Reviewer read: the plan is right to say "inspect first", but it has not yet used the strongest evidence in the room. Scout's point about TransitionableComponent.ts (2750 bytes) and the README's done item "Automated props interpolation (position, rotation, size)" changes the first milestone. If interpolation already lives in the component layer, the open question is not "can we write Rotate" but "what does Move.ts call, and do Rotation and Scale expose the same interface as Position?" The plan's milestone 1 should name TransitionableComponent.ts and Action.ts explicitly. I disagree with one thing in the plan: it says Rotate and Scale come "modeled on Move". Move is the only real leaf action, so a sample of one may encode Position-specific assumptions. Scale also has 2D x/y values, so a Scale action may need a decision on uniform versus per-axis. That is a design question to put in the maintainer issue. The 18-byte files are identical but unread. They could hold an empty export, a comment, or a stub that other files import. The README says tweens are done, so the maintainer may treat these as deliberately deferred. Permissions are unchanged. MIT is present, canPush is false, and there is no maintainer permission or endorsement. Nothing has been changed or run.

Inspect tween and component code · This step could not finish. Saved work is retained. The workspace needs review before another attempt.

Builder here. Scout's point changes my plan. If TransitionableComponent.ts (2750 bytes) and the Position, Rotation and Scale components already handle interpolation, Rotate and Scale may be thin wrappers around them, with Move.ts (1001 bytes) as the template. That would be a very small release: two actions, each with a test and a Sequence example. I haven't opened any file, so this is inferred from names and sizes. I disagree on Fade. Scout suggests it may be easier, but the file list shows no alpha component, so Fade may need new component work. It stays out until inspection shows otherwise. The finish line is concrete: a PixiJS developer writes Sequence(Move, Rotate, Scale, Delay, Callback) and it runs without hand-written tweening. The first check is what Move.ts imports and whether Rotation and Scale expose the same interface as Position. The repo is MIT, but canPush is false and we have no maintainer permission or endorsement. This stays a fork-and-PR proposal, opened as a maintainer-facing issue first, since the README marks tweens as done. Nothing has been changed or run.

Scout read, adding one point the team hasn't weighed. The tween gap may be narrower than "five missing actions". The README roadmap has a separate item, "Automated props interpolation (position, rotation, size)", marked done, and the tree has TransitionableComponent.ts (2750 bytes) next to Position, Rotation and Scale components. Rotation and scale tweening may already live in that component layer. If so, the unfinished work is not a new interpolation engine but exposing it as chainable actions, in line with the Cocos2d-style goal. That would shrink the job and shift the check from "can we write Rotate" to "does the component layer already do the work, and what does Move.ts call?" I'd also question one thing in the plan: Fade might be easier than Builder assumes. Fade needs an alpha property, and I can't tell from the file list whether any component covers it. That is also unread. User and outcome stay the same: a 2D PixiJS developer chaining Move, Rotate, Scale, Delay and Callback in a Sequence. MIT is present, but canPush is false and there is no maintainer permission or endorsement, so this stays a proposal to the maintainer first. Nothing has been opened, changed or run.

Steward read: the team converges on the tween/action layer as the first milestone, not the API/Core split (#2, #4, #5, #6) or 3D (#3, #7). Those are architectural, and several issue bodies are empty. Reviewer is right that the premise is unverified. Fade, Rotate, Scale, Skew and Tint share one 18-byte blob SHA, so they are identical files, but nobody has opened them. Builder's reuse of the Rotation and Scale components is also inferred from names and sizes. So the first step is inspection, not building. User: a 2D web game developer on PixiJS who wants Cocos2d-style action chains. Once it works, they could compose Move, Rotate, Scale, Delay and Callback in a Sequence without hand-written tweening. Limits: Rotate and Scale only, with Fade only if inspection shows it fits. Tint, Skew, audio, 3D and the module split stay out. Permissions: the repo is MIT, but canPush is false and we have no maintainer permission or endorsement. Work would be a fork-and-PR proposal, and a maintainer-facing issue should come first. Nothing has been changed or run. I'm marking this revise, not ready, until the files are read and the maintainer has had a chance to respond.

Open a maintainer-facing proposal · proposed

Implement Rotate and Scale with tests on a fork · proposed