Merge branch 'develop' into pr1007

Fixes Conflicts:
	test/test_binomial.cpp

.github/workflows/ci.yml

@@ -181,6 +181,7 @@ jobs:
         shell: cmd
     env:
       ARGS: toolset=${{ matrix.toolset }} address-model=64 cxxstd=${{ matrix.standard }}
+      ARGSLATEST: toolset=${{ matrix.toolset }} address-model=64 cxxstd=latest
     strategy:
       fail-fast: false
       matrix:
@@ -214,9 +215,14 @@ jobs:
       - name: Config info
         run: config_info_travis
         working-directory: ../boost-root/libs/config/test
-      - name: Test
+      - name: Test std-14 vc140 and std-14-17 vc142
+        if: ${{ matrix.toolset != 'msvc-14.0' || matrix.standard != '17' }}
         run: ..\..\..\b2 --hash %ARGS% define=CI_SUPPRESS_KNOWN_ISSUES debug-symbols=off ${{ matrix.suite }} pch=off
         working-directory: ../boost-root/libs/math/test
+      - name: Test std-latest vc140
+        if: ${{ matrix.toolset == 'msvc-14.0' && matrix.standard == '17' }}
+        run: ..\..\..\b2 --hash %ARGSLATEST% define=CI_SUPPRESS_KNOWN_ISSUES debug-symbols=off ${{ matrix.suite }} pch=off
+        working-directory: ../boost-root/libs/math/test
   windows_gcc:
     runs-on: windows-2019
     defaults:

include/boost/math/concepts/real_concept.hpp

@@ -45,8 +45,12 @@
 #  include <cstdio>
 #endif
 
-#if __has_include(<stdfloat>)
-#  include <stdfloat>
+#if defined __has_include
+#  if __cplusplus > 202002L || _MSVC_LANG > 202002L 
+#    if __has_include (<stdfloat>)
+#    include <stdfloat>
+#    endif
+#  endif
 #endif
 
 namespace boost{ namespace math{

include/boost/math/tools/big_constant.hpp

@@ -60,7 +60,7 @@ inline T make_big_value(largest_float, const char* s, std::false_type const&, st
 }
 #else
 template <typename T>
-inline T make_big_value(largest_float, const char* s, std::false_type const&, std::false_type const&)
+inline T make_big_value(largest_float, const char*, std::false_type const&, std::false_type const&)
 {
    static_assert(sizeof(T) == 0, "Type is unsupported in standalone mode. Please disable and try again.");
 }

include/boost/math/tools/config.hpp

@@ -152,11 +152,11 @@
      // libstdc++3 only defines to/from_chars for std::float128_t when one of these defines are set
      // otherwise we're right out of luck...
 #    if defined(_GLIBCXX_LDOUBLE_IS_IEEE_BINARY128) || defined(_GLIBCXX_HAVE_FLOAT128_MATH)
-#    include <cstring> // std::strlen is used with from_chars
-#    include <charconv>
-#    include <stdfloat>
-#    define BOOST_MATH_USE_CHARCONV_FOR_CONVERSION
-#endif
+#      include <cstring> // std::strlen is used with from_chars
+#      include <charconv>
+#      include <stdfloat>
+#      define BOOST_MATH_USE_CHARCONV_FOR_CONVERSION
+#    endif
 #  endif
 #endif
 

include/boost/math/tools/promotion.hpp

@@ -26,8 +26,12 @@
 #include <boost/math/tools/config.hpp>
 #include <type_traits>
 
-#if __has_include(<stdfloat>)
-#  include <stdfloat>
+#if defined __has_include
+#  if __cplusplus > 202002L || _MSVC_LANG > 202002L 
+#    if __has_include (<stdfloat>)
+#    include <stdfloat>
+#    endif
+#  endif
 #endif
 
 namespace boost

test/Jamfile.v2

@@ -983,7 +983,7 @@ test-suite misc :
    [ run centered_continued_fraction_test.cpp : : : [ requires cxx17_if_constexpr cxx17_std_apply ] [ check-target-builds ../config//has_float128 "GCC libquadmath and __float128 support" : <linkflags>-lquadmath ] ]
    [ run luroth_expansion_test.cpp : : : [ requires cxx17_if_constexpr cxx17_std_apply ] [ check-target-builds ../config//has_float128 "GCC libquadmath and __float128 support" : <linkflags>-lquadmath ] ]
    [ run engel_expansion_test.cpp : : : [ requires cxx17_if_constexpr cxx17_std_apply ] [ check-target-builds ../config//has_float128 "GCC libquadmath and __float128 support" : <linkflags>-lquadmath ] ]
-   [ run test_classify.cpp pch ../../test/build//boost_unit_test_framework  ]
+   [ run test_classify.cpp pch ../../test/build//boost_unit_test_framework : : : <toolset>msvc:<cxxflags>/bigobj ]
    [ run test_error_handling.cpp ../../test/build//boost_unit_test_framework  ]
    [ run legendre_stieltjes_test.cpp ../../test/build//boost_unit_test_framework : : : [ requires cxx11_auto_declarations cxx11_range_based_for ] [ check-target-builds ../config//has_float128 "GCC libquadmath and __float128 support" : <linkflags>-lquadmath ]  ]
    [ run test_minima.cpp pch ../../test/build//boost_unit_test_framework  ]

test/test_autodiff_7.cpp

@@ -23,45 +23,4 @@ BOOST_AUTO_TEST_CASE_TEMPLATE(expm1_hpp, T, all_float_types) {
   }
 }
 
-BOOST_AUTO_TEST_CASE_TEMPLATE(fpclassify_hpp, T, all_float_types) {
-  using boost::math::fpclassify;
-  using boost::math::isfinite;
-  using boost::math::isinf;
-  using boost::math::isnan;
-  using boost::math::isnormal;
-  using boost::multiprecision::fpclassify;
-  using boost::multiprecision::isfinite;
-  using boost::multiprecision::isinf;
-  using boost::multiprecision::isnan;
-  using boost::multiprecision::isnormal;
-
-  using test_constants = test_constants_t<T>;
-  static constexpr auto m = test_constants::order;
-  test_detail::RandomSample<T> x_sampler{-1000, 1000};
-  for (auto i : boost::irange(test_constants::n_samples)) {
-    std::ignore = i;
-
-    BOOST_CHECK_EQUAL(fpclassify(make_fvar<T, m>(0)), FP_ZERO);
-    BOOST_CHECK_EQUAL(fpclassify(make_fvar<T, m>(10)), FP_NORMAL);
-    BOOST_CHECK_EQUAL(
-        fpclassify(make_fvar<T, m>(std::numeric_limits<T>::infinity())),
-        FP_INFINITE);
-    BOOST_CHECK_EQUAL(
-        fpclassify(make_fvar<T, m>(std::numeric_limits<T>::quiet_NaN())),
-        FP_NAN);
-    if (std::numeric_limits<T>::has_denorm != std::denorm_absent) {
-      BOOST_CHECK_EQUAL(
-          fpclassify(make_fvar<T, m>(std::numeric_limits<T>::denorm_min())),
-          FP_SUBNORMAL);
-    }
-
-    BOOST_CHECK(isfinite(make_fvar<T, m>(0)));
-    BOOST_CHECK(isnormal(make_fvar<T, m>((std::numeric_limits<T>::min)())));
-    BOOST_CHECK(
-        !isnormal(make_fvar<T, m>(std::numeric_limits<T>::denorm_min())));
-    BOOST_CHECK(isinf(make_fvar<T, m>(std::numeric_limits<T>::infinity())));
-    BOOST_CHECK(isnan(make_fvar<T, m>(std::numeric_limits<T>::quiet_NaN())));
-  }
-}
-
 BOOST_AUTO_TEST_SUITE_END()

test/test_classify.cpp

@@ -16,6 +16,8 @@
 #include <iostream>
 #include <iomanip>
 
+#include "test_autodiff.hpp"
+
 #ifdef _MSC_VER
 #pragma warning(disable: 4127 4146) //  conditional expression is constant
 #endif
@@ -256,6 +258,9 @@ void test_classify(T t, const char* type)
 #endif
 }
 
+
+BOOST_AUTO_TEST_SUITE(test_fpclassify)
+
 BOOST_AUTO_TEST_CASE( test_main )
 {
    BOOST_MATH_CONTROL_FP;
@@ -288,6 +293,16 @@ BOOST_AUTO_TEST_CASE( test_main )
    test_classify(unsigned(0), "unsigned");
 }
 
+BOOST_AUTO_TEST_CASE_TEMPLATE(fpclassify_autodiff, T, all_float_types) {
+   test_classify(boost::math::differentiation::make_fvar<T, 1>(0), "autodiff float");
+   test_classify(boost::math::differentiation::make_fvar<T, 2>(0), "autodiff float");
+   test_classify(boost::math::differentiation::make_fvar<T, 3>(0), "autodiff float");
+   test_classify(boost::math::differentiation::make_fvar<T, 7>(0), "autodiff float");
+   test_classify(boost::math::differentiation::make_fvar<T, 12>(0), "autodiff float");
+}
+
+BOOST_AUTO_TEST_SUITE_END()
+
 /*
 Autorun "i:\Boost-sandbox\math_toolkit\libs\math\test\MSVC80\debug\test_classify.exe"
 Running 1 test case...