Round float at the decimal level to match CPython's _Py_dg_dtoa by changjoon-park · Pull Request #7761 · RustPython/RustPython

changjoon-park · 2026-05-02T10:23:59Z

Summary

CPython's float.__round__ (Objects/floatobject.c) routes through _Py_dg_dtoa and rounds at the decimal level. The previous round_float_digits multiplied by 10**ndigits and rounded at the IEEE 754 binary level, which diverges for values that aren't exactly representable.

>>> round(2.675, 2)
# CPython 3.14.4: 2.67
# RustPython main: 2.68 ❌

2.675 stores as 2.67499...; dtoa correctly rounds it down to 2.67, but (2.675 * 100.0).round() / 100.0 lands on 2.68 because the multiplication produces a phantom 267.5 tie that round-half-to-even snaps up.

Rust's {:.*} float formatting uses dtoa-style algorithms (Grisu3 + Dragon4 fallback) and matches CPython's _Py_dg_dtoa byte-for-byte. Replacing the multiply-then-round step with format!("{:.*}", ndigits, x).parse() removes the entire phantom-tie class for ndigits >= 0.

For ndigits < 0 the existing divide-then-round path is preserved because dividing typical inputs by 10**|ndigits| produces genuine half-integer ties rather than synthesizing them.

Verification (CPython 3.14.4)

Probe set	Cases	Match
Targeted half-tie probes (positive ndigits)	34	34/34
Negative ndigits regression	17	17/17
Special values (NaN, ±Inf, ±0.0, NDIGITS bounds)	12	12/12
No-arg `round(x)` (different code path)	10	10/10
Subnormal / denormal	6	6/6
Random fuzz (mantissa × ndigits ∈ {-5..50})	4332	4332/4332

Tests unmasked

Lib/test/test_float.py::RoundTestCase:

test_matches_float_format — rounds against float(format(x, '.{n}f')) across many values; this is exactly what the fix does internally.
test_previous_round_bugs — explicit "round-half-even" cases (e.g. round(25.0, -1) == 20.0).

Test runs

cargo run --release -- -m test test_float test_builtin test_decimal \
    test_fractions test_complex test_math test_format test_long test_int \
    test_statistics test_re

1573 tests pass, 0 regressions. All 188 extra_tests/snippets/*.py pass under the CI feature set (stdlib,importlib,stdio,encodings,sqlite,ssl-rustls,host_env).

A regression block is added to extra_tests/snippets/builtin_round.py.

Summary by CodeRabbit

Bug Fixes
- Improved rounding to more closely match CPython’s behavior: better handling of halfway ties (banker’s rounding), negative ndigits, signed zeros, and non-finite inputs.
Tests
- Expanded rounding tests to cover banker's rounding, negative ndigits, exact-half cases, NaN/infinity behavior, signed zero, and out-of-range ndigits.

CPython's `float.__round__` (Objects/floatobject.c) routes through `_Py_dg_dtoa` and rounds at the decimal level. The previous `round_float_digits` multiplied by 10**ndigits and rounded at the IEEE 754 binary level, which diverges for values that aren't exactly representable. For example, 2.675 stores as 2.67499...; dtoa correctly rounds it down to 2.67, but `(2.675 * 100.0).round() / 100.0` lands on 2.68 because the multiplication produces a phantom 267.5 tie that round-half-to-even snaps up. Rust's `{:.*}` float formatting uses dtoa-style algorithms (Grisu3 + Dragon4 fallback) and matches CPython's `_Py_dg_dtoa` byte-for-byte. Replace the multiply-then-round step with `format!` + `parse` for ndigits >= 0. The ndigits < 0 path is unchanged because dividing typical inputs by 10**|ndigits| produces genuine ties rather than synthesizing them. Verified byte-identical with CPython 3.14.4 over a 108-case random fuzz plus targeted half-tie probes. Unmasks `test_float.RoundTestCase.test_matches_float_format` and `test_previous_round_bugs`.

coderabbitai · 2026-05-02T10:24:14Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yml

Review profile: CHILL

Plan: Pro

Run ID: 741c2592-e9ac-4826-81f0-ef8d550e2a49

📥 Commits

Reviewing files that changed from the base of the PR and between 2310495 and f039f28.

📒 Files selected for processing (1)

crates/common/src/float_ops.rs

✅ Files skipped from review due to trivial changes (1)

crates/common/src/float_ops.rs

📝 Walkthrough

Walkthrough

Rewrites round_float_digits() to mirror CPython: early non-finite returns, clamps ndigits, uses format+parse for ndigits >= 0, divide-then-round with explicit 0.5 tie correction for ndigits < 0, and validates finiteness of the result; adds tests exercising banker's rounding and edge cases.

Changes

Float Rounding Implementation

Layer / File(s)	Summary
Imports `crates/common/src/float_ops.rs`	Removed unused `Float` and `Zero` from `num_traits` imports.
Behavioral/Data Shape `crates/common/src/float_ops.rs`	Added early return for non-finite inputs and computed `NDIGITS_MAX`/`NDIGITS_MIN` bounds for clamping `ndigits`.
Core Implementation `crates/common/src/float_ops.rs`	Replaced previous pow-splitting/overflow-avoidance and special-case `ndigits == 0` logic. For `ndigits >= 0` use `format!("{:.*}", ndigits, x)` then `parse()` back to `f64`; for `ndigits < 0` use divide-then-round via `powi` with explicit 0.5 tie handling (banker's rounding). Final result is checked for finiteness and mapped to `None` if non-finite.
Tests / Examples `extra_tests/snippets/builtin_round.py`	Added `import math` and ~50 assertions covering decimal banker's rounding (including IEEE-754 artifacts), exact-halfway ties, negative `ndigits`, NaN/infinity behavior with `ndigits`, signed zero preservation, and out-of-range `ndigits` short-circuit cases.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

Fix process abort on large float format precision #7633: Adds precision-clamping helpers for runtime precision in format!("{:.*}", ...); relevant because this PR newly uses runtime-format precision for ndigits >= 0.

Suggested reviewers

ShaharNaveh
youknowone

Poem

🐇 I hop on floats both near and far,

I nudge the digits, round like a star —
Banker's balance, ties resolved clean,
Signed zeros kept, and NaN sits serene.
Hop, hop, let's ship this rounding scene!

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title directly describes the main change: updating float rounding to match CPython's behavior using decimal-level rounding via _Py_dg_dtoa.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Review rate limit: 7/8 reviews remaining, refill in 7 minutes and 30 seconds.}

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

github-actions · 2026-05-02T10:24:47Z

📦 Library Dependencies

The following Lib/ modules were modified. Here are their dependencies:

[x] test: cpython/Lib/test/test_float.py (TODO: 4)
[x] test: cpython/Lib/test/test_strtod.py (TODO: 6)

dependencies:

dependent tests: (no tests depend on float)

Legend:

[+] path exists in CPython
[x] up-to-date, [ ] outdated

ShaharNaveh

overall lgtm!
I had a minor nitpick

tysm for the patches:)

ShaharNaveh · 2026-05-02T10:47:02Z

+        let pow1 = 10.0f64.powi(-ndigits);
+        let y = x / pow1;
+        let z = y.round();
+        #[allow(clippy::float_cmp)]


Suggested change

#[allow(clippy::float_cmp)]

#[expect(clippy::float_cmp, reason = "...")]

Can you please change it to expect and add a reason to this cfg?

good catch i'll keep this convention following prs too thanks !

Co-authored-by: ShaharNaveh <[email protected]>

ShaharNaveh

tysm:)

changjoon-park · 2026-05-02T11:13:40Z

tysm:)

thanks for the quick review !!

ShaharNaveh requested changes May 2, 2026

View reviewed changes

Use #[expect] with reason for float_cmp suppression

f039f28

Co-authored-by: ShaharNaveh <[email protected]>

ShaharNaveh approved these changes May 2, 2026

View reviewed changes

youknowone approved these changes May 2, 2026

View reviewed changes

youknowone merged commit c2910c0 into RustPython:main May 2, 2026
21 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Round float at the decimal level to match CPython's _Py_dg_dtoa#7761

Round float at the decimal level to match CPython's _Py_dg_dtoa#7761
youknowone merged 2 commits into
RustPython:mainfrom
changjoon-park:fix-round-float-digits-dtoa

changjoon-park commented May 2, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented May 2, 2026 •

edited

Loading

Walkthrough

Changes

Estimated code review effort

Possibly related PRs

Suggested reviewers

Poem

❌ Failed checks (1 warning)

Uh oh!

github-actions Bot commented May 2, 2026 •

edited

Loading

Uh oh!

ShaharNaveh left a comment

Uh oh!

ShaharNaveh May 2, 2026

Uh oh!

changjoon-park May 2, 2026

Uh oh!

ShaharNaveh left a comment

Uh oh!

changjoon-park commented May 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

	#[allow(clippy::float_cmp)]
	#[expect(clippy::float_cmp, reason = "...")]

Conversation

changjoon-park commented May 2, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verification (CPython 3.14.4)

Tests unmasked

Test runs

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented May 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Estimated code review effort

Possibly related PRs

Suggested reviewers

Poem

❌ Failed checks (1 warning)

Uh oh!

github-actions Bot commented May 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📦 Library Dependencies

Uh oh!

ShaharNaveh left a comment

Choose a reason for hiding this comment

Uh oh!

ShaharNaveh May 2, 2026

Choose a reason for hiding this comment

Uh oh!

changjoon-park May 2, 2026

Choose a reason for hiding this comment

Uh oh!

ShaharNaveh left a comment

Choose a reason for hiding this comment

Uh oh!

changjoon-park commented May 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

changjoon-park commented May 2, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented May 2, 2026 •

edited

Loading

github-actions Bot commented May 2, 2026 •

edited

Loading