py/objfloat: Fix binary ops with incompatible objects.

These are now returned as "operation not supported" instead of raising
TypeError. In particular, this fixes equality for float vs incompatible
types, which now properly results in False instead of exception. This
also paves the road to support reverse operation (e.g. __radd__) with
float objects.

This is achieved by introducing mp_obj_get_float_maybe(), similar to
existing mp_obj_get_int_maybe().

py/obj.c

@@ -264,20 +264,33 @@ bool mp_obj_get_int_maybe(mp_const_obj_t arg, mp_int_t *value) {
 }
 
 #if MICROPY_PY_BUILTINS_FLOAT
-mp_float_t mp_obj_get_float(mp_obj_t arg) {
+bool mp_obj_get_float_maybe(mp_obj_t arg, mp_float_t *value) {
+    mp_float_t val;
+
     if (arg == mp_const_false) {
-        return 0;
+        val = 0;
     } else if (arg == mp_const_true) {
-        return 1;
+        val = 1;
     } else if (MP_OBJ_IS_SMALL_INT(arg)) {
-        return MP_OBJ_SMALL_INT_VALUE(arg);
+        val = MP_OBJ_SMALL_INT_VALUE(arg);
     #if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
     } else if (MP_OBJ_IS_TYPE(arg, &mp_type_int)) {
-        return mp_obj_int_as_float_impl(arg);
+        val = mp_obj_int_as_float_impl(arg);
     #endif
     } else if (mp_obj_is_float(arg)) {
-        return mp_obj_float_get(arg);
+        val = mp_obj_float_get(arg);
     } else {
+        return false;
+    }
+
+    *value = val;
+    return true;
+}
+
+mp_float_t mp_obj_get_float(mp_obj_t arg) {
+    mp_float_t val;
+
+    if (!mp_obj_get_float_maybe(arg, &val)) {
         if (MICROPY_ERROR_REPORTING == MICROPY_ERROR_REPORTING_TERSE) {
             mp_raise_TypeError("can't convert to float");
         } else {
@@ -285,6 +298,8 @@ mp_float_t mp_obj_get_float(mp_obj_t arg) {
                 "can't convert %s to float", mp_obj_get_type_str(arg)));
         }
     }
+
+    return val;
 }
 
 #if MICROPY_PY_BUILTINS_COMPLEX

py/obj.h

@@ -686,6 +686,7 @@ mp_int_t mp_obj_get_int_truncated(mp_const_obj_t arg);
 bool mp_obj_get_int_maybe(mp_const_obj_t arg, mp_int_t *value);
 #if MICROPY_PY_BUILTINS_FLOAT
 mp_float_t mp_obj_get_float(mp_obj_t self_in);
+bool mp_obj_get_float_maybe(mp_obj_t arg, mp_float_t *value);
 void mp_obj_get_complex(mp_obj_t self_in, mp_float_t *real, mp_float_t *imag);
 #endif
 //qstr mp_obj_get_qstr(mp_obj_t arg);

py/objfloat.c

@@ -240,7 +240,11 @@ STATIC void mp_obj_float_divmod(mp_float_t *x, mp_float_t *y) {
 }
 
 mp_obj_t mp_obj_float_binary_op(mp_binary_op_t op, mp_float_t lhs_val, mp_obj_t rhs_in) {
-    mp_float_t rhs_val = mp_obj_get_float(rhs_in); // can be any type, this function will convert to float (if possible)
+    mp_float_t rhs_val;
+    if (!mp_obj_get_float_maybe(rhs_in, &rhs_val)) {
+        return MP_OBJ_NULL; // op not supported
+    }
+
     switch (op) {
         case MP_BINARY_OP_ADD:
         case MP_BINARY_OP_INPLACE_ADD: lhs_val += rhs_val; break;

tests/float/float_compare.py

@@ -0,0 +1,22 @@
+# Extended float comparisons
+
+class Foo:
+    pass
+
+foo = Foo()
+
+print(foo == 1.0)
+print(1.0 == foo)
+print(1.0 == Foo)
+print(1.0 == [])
+print(1.0 == {})
+
+try:
+    print(foo < 1.0)
+except TypeError:
+    print("TypeError")
+
+try:
+    print(1.0 < foo)
+except TypeError:
+    print("TypeError")