Skip to content

Fix integer-to-float overflow rounding boundary #9584

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
youknowone merged 2 commits into from
Nov 4, 2026
Merged

Fix integer-to-float overflow rounding boundary #9584

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
dea6c93
Fix integer-to-float overflow rounding boundary
youknowdot Nov 3, 2026
4fd589a
Retire passing float conversion expected failure
youknowdot Nov 3, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion Lib/test/test_long.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -403,7 +403,6 @@ def check_float_conversion(self, n):
"Got {}, expected {}.".format(n, actual, expected))
self.assertEqual(actual, expected, msg)

@unittest.expectedFailure # TODO: RUSTPYTHON
@support.requires_IEEE_754
def test_float_conversion(self):

Expand Down
20 changes: 11 additions & 9 deletions crates/common/src/int.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,13 +6,10 @@ use num_traits::{One, ToPrimitive, Zero};
#[must_use]
pub fn true_div(numerator: &BigInt, denominator: &BigInt) -> f64 {
let rational = Rational::from_integers_ref(numerator.into(), denominator.into());
match rational.rounding_into(RoundingMode::Nearest) {
// returned value is $t::MAX but still less than the original
(val, core::cmp::Ordering::Less) if val == f64::MAX => f64::INFINITY,
// returned value is $t::MIN but still greater than the original
(val, core::cmp::Ordering::Greater) if val == f64::MIN => f64::NEG_INFINITY,
(val, _) => val,
}
// Nearest rounding can return the largest finite value for a quotient
// slightly beyond it; overflow starts at the halfway rounding boundary.
let (rounded, _) = rational.rounding_into(RoundingMode::Nearest);
rounded
}

#[must_use]
Expand Down Expand Up @@ -202,11 +199,16 @@ fn word_all_ascii_digits(word: u64) -> bool {
(below | above) & 0x8080_8080_8080_8080 == 0
}

// num-bigint now returns Some(inf) for to_f64() in some cases, so just keep that the same for now
/// Convert an integer to a finite float, or return None when it overflows.
#[inline(always)]
#[must_use]
pub fn bigint_to_finite_float(int: &BigInt) -> Option<f64> {
int.to_f64().filter(|f| f.is_finite())
// malachite-bigint's ToPrimitive returns None when the input exceeds
// the largest finite magnitude, even when nearest rounding is finite.
let value = int
.to_f64()
.unwrap_or_else(|| true_div(int, &BigInt::one()));
value.is_finite().then_some(value)
}

#[cfg(test)]
Expand Down
Loading