Phase 1 (1/5): Package scaffold, CI, and core math layer by ronaldmannak · Pull Request #1 · PicoMLX/PicoManim

ronaldmannak · 2026-07-05T14:31:20Z

Summary

Bootstraps the PicoManim Swift package (Swift 6.2, macOS 15+ / iOS 18+) with CI and the geometry primitives every later layer builds on:

Vec2 — SIMD2<Double> alias with rotation, lerp, and unit-direction helpers.
Transform2D — affine transform kept decomposed (scale → rotate → translate) so animations can interpolate rotation and scale without shearing.
ManimColor — RGBA color with the classic Manim palette (.blue, .red, .yellow, …).
CI — GitHub Actions: swift build + swift test on macos-26 and on Linux (swift:6.2 container). The Linux job exists to keep the core platform-neutral (SwiftUI code in later PRs is fenced behind #if canImport(SwiftUI)). The pull_request trigger deliberately has no base-branch filter so the stacked PRs in this series get CI too.

Stack (Phase 1: shapes + animation + ManimView)

This is PR 1/5, base main. The stack, in review/merge order:

→ this PR — scaffold, CI, core math
Bézier engine (CubicCurve, BezierPath)
Mobject model + shape factories
Animation timeline (RateFunction, ManimAnimation, ManimScene)
ManimView SwiftUI preview + demo scene + full docs

Each later PR is based on its parent branch; they will be retargeted to main as parents merge.

Intentionally not included

Paths/shapes, animation, and rendering — they arrive in PRs 2–5 so each diff stays reviewable on its own.

Test strategy

Exact-value math tests (rotation quarter-turn, lerp midpoints, transform composition order, hex color round-trip) — plus this PR proves out the CI pipeline itself for the rest of the stack.

Concurrency

All types are Sendable value types. No @unchecked Sendable anywhere in this stack.

🤖 Generated with Claude Code

https://claude.ai/code/session_01YGb3ukB5dQy4vgz7Z1hags

Generated by Claude Code

Bootstrap the PicoManim Swift package (Swift 6.2, macOS 15+/iOS 18+) with the geometry primitives everything else builds on: - Vec2: SIMD2<Double> alias with rotation, lerp, and direction helpers. - Transform2D: scale -> rotate -> translate affine transform kept in decomposed form so animations can interpolate components directly. - ManimColor: RGBA color with the classic Manim palette. - GitHub Actions CI: swift build + swift test on macos-26 and on Linux (swift:6.2 container) to keep the core platform-neutral. The pull_request trigger has no base-branch filter so stacked PRs get CI. Part 1/5 of the Phase 1 stack (shapes -> animation -> ManimView). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YGb3ukB5dQy4vgz7Z1hags

gemini-code-assist

Code Review

This pull request introduces the core math layer for the PicoManim animation library, adding basic structures such as Vec2, Transform2D, and ManimColor along with their corresponding unit tests. The review feedback suggests several improvements to robustness and code reuse: using hypot for safer vector length calculations, clamping color components within the 0...1 range during interpolation to handle overshooting animations, and leveraging the existing lerp helper function in Transform2D to avoid manual interpolation logic.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

gemini-code-assist · 2026-07-05T14:38:14Z

+    public var length: Double {
+        (x * x + y * y).squareRoot()
+    }


Using (x * x + y * y).squareRoot() to calculate the Euclidean length of a vector can lead to overflow or underflow for very large or very small values of x or y. Using hypot(x, y) is more robust as it is specifically designed to avoid intermediate overflow and underflow.

Suggested change

public var length: Double {

(x * x + y * y).squareRoot()

}

public var length: Double {

hypot(x, y)

}

Fixed in d6e9dfa. Vec2.length now uses Foundation.hypot(x, y), avoiding intermediate overflow/underflow. Existing length tests cover the common cases; no behavior change for in-range values. No impact on later PRs in the stack (they consume length unchanged).

Generated by Claude Code

gemini-code-assist · 2026-07-05T14:38:14Z

+    public static func lerp(_ a: ManimColor, _ b: ManimColor, _ t: Double) -> ManimColor {
+        ManimColor(
+            red: a.red + (b.red - a.red) * t,
+            green: a.green + (b.green - a.green) * t,
+            blue: a.blue + (b.blue - a.blue) * t,
+            alpha: a.alpha + (b.alpha - a.alpha) * t
+        )
+    }


The lerp function does not clamp the interpolated color components to the 0...1 range. During animations with overshooting rate functions (such as back or elastic easing), t can go outside 0...1, causing color components to go out of bounds. This violates the documented invariant "components in the 0...1 range" and can cause rendering issues. Using the clamp and lerp helper functions defined in Vec2.swift ensures the components remain valid.

Suggested change

public static func lerp(_ a: ManimColor, _ b: ManimColor, _ t: Double) -> ManimColor {

ManimColor(

red: a.red + (b.red - a.red) * t,

green: a.green + (b.green - a.green) * t,

blue: a.blue + (b.blue - a.blue) * t,

alpha: a.alpha + (b.alpha - a.alpha) * t

)

}

public static func lerp(_ a: ManimColor, _ b: ManimColor, _ t: Double) -> ManimColor {

ManimColor(

red: clamp(lerp(a.red, b.red, t), 0...1),

green: clamp(lerp(a.green, b.green, t), 0...1),

blue: clamp(lerp(a.blue, b.blue, t), 0...1),

alpha: clamp(lerp(a.alpha, b.alpha, t), 0...1)

)

}

Fixed in d6e9dfa, with a required correction in 65d2984: the suggestion as written doesn't compile — inside the static lerp member, the unqualified lerp(a.red, ...) call resolves to the enclosing member itself (type scope shadows the module-level scalar helper), which broke CI. The final version module-qualifies the helper (PicoManim.lerp) and clamps every component to 0...1, preserving the documented invariant under overshooting rate functions. A source comment explains the qualification so it doesn't get "simplified" back. Later PRs benefit automatically (PR #4's transform color interpolation goes through this function).

Generated by Claude Code

gemini-code-assist · 2026-07-05T14:38:14Z

+    public static func lerp(_ a: Transform2D, _ b: Transform2D, _ t: Double) -> Transform2D {
+        Transform2D(
+            translation: Vec2.lerp(a.translation, b.translation, t),
+            rotation: a.rotation + (b.rotation - a.rotation) * t,
+            scale: Vec2.lerp(a.scale, b.scale, t)
+        )
+    }


The lerp function manually computes linear interpolation for the rotation angle, duplicating the logic of the scalar lerp helper function defined in Vec2.swift. Using the helper function improves code reuse and maintainability.

Suggested change

public static func lerp(_ a: Transform2D, _ b: Transform2D, _ t: Double) -> Transform2D {

Transform2D(

translation: Vec2.lerp(a.translation, b.translation, t),

rotation: a.rotation + (b.rotation - a.rotation) * t,

scale: Vec2.lerp(a.scale, b.scale, t)

)

}

public static func lerp(_ a: Transform2D, _ b: Transform2D, _ t: Double) -> Transform2D {

Transform2D(

translation: Vec2.lerp(a.translation, b.translation, t),

rotation: lerp(a.rotation, b.rotation, t),

scale: Vec2.lerp(a.scale, b.scale, t)

)

}

Fixed in d6e9dfa. Transform2D.lerp now reuses the shared scalar helper for rotation, module-qualified (PicoManim.lerp) because the unqualified name resolves to the enclosing static member rather than the global helper. No functional change; the existing transformLerpEndpoints test still pins the behavior.

Generated by Claude Code

- Vec2.length uses hypot to avoid intermediate overflow/underflow. - ManimColor.lerp clamps components to 0...1 so overshooting rate functions can't violate the documented component range. - Transform2D.lerp reuses the shared scalar lerp helper. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YGb3ukB5dQy4vgz7Z1hags

Inside the static lerp member, unqualified lerp(a.red, ...) resolved to the enclosing member instead of the module-level scalar helper and failed to type-check. Qualify with the module name, matching Transform2D.lerp. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YGb3ukB5dQy4vgz7Z1hags

ronaldmannak · 2026-07-05T14:58:11Z

@codex review

Generated by Claude Code

chatgpt-codex-connector · 2026-07-05T15:02:28Z

Codex Review: Didn't find any major issues. Swish!

Reviewed commit: 65d29847d5

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

ronaldmannak · 2026-07-05T15:02:54Z

Ready for human review/merge.

Checklist:

✅ CI green on latest commit 65d2984 (test-macos + test-linux, swift build + swift test) — this PR also bootstraps the CI both jobs run on
✅ Codex reviewed latest commit: "Didn't find any major issues"
✅ All 3 Gemini review comments addressed and individually answered: hypot for Vec2.length, clamped ManimColor.lerp components, shared scalar-lerp helper reuse — fixed in d6e9dfa + 65d2984 (the follow-up fixed a member-shadowing compile error from applying the lerp suggestion verbatim; explained in the thread and in a source comment)
✅ Stacked PRs current: Phase 1 (2/5): Cubic Bézier engine — curves, paths, partials, morph alignment #2–Phase 1 (5/5): ManimView SwiftUI preview, demo scene, and docs #5 all rebased on this branch's latest

Remaining notes:

This is the stack root (base main); it should merge first. I'll retarget Phase 1 (2/5): Cubic Bézier engine — curves, paths, partials, morph alignment #2 to main after merge.
One transient red CI run existed mid-review (the shadowing issue above); the head commit is green on both platforms.
No @unchecked Sendable; pure value types.

Generated by Claude Code

This was referenced Jul 5, 2026

Phase 1 (2/5): Cubic Bézier engine — curves, paths, partials, morph alignment #2

Merged

Phase 1 (5/5): ManimView SwiftUI preview, demo scene, and docs #5

Merged

Phase 2 umbrella: text/LaTeX mobjects, groups, axes, updaters, video export #6

Open

gemini-code-assist Bot reviewed Jul 5, 2026

View reviewed changes

claude added 2 commits July 5, 2026 14:39

ronaldmannak mentioned this pull request Jul 5, 2026

Phase 1 (3/5): Mobject model and Manim-style shape factories #3

Merged

ronaldmannak mentioned this pull request Jul 5, 2026

Phase 1 (4/5): Animation timeline — rate functions, animations, ManimScene #4

Merged

ronaldmannak merged commit 254ed60 into main Jul 5, 2026
2 checks passed

ronaldmannak mentioned this pull request Jul 6, 2026

Add PicoManim: Manim-inspired animation library for Swift #10

Closed

Navigation Menu

Uh oh!

Phase 1 (1/5): Package scaffold, CI, and core math layer - #1

Phase 1 (1/5): Package scaffold, CI, and core math layer#1
ronaldmannak merged 3 commits into
mainfrom
claude/picomanim-phase1-01-scaffold

ronaldmannak commented Jul 5, 2026

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot Jul 5, 2026

Uh oh!

ronaldmannak Jul 5, 2026

Uh oh!

gemini-code-assist Bot Jul 5, 2026

Uh oh!

ronaldmannak Jul 5, 2026

Uh oh!

gemini-code-assist Bot Jul 5, 2026

Uh oh!

ronaldmannak Jul 5, 2026

Uh oh!

ronaldmannak commented Jul 5, 2026

Uh oh!

chatgpt-codex-connector Bot commented Jul 5, 2026

Uh oh!

ronaldmannak commented Jul 5, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

ronaldmannak commented Jul 5, 2026

Summary

Stack (Phase 1: shapes + animation + ManimView)

Intentionally not included

Test strategy

Concurrency

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Jul 5, 2026

Choose a reason for hiding this comment

Uh oh!

ronaldmannak Jul 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Jul 5, 2026

Choose a reason for hiding this comment

Uh oh!

ronaldmannak Jul 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Jul 5, 2026

Choose a reason for hiding this comment

Uh oh!

ronaldmannak Jul 5, 2026

Choose a reason for hiding this comment

Uh oh!

ronaldmannak commented Jul 5, 2026

Uh oh!

chatgpt-codex-connector Bot commented Jul 5, 2026

Uh oh!

ronaldmannak commented Jul 5, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants