Add tests for the new Pose class

[Micheal Parks (micheal-parks)]

Overview

Adds a comprehensive set of unit tests for the new Pose class, giving the visualizer's core math library a solid bump in coverage.

[changeset-bot]

🦋 Changeset detected

Latest commit: 5e404fb

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@viamrobotics/motion-tools	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[Micheal Parks (micheal-parks)]

Claude (@claude) review

[claude]

Claude finished Micheal Parks (@micheal-parks)'s task in 2m 58s —— View job

---

Review

• Read CLAUDE.md and repo conventions
• Read the test file (pose.spec.ts)
• Read the Pose class implementation
• Read Frame type for context
• Post inline comments for line-specific issues
• Post final summary

---

This is a well-structured test suite. The top-of-file comment explaining why round-trips alone aren't sufficient, and the deliberate use of absolute assertions to pin unit constants, is exactly the right approach. The it.each usage, the fields() snapshot helper, and the oZ-defaulting case matrix are all solid.

Issues

Two coverage gaps worth fixing (see inline comments):

1. setFromObject3D only asserts x, not y/z (line 194) — the test name says "reads position back as millimetres" but only one axis is checked. An axis-swap bug would pass.

2. ov_radians and ov_degrees tests only check theta (lines 382, 390) — the ov.copy() and ov.set() calls also copy the x/y/z components of the orientation vector. A regression that zeroed those fields would go undetected.

Minor

• The destination-parameter variants of toMatrix4, toVector3, and toObject3D aren't exercised in the "shared scratch objects" suite — only toQuaternion(new Quaternion()) gets that treatment (line 295). These are the same default-parameter pattern, so a test like pose.toMatrix4(new Matrix4()) returning a correctly-written matrix would round out that group.

[claude]

The test name promises "reads position back as millimetres" but only asserts x. y and z go through the same setFromVector3 path (each independently multiplied by M_TO_MM), so a bug like an axis swap would pass undetected here.

Suggested change

	expect(pose.theta).toBeCloseTo(90, 6)
	expect(pose.x).toBeCloseTo(1000, 6)
	expect(pose.y).toBeCloseTo(2000, 6)
	expect(pose.z).toBeCloseTo(3000, 6)
	expect(pose.theta).toBeCloseTo(90, 6)

[claude]

Only theta is checked. The ov_radians path calls ov.copy(frame.orientation.value), which copies x, y, and z too — but a regression that dropped those fields would sail through here. Worth adding the full orientation assertion:

Suggested change

	})
	const pose = new Pose().setFromFrame({
	orientation: { type: 'ov_radians', value: { x: 0, y: 0, z: 1, th: Math.PI / 2 } },
	})
	expect(pose.oX).toBeCloseTo(0, 6)
	expect(pose.oY).toBeCloseTo(0, 6)
	expect(pose.oZ).toBeCloseTo(1, 6)
	expect(pose.theta).toBeCloseTo(90, 6)

[github-actions]

PR Preview Action v1.8.1
Preview removed because the pull request was closed.
2026-07-28 20:26 UTC