Fix compilation on ppc64

include/fmt/format.h

@@ -755,6 +755,9 @@ struct is_fast_float : bool_constant<std::numeric_limits<T>::is_iec559 &&
                                      sizeof(T) <= sizeof(double)> {};
 template <typename T> struct is_fast_float<T, false> : std::false_type {};
 
+template <typename T>
+using is_double_double = bool_constant<std::numeric_limits<T>::digits == 106>;
+
 #ifndef FMT_USE_FULL_CACHE_DRAGONBOX
 #  define FMT_USE_FULL_CACHE_DRAGONBOX 0
 #endif
@@ -1314,15 +1317,9 @@ struct float_info<T, enable_if_t<std::numeric_limits<T>::digits == 64 ||
   static const int exponent_bits = 15;
 };
 
-template <int> constexpr int check_digits();
-
 // A double-double floating point number.
 template <typename T>
-struct float_info<T, enable_if_t<!(std::numeric_limits<T>::digits == 64 ||
-                                   std::numeric_limits<T>::digits == 113 ||
-                                   is_float128<T>::value) &&
-                                 std::is_same<T, long double>::value>> {
-  static constexpr int check = check_digits<std::numeric_limits<T>::digits>();
+struct float_info<T, enable_if_t<is_double_double<T>::value>> {
   using carrier_uint = detail::uint128_t;
 };
 
@@ -1400,8 +1397,7 @@ template <typename F> struct basic_fp {
   template <typename Float> FMT_CONSTEXPR basic_fp(Float n) { assign(n); }
 
   // Assigns n to this and return true iff predecessor is closer than successor.
-  template <typename Float,
-            FMT_ENABLE_IF(std::numeric_limits<Float>::is_iec559)>
+  template <typename Float, FMT_ENABLE_IF(!is_double_double<Float>::value)>
   FMT_CONSTEXPR auto assign(Float n) -> bool {
     static_assert(std::numeric_limits<Float>::digits <= 113, "unsupported FP");
     // Assume Float is in the format [sign][exponent][significand].
@@ -1426,8 +1422,7 @@ template <typename F> struct basic_fp {
     return is_predecessor_closer;
   }
 
-  template <typename Float,
-            FMT_ENABLE_IF(!std::numeric_limits<Float>::is_iec559)>
+  template <typename Float, FMT_ENABLE_IF(is_double_double<Float>::value)>
   FMT_CONSTEXPR auto assign(Float n) -> bool {
     static_assert(std::numeric_limits<double>::is_iec559, "unsupported FP");
     return assign(static_cast<double>(n));

test/format-impl-test.cc

@@ -379,17 +379,17 @@ bool operator>=(const double_double& lhs, const double_double& rhs) {
 namespace std {
 template <> struct is_floating_point<double_double> : std::true_type {};
 template <> struct numeric_limits<double_double> {
-  static constexpr bool is_iec559 = false;
+  // is_iec559 is true for double-double in libstdc++.
+  static constexpr bool is_iec559 = true;
   static constexpr int digits = 106;
 };
 }  // namespace std
 
 TEST(format_impl_test, write_double_double) {
-  // TODO: restore
-  // auto s = std::string();
-  // fmt::detail::write<char>(std::back_inserter(s), double_double(42), {});
+  auto s = std::string();
+  fmt::detail::write<char>(std::back_inserter(s), double_double(42), {});
 #ifndef _MSC_VER  // MSVC has an issue with specializing is_floating_point.
-  // EXPECT_EQ(s, "42");
+  EXPECT_EQ(s, "42");
 #endif
 }
 

test/format-test.cc

@@ -1984,7 +1984,7 @@ TEST(format_test, to_string) {
   EXPECT_EQ(fmt::to_string(zero), "0");
 
 #if FMT_USE_FLOAT128
-  EXPECT_EQ(fmt::to_string(__float128(0.42)), "0.42");
+  EXPECT_EQ(fmt::to_string(__float128(0.5)), "0.5");
 #endif
 }