Repository navigation
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #19904 +/- ##
==========================================
+ Coverage 98.57% 98.60% +0.03%
==========================================
Files 182 182
Lines 23378 23380 +2
Branches 5 5
==========================================
+ Hits 23044 23054 +10
+ Misses 333 325 -8
Partials 1 1
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
|
Code size report: |
0xABCD01
force-pushed
the
fix/19874-int-subclass-float
branch
from
November 7, 2026 12:05
0eae561 to
9024ae7
Compare
Contributor
|
There appears to be some room left for optimization A quick experiment diff --git a/py/objint_longlong.c b/py/objint_longlong.c
--- a/py/objint_longlong.c
+++ b/py/objint_longlong.c
@@ -107,10 +107,6 @@ mp_obj_t mp_obj_int_unary_op(mp_unary_op_t op, mp_obj_t o_in) {
}
case MP_UNARY_OP_INT_MAYBE:
return o_in;
- #if MICROPY_PY_BUILTINS_FLOAT
- case MP_UNARY_OP_FLOAT_MAYBE:
- return mp_obj_new_float(mp_obj_int_as_float_impl(o_in));
- #endif
default:
return MP_OBJ_NULL; // op not supported
}
diff --git a/py/objint_mpz.c b/py/objint_mpz.c
--- a/py/objint_mpz.c
+++ b/py/objint_mpz.c
@@ -161,10 +161,6 @@ mp_obj_t mp_obj_int_unary_op(mp_unary_op_t op, mp_obj_t o_in) {
}
case MP_UNARY_OP_INT_MAYBE:
return o_in;
- #if MICROPY_PY_BUILTINS_FLOAT
- case MP_UNARY_OP_FLOAT_MAYBE:
- return mp_obj_new_float(mp_obj_int_as_float_impl(o_in));
- #endif
default:
return MP_OBJ_NULL; // op not supported
}
diff --git a/py/objtype.c b/py/objtype.c
--- a/py/objtype.c
+++ b/py/objtype.c
@@ -433,6 +433,11 @@ static mp_obj_t instance_unary_op(mp_unary_op_t op, mp_obj_t self_in) {
};
mp_obj_class_lookup(&lookup, self->base.type);
if (member[0] == MP_OBJ_SENTINEL) {
+ #if MICROPY_PY_BUILTINS_FLOAT
+ if (op == MP_UNARY_OP_FLOAT_MAYBE && mp_obj_is_int(self->subobj[0])) {
+ return mp_obj_new_float(mp_obj_get_float(self->subobj[0]));
+ }
+ #endif
return mp_unary_op(op, self->subobj[0]);
} else if (member[0] != MP_OBJ_NULL) {
mp_obj_t val = mp_call_function_1(member[0], self_in);
diff --git a/py/runtime.c b/py/runtime.c
--- a/py/runtime.c
+++ b/py/runtime.c
@@ -314,10 +314,6 @@ mp_obj_t mp_unary_op(mp_unary_op_t op, mp_obj_t arg) {
case MP_UNARY_OP_POSITIVE:
case MP_UNARY_OP_INT_MAYBE:
return arg;
- #if MICROPY_PY_BUILTINS_FLOAT
- case MP_UNARY_OP_FLOAT_MAYBE:
- return mp_obj_new_float((mp_float_t)val);
- #endif
case MP_UNARY_OP_NEGATIVE:
// check for overflow
if (val == MP_SMALL_INT_MIN) {Cost ~48 bytes on the Unix port. Also the changed longlong backend — I usually prefer the code in a seperate commit from the tests, |
float() on an instance of an int subclass raised TypeError, because mp_obj_get_float_maybe() only recognises small ints and objects of exact type int, and its MP_UNARY_OP_FLOAT_MAYBE fallback could not resolve for a subclass instance: instance_unary_op() delegates the op to the instance's native sub-object, but nothing converted the sub-object when it was an int. Handle MP_UNARY_OP_FLOAT_MAYBE in instance_unary_op(), at the point where the op is delegated to the native sub-object: if that sub-object is an int, convert it directly. A __float__ method defined by the subclass still takes precedence, because the member lookup finds it before this delegation happens. This also fixes complex(), the math module functions, float arithmetic and % formatting with int subclass instances, which all convert through mp_obj_get_float_maybe(). Signed-off-by: Hafiar <[email protected]>
Covers float(), complex(), math functions, string formatting and float arithmetic with int subclass instances, for small-int values, big-int values, and int64-range heap int values (the latter also runs with the longlong long-int implementation). Signed-off-by: Hafiar <[email protected]>
0xABCD01
force-pushed
the
fix/19874-int-subclass-float
branch
from
November 9, 2026 04:08
9024ae7 to
627b834
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #19874.
float()raisesTypeError: can't convert MyInt to floatfor instances of anintsubclass.complex(),math.sqrt(),'%f'formatting and arithmetic with a float operand fail the same way. CPython accepts all of these.An
intsubclass instance is a generic instance object wrapping the int value, so the float conversion path does not recognise it: theFLOAT_MAYBEunary op is not implemented for the wrapped int, while the matchingINT_MAYBEop is — which is whyint()already works on the same objects.This change implements
MP_UNARY_OP_FLOAT_MAYBEnext to the existingMP_UNARY_OP_INT_MAYBEcases inpy/runtime.c(small ints),py/objint_mpz.candpy/objint_longlong.c. A subclass that defines its own__float__still takes precedence, because conversion still goes through the unary-op dispatch on the instance first.Tests: new tests in
tests/float/(int_subclass_convert.py,int_subclass_convert_intbig.py,complex_int_subclass.py). run-tests results: basics 552/552, float 72/72, extmod 190/190, micropython 104/104, misc 7/7, cmdline 25/25.Disclosure: I used AI assistance to investigate this issue and prepare this change, and I reviewed the diff and test results before submitting.