FIX: snap near-integer arc windings to a full circle on polar plots (#20388, #26972) by SharadhNaidu · Pull Request #31844 · matplotlib/matplotlib

SharadhNaidu · 2026-06-05T19:30:24Z

PR summary

On polar plots, radial gridlines (#20388) and the outer spine (#26972) could collapse to a near-empty arc for certain set_theta_zero_location / set_theta_offset / set_theta_direction combinations.

The root cause is in Path.arc's angle-unwrap step. A span of 360*n plus a tiny floating-point rounding error was reduced modulo 360 to a near-zero arc instead of a full circle:

eta2 = theta2 - 360 * np.floor((theta2 - theta1) / 360)   # 360 + 1e-9 -> 1e-9

Fix

Detect spans that are within floating-point tolerance of a whole number of turns and draw a complete circle; otherwise unwrap theta2 to the shortest arc within 360 degrees exactly as before:

n_turns = (theta2 - theta1) / 360
if (theta2 != theta1 and round(n_turns) >= 1
        and abs(n_turns - round(n_turns)) < 1e-9):
    eta2 = theta1 + 360            # complete circle
else:
    eta2 = theta2 - 360 * np.floor(n_turns)   # shortest arc

The change is byte-identical for all non-degenerate inputs (e.g. 90°, 359.99°, 360°, 410°→50°, 540°→180°, 720°→full circle) and only affects the 360*n + ε collapse cases, so no call sites need to change.

Closes #20388
Closes #26972

Tests

test_arc_full_circle_snap / test_arc_unwrap_partial_turn in test_path.py cover the snap-vs-unwrap behaviour at the unit level.
test_path.py and end-to-end polar gridline / outer-spine regression tests in test_polar.py.

PR checklist

Code is PEP8 compliant.
New and changed code is tested.
Plot-related changes verified in CI (no image baselines change; behaviour is byte-identical outside the bug case).

AI Disclosure

AI was used to help proofread the grammar of this PR description and to understand some parts of the existing codebase during development. The code changes and design decisions were developed independently.

rcomer

Thanks @SharadhNaidu, this fix makes sense to me and I have even learned something as I did not know about np.nextafter!

I am wondering if we actually need the polar-specific tests. Obviously for the purpose of this work is it necessary to verify that the reported polar cases are fixed. However, the fix is entirely within path.py, and the underlying problem is now covered by the computationally cheaper tests in test_path.py.

rcomer · 2026-06-06T13:33:36Z

I suggest to also rebase you branch against main, as it is a long way behind.

SharadhNaidu · 2026-06-06T13:52:54Z

I suggest to also rebase you branch against main, as it is a long way behind.

done , rebased and working on the current main

matplotlib#26972) On polar plots, radial gridlines and the outer spine could collapse to a near-empty arc. The root cause is Path.arc's angle-unwrap step: a span of 360*n plus a tiny floating-point rounding error was reduced modulo 360 to a near-zero arc instead of a full circle. Detect spans that are within floating-point tolerance of a whole number of turns and draw a complete circle; otherwise unwrap theta2 to the shortest arc within 360 degrees as before. The change is byte-identical for all non-degenerate inputs, so no call sites need to change. Adds unit coverage for the snap/unwrap behaviour and end-to-end polar gridline and spine regression tests.

- Replace the 1e-9 magic tolerance with 1e-12 turns, justified by the observed error from the polar transform pipeline (machine-epsilon level, under 1e-15 turns), via np.rint(n_turns). - Snap on any non-zero whole number of turns (abs winding), restoring the legacy full-circle behaviour for exact negative multiples such as Path.arc(0, -360); the previous rework collapsed those to an empty arc. - Add regression tests for negative full circles and a tolerance-tightness guard (a span just past a full turn must not snap).

- Drop the contrast and GitHub issue references from the Path.arc docstring; just state that a complete circle is drawn. - Remove issue references and reword the test_path.py comments for a reader unfamiliar with this PR's history. - Make test_arc_full_circle_snap compare all vertices against the reference circle, consistent with (and slightly stricter than) the neighbouring test. - Remove test_polar_transform_constant_r_arc (it passes against main and no longer guards the fix) and merge the two outer-spine tests into a single test_polar_outer_spine_not_collapsed using the spine bbox.

SharadhNaidu · 2026-06-07T01:15:53Z

@rcomer , I am sorry if i am violating the rules of directly tagging you but , is there any changes that i have to make ?

rcomer

Thanks @SharadhNaidu

I am still not sure about

I am wondering if we actually need the polar-specific tests. Obviously for the purpose of this work is it necessary to verify that the reported polar cases are fixed. However, the fix is entirely within path.py, and the underlying problem is now covered by the computationally cheaper tests in test_path.py.

But I will leave it for the second reviewer or other developer to weigh in on that. I note there are now fewer of these tests than when I wrote the comment.

For code changes, we require approvals from two core developers, so this will have to wait a bit longer before merge.

I am glad that your experience in #31674 has not put you off!

SharadhNaidu · 2026-06-08T16:43:32Z

For code changes, we require approvals from two core developers, so this will have to wait a bit longer before merge.

no worries , I will be waiting for their review and make the recommended changes , if any.

I am glad that your experience in #31844 31674 has not put you off!

Yup , learnt a lot in that process.

SharadhNaidu · 2026-06-09T10:05:16Z

applied the recommended changes , is_full_circle reads much cleaner.

will improve interms of writing comments.

SharadhNaidu · 2026-06-09T12:06:34Z

@timhoffm I am sorry if i am violating the rules of directly tagging you but , is there any changes that i have to make ?

timhoffm · 2026-06-09T12:28:15Z

Thanks @SharadhNaidu, and congatulations on your first PR to Matplotlib! 🎉 We hope to see you back.

SharadhNaidu · 2026-06-09T12:30:44Z

@timhoffm thanks and looking to contribute as much as i can.

…cle on polar plots (#20388, #26972)

…844-on-v3.11.x Backport PR #31844 on branch v3.11.x (FIX: snap near-integer arc windings to a full circle on polar plots (#20388, #26972))

SharadhNaidu mentioned this pull request Jun 5, 2026

FIX: fix radial gridlines and outer spine collapsing on polar plots (#20388, #26972) #31701

Closed

github-actions Bot added the topic: path handling label Jun 5, 2026

rcomer reviewed Jun 6, 2026

View reviewed changes

SharadhNaidu added 3 commits June 6, 2026 19:55

SharadhNaidu force-pushed the fix-arc-winding-snap branch from 560f59f to bd946aa Compare June 6, 2026 14:25

rcomer approved these changes Jun 8, 2026

View reviewed changes

timhoffm reviewed Jun 9, 2026

View reviewed changes

Comment thread lib/matplotlib/path.py Outdated

Address review: extract is_full_circle and clarify arc snap comment

e9ed519

timhoffm approved these changes Jun 9, 2026

View reviewed changes

timhoffm modified the milestones: v3.12.0, v3.11.0 Jun 9, 2026

timhoffm merged commit 319174c into matplotlib:main Jun 9, 2026
37 of 40 checks passed

meeseeksmachine mentioned this pull request Jun 9, 2026

Backport PR #31844 on branch v3.11.x (FIX: snap near-integer arc windings to a full circle on polar plots (#20388, #26972)) #31861

Merged

timhoffm added a commit that referenced this pull request Jun 9, 2026

Backport PR #31844: FIX: snap near-integer arc windings to a full cir…

bb14123

…cle on polar plots (#20388, #26972)

timhoffm added a commit that referenced this pull request Jun 9, 2026

Merge pull request #31861 from meeseeksmachine/auto-backport-of-pr-31…

f6e82c7

…844-on-v3.11.x Backport PR #31844 on branch v3.11.x (FIX: snap near-integer arc windings to a full circle on polar plots (#20388, #26972))

rcomer mentioned this pull request Jun 9, 2026

fix floating comparison in path.arc #26991

Closed

3 tasks

This was referenced Jun 12, 2026

[Bug]: numerical instability of Path.arc for 0 degree or 360 degree arc #29953

Closed

FIX: update arc to do better rounding and allow wrap #29962

Closed

Uh oh!

Conversation

SharadhNaidu commented Jun 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR summary

Fix

Tests

PR checklist

Uh oh!

rcomer left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

rcomer commented Jun 6, 2026

Uh oh!

SharadhNaidu commented Jun 6, 2026

Uh oh!

SharadhNaidu commented Jun 7, 2026

Uh oh!

rcomer left a comment

Choose a reason for hiding this comment

Uh oh!

SharadhNaidu commented Jun 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

SharadhNaidu commented Jun 9, 2026

Uh oh!

SharadhNaidu commented Jun 9, 2026

Uh oh!

Uh oh!

timhoffm commented Jun 9, 2026

Uh oh!

SharadhNaidu commented Jun 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

SharadhNaidu commented Jun 5, 2026 •

edited

Loading

SharadhNaidu commented Jun 8, 2026 •

edited

Loading