static inline bool fixed_point_ne(fixed_point x, fixed_point y) { return x.data != y.data; }
#define FIXED_POINT_I(x, y) { ((x) * 256) + ((y) * 256 / 1000) }
+#define FIXED_INT_I(x) FIXED_POINT_I(x, 0)
-static inline fixed_point FIXED_POINT(unsigned x, unsigned y) {
- fixed_point r = { ((int) x * 256) + ((int) y * 256 / 1000) };
+static inline fixed_point FIXED_POINT(int32_t x, int32_t y) {
+ fixed_point r = FIXED_POINT_I(x, y);
return r;
}
-static inline int fixed_point_cast_int(fixed_point x) { return x.data / 256; }
+static inline fixed_point FIXED_INT(int32_t x) { return FIXED_POINT(x, 0); }
+
+// sign bit is shifted in if x.data < 0, so this is x.data / 256 - (x.data < 0).
+// This means 0.123 is cast to 1, which is what we want when we cast a model coordinate
+// to a screen coordinate.
+static inline int fixed_point_cast_int(fixed_point x) { return x.data >> 8; }
static inline fixed_point fixed_point_min(fixed_point x, fixed_point y) { return fixed_point_lt(x, y) ? x : y; }
static inline fixed_point fixed_point_max(fixed_point x, fixed_point y) { return fixed_point_gt(x, y) ? x : y; }