Leave decltype(max_ulps_) alone and cast, not sure this is better

googlemock/test/gmock-matchers_test.cc

@@ -2708,19 +2708,21 @@ class FloatingPointTest : public testing::Test {
         zero_bits_(Floating(0).bits()),
         one_bits_(Floating(1).bits()),
         infinity_bits_(Floating(Floating::Infinity()).bits()),
-        close_to_positive_zero_(
-            Floating::ReinterpretBits(zero_bits_ + max_ulps_/2)),
-        close_to_negative_zero_(
-            -Floating::ReinterpretBits(zero_bits_ + max_ulps_ - max_ulps_/2)),
+        close_to_positive_zero_(Floating::ReinterpretBits(
+            static_cast<Bits>(zero_bits_ + max_ulps_/2))),
+        close_to_negative_zero_(-Floating::ReinterpretBits(
+            static_cast<Bits>(zero_bits_ + max_ulps_ - max_ulps_/2))),
         further_from_negative_zero_(-Floating::ReinterpretBits(
-            zero_bits_ + max_ulps_ + 1 - max_ulps_/2)),
-        close_to_one_(Floating::ReinterpretBits(one_bits_ + max_ulps_)),
-        further_from_one_(Floating::ReinterpretBits(one_bits_ + max_ulps_ + 1)),
+            static_cast<Bits>(zero_bits_ + max_ulps_ + 1 - max_ulps_/2))),
+        close_to_one_(Floating::ReinterpretBits(
+            static_cast<Bits>(one_bits_ + max_ulps_))),
+        further_from_one_(Floating::ReinterpretBits(
+            static_cast<Bits>(one_bits_ + max_ulps_ + 1))),
         infinity_(Floating::Infinity()),
-        close_to_infinity_(
-            Floating::ReinterpretBits(infinity_bits_ - max_ulps_)),
-        further_from_infinity_(
-            Floating::ReinterpretBits(infinity_bits_ - max_ulps_ - 1)),
+        close_to_infinity_(Floating::ReinterpretBits(
+            static_cast<Bits>(infinity_bits_ - max_ulps_))),
+        further_from_infinity_(Floating::ReinterpretBits(
+            static_cast<Bits>(infinity_bits_ - max_ulps_ - 1))),
         max_(Floating::Max()),
         nan1_(Floating::ReinterpretBits(Floating::kExponentBitMask | 1)),
         nan2_(Floating::ReinterpretBits(Floating::kExponentBitMask | 200)) {
@@ -2778,7 +2780,7 @@ class FloatingPointTest : public testing::Test {
 
   // Pre-calculated numbers to be used by the tests.
 
-  const unsigned int max_ulps_;
+  const size_t max_ulps_;
 
   const Bits zero_bits_;  // The bits that represent 0.0.
   const Bits one_bits_;  // The bits that represent 1.0.