Merge pull request #237 from fastfloat/release610_candidate

preparing version 6.1.0

CMakeLists.txt

@@ -1,6 +1,8 @@
 cmake_minimum_required(VERSION 3.9)
 
 project(fast_float VERSION 6.0.0 LANGUAGES CXX)
+set(FASTFLOAT_CXX_STANDARD 11 CACHE STRING "the C++ standard to use for fastfloat")
+set(CMAKE_CXX_STANDARD ${FASTFLOAT_CXX_STANDARD})
 option(FASTFLOAT_TEST "Enable tests" OFF)
 if(FASTFLOAT_TEST)
   enable_testing()

README.md

@@ -165,6 +165,16 @@ constexpr double constexptest() {
 }
 ```
 
+## C++23: Fixed width floating-point types
+
+The library also supports fixed-width floating-point types such as `std::float32_t` and `std::float64_t`. E.g., you can write:
+
+```C++
+std::float32_t result;
+auto answer = fast_float::from_chars(f.data(), f.data() + f.size(), result);
+``````
+
+
 ## Non-ASCII Inputs
 
 We also support UTF-16 and UTF-32 inputs, as well as ASCII/UTF-8, as in the following example:
@@ -371,7 +381,7 @@ the command line help.
 
 You may directly download automatically generated single-header files:
 
-https://github.com/fastfloat/fast_float/releases/download/v6.0.0/fast_float.h
+https://github.com/fastfloat/fast_float/releases/download/v6.1.0/fast_float.h
 
 ## RFC 7159 
 

include/fast_float/ascii_number.h

@@ -529,8 +529,8 @@ from_chars_result_t<UC> parse_int_string(UC const* p, UC const* pend, T& value,
     // this weird workaround is required because:
     // - converting unsigned to signed when its value is greater than signed max is UB pre-C++23.
     // - reinterpret_casting (~i + 1) would work, but it is not constexpr
-    // this is always optimized into a neg instruction.
-    value = T(-std::numeric_limits<T>::max() - T(i - std::numeric_limits<T>::max()));
+    // this is always optimized into a neg instruction (note: T is an integer type)
+    value = T(-std::numeric_limits<T>::max() - T(i - uint64_t(std::numeric_limits<T>::max())));
 #ifdef FASTFLOAT_VISUAL_STUDIO
 #pragma warning(pop)
 #endif

include/fast_float/float_common.h

@@ -7,7 +7,11 @@
 #include <cstring>
 #include <type_traits>
 #include <system_error>
-
+#ifdef __has_include
+  #if __has_include(<stdfloat>)
+    #include <stdfloat>
+  #endif
+#endif
 #include "constexpr_feature_detect.h"
 
 namespace fast_float {
@@ -188,7 +192,14 @@ fastfloat_really_inline constexpr bool cpp20_and_in_constexpr() {
 
 template <typename T>
 fastfloat_really_inline constexpr bool is_supported_float_type() {
-  return std::is_same<T, float>::value || std::is_same<T, double>::value;
+  return std::is_same<T, float>::value || std::is_same<T, double>::value
+#if __STDCPP_FLOAT32_T__
+    || std::is_same<T, std::float32_t>::value
+#endif
+#if __STDCPP_FLOAT64_T__
+    || std::is_same<T, std::float64_t>::value
+#endif
+  ;
 }
 
 template <typename UC>

include/fast_float/parse_number.h

@@ -10,11 +10,6 @@
 #include <cstring>
 #include <limits>
 #include <system_error>
-#ifdef __has_include
-  #if __has_include(<stdfloat>)
-    #include <stdfloat>
-  #endif
-#endif
 namespace fast_float {
 
 
@@ -186,6 +181,7 @@ struct from_chars_caller<std::float64_t>
 
 
 template<typename T, typename UC, typename>
+FASTFLOAT_CONSTEXPR20
 from_chars_result_t<UC> from_chars(UC const * first, UC const * last,
                              T &value, chars_format fmt /*= chars_format::general*/)  noexcept  {
   return from_chars_caller<T>::call(first, last, value, parse_options_t<UC>(fmt));
@@ -196,7 +192,7 @@ FASTFLOAT_CONSTEXPR20
 from_chars_result_t<UC> from_chars_advanced(UC const * first, UC const * last,
                                       T &value, parse_options_t<UC> options)  noexcept  {
 
-  static_assert (is_supported_float_type<T>(), "only float and double are supported");
+  static_assert (is_supported_float_type<T>(), "only some floating-point types are supported");
   static_assert (is_supported_char_type<UC>(), "only char, wchar_t, char16_t and char32_t are supported");
 
   from_chars_result_t<UC> answer;

tests/CMakeLists.txt

@@ -84,11 +84,14 @@ fast_float_add_cpp_test(fast_int)
 target_compile_features(fast_int PRIVATE cxx_std_17)
 fast_float_add_cpp_test(json_fmt)
 fast_float_add_cpp_test(fortran)
-
-option(FASTFLOAT_FIXEDWIDTH_TESTS "Require fixed width test for C++23 (build will fail if the compiler won't support it)" OFF)
+if(CMAKE_CXX_STANDARD GREATER_EQUAL 23)
+  option(FASTFLOAT_FIXEDWIDTH_TESTS "Require fixed width test for C++23 (build will fail if the compiler won't support it)" ON)
+else()
+  option(FASTFLOAT_FIXEDWIDTH_TESTS "Require fixed width test for C++23 (build will fail if the compiler won't support it)" OFF)
+endif()
 if (FASTFLOAT_FIXEDWIDTH_TESTS)
   fast_float_add_cpp_test(fixedwidthtest)
-  target_compile_features(fixedwidthtest PRIVATE cxx_std_23)
+  target_compile_features(fixedwidthtest PUBLIC cxx_std_23)
 endif()
 
 option(FASTFLOAT_EXHAUSTIVE "Exhaustive tests" OFF)