inf",
+                                     " \r\n\t\f\v2.82842712474619 \r\n\t\f\v",
+                                     "nan",
+                                     " \r\n\t\f\v2.44948974278318 \r\n\t\f\v",
+                                     "Inf",
+                                     " \r\n\t\f\v2 \r\n\t\f\v",
+                                     "-infinity",
+                                     " \r\n\t\f\v0 \r\n\t\f\v",
+                                     " \r\n\t\f\v1.73205080756888 \r\n\t\f\v",
+                                     " \r\n\t\f\v1 \r\n\t\f\v",
+                                     " \r\n\t\f\v1.4142135623731 \r\n\t\f\v",
+                                     " \r\n\t\f\v2.23606797749979 \r\n\t\f\v",
+                                     "1970-01-02  ",
+                                     " \r\n\t\f\v2.64575131106459 \r\n\t\f\v",
+                                     "inf",
+                                     "-nan",
+                                     "NaN",
+                                     "",
+                                     "-Inf",
+                                     "+2.2"};
+  std::vector<std::pair<bool, double>> expected_results = {
+      {true, std::numeric_limits<double>::infinity()},
+      {true, 3.16227766016838},
+      {true, 3},
+      {false, -1},
+      {true, std::numeric_limits<double>::quiet_NaN()},
+      {true, -std::numeric_limits<double>::infinity()},
+      {true, 2.82842712474619},
+      {true, std::numeric_limits<double>::quiet_NaN()},
+      {true, 2.44948974278318},
+      {true, std::numeric_limits<double>::infinity()},
+      {true, 2},
+      {true, -std::numeric_limits<double>::infinity()},
+      {true, 0},
+      {true, 1.73205080756888},
+      {true, 1},
+      {true, 1.4142135623731},
+      {true, 2.23606797749979},
+      {false, -1},
+      {true, 2.64575131106459},
+      {true, std::numeric_limits<double>::infinity()},
+      {true, std::numeric_limits<double>::quiet_NaN()},
+      {true, std::numeric_limits<double>::quiet_NaN()},
+      {false, -1},
+      {true, -std::numeric_limits<double>::infinity()},
+      {true, 2.2}};
+  for (size_t i = 0; i < inputs.size(); i++) {
+    std::string &input = inputs[i];
+    std::pair<bool, double> expected = expected_results[i];
+    double result;
+    // answer contains a error code and a pointer to the end of the
+    // parsed region (on success).
+    auto answer = fast_float::from_chars(input.data(),
+                                         input.data() + input.size(), result);
+    if (answer.ec != std::errc()) {
+      std::cout << "could not parse" << std::endl;
+      if (expected.first) {
+        return false;
+      }
+      continue;
+    }
+    bool non_space_trailing_content = false;
+    if (answer.ptr != input.data() + input.size()) {
+      // check that there is no content left
+      for (const char *leftover = answer.ptr;
+           leftover != input.data() + input.size(); leftover++) {
+        if (!fast_float::is_space(uint8_t(*leftover))) {
+          non_space_trailing_content = true;
+          break;
+        }
+      }
+    }
+    if (non_space_trailing_content) {
+      std::cout << "found trailing content " << std::endl;
+    }
+
+    if (non_space_trailing_content) {
+      if (!expected.first) {
+        continue;
+      } else {
+        return false;
+      }
+    }
+    std::cout << "parsed " << result << std::endl;
+    if (!expected.first) {
+      return false;
+    }
+    if (result != expected.second) {
+      if (std::isnan(result) && std::isnan(expected.second)) {
+        continue;
+      }
+      std::cout << "results do not match. Expected "<<  expected.second << std::endl;
+      return false;
+    }
+  }
+  return true;
+}
+
+int main() {
+  if (!eddelbuettel()) {
+    printf("Bug.\n");
+    return EXIT_FAILURE;
+  }
+  printf("All ok.\n");
+  return EXIT_SUCCESS;
+}

tests/string_test.cpp

@@ -44,8 +44,8 @@ float cygwin_strtof_l(const char* start, char** end) {
 
 inline void Assert(bool Assertion) {
 #if defined(__CYGWIN__) || defined(__MINGW32__) || defined(__MINGW64__)  || defined(sun) || defined(__sun)
-  if (!Assertion) { std::cerr << "Omitting hard falure on msys/cygwin/sun systems."; }
+  if (!Assertion) { std::cerr << "Omitting hard failure on msys/cygwin/sun systems."; }
-#else 
+#else
   if (!Assertion) { throw std::runtime_error("bug"); }
 #endif
 }