+/* Floating-point network-type support.
+ These functions must convert to/from a 32-bit big-endian integer that follows
+ the layout of Java's java.lang.float.floatToRawIntBits method.
+ Conveniently, for the AVR family, this is a straight byte copy...
+*/
+
+typedef float nx_float __attribute__((nx_base_be(afloat)));
+
+inline float __nesc_ntoh_afloat(const void *COUNT(sizeof(float)) source) @safe() {
+ float f;
+ memcpy(&f, source, sizeof(float));
+ return f;
+}
+
+inline float __nesc_hton_afloat(void *COUNT(sizeof(float)) target, float value) @safe() {
+ memcpy(target, &value, sizeof(float));
+ return value;
+}