diff --git a/double-conversion/bignum-dtoa.h b/double-conversion/bignum-dtoa.h
index 92455ec..34b9619 100644
--- a/double-conversion/bignum-dtoa.h
+++ b/double-conversion/bignum-dtoa.h
@@ -28,7 +28,7 @@
 #ifndef DOUBLE_CONVERSION_BIGNUM_DTOA_H_
 #define DOUBLE_CONVERSION_BIGNUM_DTOA_H_
 
-#include "double-conversion/utils.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/bignum.cc b/double-conversion/bignum.cc
index 7c88ff7..8892de8 100644
--- a/double-conversion/bignum.cc
+++ b/double-conversion/bignum.cc
@@ -25,8 +25,8 @@
 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-#include "double-conversion/bignum.h"
-#include "double-conversion/utils.h"
+#include "bignum.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/bignum.h b/double-conversion/bignum.h
index 57e4230..c385f22 100644
--- a/double-conversion/bignum.h
+++ b/double-conversion/bignum.h
@@ -28,7 +28,7 @@
 #ifndef DOUBLE_CONVERSION_BIGNUM_H_
 #define DOUBLE_CONVERSION_BIGNUM_H_
 
-#include "double-conversion/utils.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/cached-powers.cc b/double-conversion/cached-powers.cc
index aa078c8..2b43f06 100644
--- a/double-conversion/cached-powers.cc
+++ b/double-conversion/cached-powers.cc
@@ -29,7 +29,7 @@
 #include <limits.h>
 #include <math.h>
 
-#include "double-conversion/utils.h"
+#include "utils.h"
 
 #include "cached-powers.h"
 
diff --git a/double-conversion/diy-fp.cc b/double-conversion/diy-fp.cc
index 60b805e..ddd1891 100644
--- a/double-conversion/diy-fp.cc
+++ b/double-conversion/diy-fp.cc
@@ -26,8 +26,8 @@
 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 
-#include "double-conversion/diy-fp.h"
-#include "double-conversion/utils.h"
+#include "diy-fp.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/diy-fp.h b/double-conversion/diy-fp.h
index a0c0c15..2edf346 100644
--- a/double-conversion/diy-fp.h
+++ b/double-conversion/diy-fp.h
@@ -28,7 +28,7 @@
 #ifndef DOUBLE_CONVERSION_DIY_FP_H_
 #define DOUBLE_CONVERSION_DIY_FP_H_
 
-#include "double-conversion/utils.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/double-conversion.cc b/double-conversion/double-conversion.cc
index a6fc0de..d82150b 100644
--- a/double-conversion/double-conversion.cc
+++ b/double-conversion/double-conversion.cc
@@ -28,14 +28,14 @@
 #include <limits.h>
 #include <math.h>
 
-#include "double-conversion/double-conversion.h"
+#include "double-conversion.h"
 
-#include "double-conversion/bignum-dtoa.h"
-#include "double-conversion/fast-dtoa.h"
-#include "double-conversion/fixed-dtoa.h"
-#include "double-conversion/ieee.h"
-#include "double-conversion/strtod.h"
-#include "double-conversion/utils.h"
+#include "bignum-dtoa.h"
+#include "fast-dtoa.h"
+#include "fixed-dtoa.h"
+#include "ieee.h"
+#include "strtod.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/double-conversion.h b/double-conversion/double-conversion.h
index 11fbcb8..6bdfa8d 100644
--- a/double-conversion/double-conversion.h
+++ b/double-conversion/double-conversion.h
@@ -28,7 +28,7 @@
 #ifndef DOUBLE_CONVERSION_DOUBLE_CONVERSION_H_
 #define DOUBLE_CONVERSION_DOUBLE_CONVERSION_H_
 
-#include "double-conversion/utils.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/fast-dtoa.h b/double-conversion/fast-dtoa.h
index 259ee3a..5f1e8ee 100644
--- a/double-conversion/fast-dtoa.h
+++ b/double-conversion/fast-dtoa.h
@@ -28,7 +28,7 @@
 #ifndef DOUBLE_CONVERSION_FAST_DTOA_H_
 #define DOUBLE_CONVERSION_FAST_DTOA_H_
 
-#include "double-conversion/utils.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/fixed-dtoa.h b/double-conversion/fixed-dtoa.h
index 43de94e..3bdd08e 100644
--- a/double-conversion/fixed-dtoa.h
+++ b/double-conversion/fixed-dtoa.h
@@ -28,7 +28,7 @@
 #ifndef DOUBLE_CONVERSION_FIXED_DTOA_H_
 #define DOUBLE_CONVERSION_FIXED_DTOA_H_
 
-#include "double-conversion/utils.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/strtod.h b/double-conversion/strtod.h
index 00cf8a7..ed0293b 100644
--- a/double-conversion/strtod.h
+++ b/double-conversion/strtod.h
@@ -28,7 +28,7 @@
 #ifndef DOUBLE_CONVERSION_STRTOD_H_
 #define DOUBLE_CONVERSION_STRTOD_H_
 
-#include "double-conversion/utils.h"
+#include "utils.h"
 
 namespace double_conversion {
 
diff --git a/double-conversion/utils.h b/double-conversion/utils.h
index ce41e62..51d5e61 100644
--- a/double-conversion/utils.h
+++ b/double-conversion/utils.h
@@ -39,9 +39,23 @@
 #ifndef UNIMPLEMENTED
 #define UNIMPLEMENTED() (abort())
 #endif
+#ifndef DOUBLE_CONVERSION_NO_RETURN
+#ifdef _MSC_VER
+#define DOUBLE_CONVERSION_NO_RETURN __declspec(noreturn)
+#else
+#define DOUBLE_CONVERSION_NO_RETURN __attribute__((noreturn))
+#endif
+#endif
 #ifndef UNREACHABLE
+#ifdef _MSC_VER
+void DOUBLE_CONVERSION_NO_RETURN abort_noreturn();
+inline void abort_noreturn() { abort(); }
+#define UNREACHABLE()   (abort_noreturn())
+#else
 #define UNREACHABLE()   (abort())
 #endif
+#endif
+
 
 // Double operations detection based on target architecture.
 // Linux uses a 80bit wide floating point stack on x86. This induces double