COOL IDEAS™: compare two checkpoint bases through bounded causal diff

[flyingrobots]

Classification

COOL IDEA™ — read-only historical comparison over real causal rope evidence.

Observation

Existing roadmap work covers previewing one historical basis and comparing current state with a selected checkpoint/anchor. A more general primitive would compare checkpoint A directly with checkpoint B without moving the canonical head, materializing full snapshots as authority, or pretending every checkpoint is a CausalAnchor.

The idea

Given two opaque Jim checkpoint identities, ask the native Echo-backed host for a bounded causal comparison of their retained rope bases.

A result could include:

• both resolved worldline and basis-head identities;
• a bounded byte/line region or summary;
• inserted/deleted/modified spans;
• supporting rewrite, diff, receipt, and tick identities;
• coverage and continuation posture;
• an explicit retention obstruction when required history is unavailable.

The operation is observation only. Reconciliation, checkout, undo, export, and anchoring remain separate intents.

Why it is cool

This unlocks a truthful “what changed between these two meaningful moments?” surface for humans and agents:

• compare autosave to manual save;
• compare pre-refactor to post-refactor checkpoints;
• inspect two retained experiment points without disturbing the live buffer;
• feed bounded evidence into review or explanation tooling;
• distinguish semantic checkpoint identity from its declaration occurrences.

Constraints

• Do not move or mint a canonical RopeHead.
• Do not compute an unbounded local full-string diff.
• Do not treat CAS snapshots, Git, or TypeScript materializations as authority.
• Resolve checkpoint identities and basis membership inside the native host.
• Same-worldline comparison is the initial scope; cross-worldline reconciliation belongs to WF-0126: Braid and diff reconciliation UX #163 unless separately designed.
• Missing, pruned, foreign, or incompatible history returns a typed obstruction.
• A comparison does not imply or create a CausalAnchor.

Acceptance criteria

• Accept two opaque checkpoint IDs and an explicit bounded comparison request.
• Resolve both checkpoint facts and validate their bases/worldlines in native code.
• Return bounded causal diff evidence with exact basis identities and continuation/coverage posture.
• Support same-checkpoint comparison as an empty, fully evidenced result.
• Define behavior when the same checkpoint proposition has multiple declaration occurrences.
• Prove the observation is read-only: no head, rewrite, diff, checkpoint, anchor, or text mutation is minted.
• Prove restart behavior from Echo history/WAL.
• Prove typed obstructions for missing checkpoint, foreign worldline, retention gap, and unsupported operation.
• Provide both human-readable and agent-readable consumers without duplicating authority in TypeScript.

Related, not duplicate

• WF-0109: Historical Basis Preview #134 — historical preview and current-versus-checkpoint views.
• Slice: Compute +N/-N from causal rewrites since selected anchor #234 — current durability delta from a selected anchor.
• WF-0126: Braid and diff reconciliation UX #163 — cross-worldline braid/reconciliation UX.
• COOL IDEAS™: expose an Echo-powered undo timeline #79 and COOL IDEAS™: add a Why Did This Happen inspector for Echo evidence #81 — timeline and evidence-inspector consumers.
• Roadmap: causal rope everywhere and only causal rope #255 — causal rope roadmap.

[coderabbitai]

🔗 Related PRs

#195 - Add honest why observation coordinates [merged]
#205 - Document graph-backed rope runtime gate [merged]
#265 - CR-01: Separate checkpoints from Echo anchor authority [merged]
#280 - Add runtime-backed causal range evidence [merged]
#282 - Route checkpoint declarations through Echo [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!