ADR 0007: Scene Tree Lab Explorer

Status

Accepted - 2025-02-14

Context

We frequently rely on the optional rmscene dependency to decode ReMarkable .rm files. When debugging new primitives or studying the device format we need a fast way to inspect the decoded structures in a human-friendly way. The previous workflow involved sprinkling print statements or dropping into a debugger, which made it hard to share findings and slowed iteration.

Decision

Add a lightweight lab script (lab_rm_reader.py) that loads a sample .rm file into rmscene’s SceneTree and walks the structure. The walker summarises key objects (groups, CRDT sequence items, lines, points, LWW values) with concise, well-indented output so developers can spot hierarchy issues at a glance.

Consequences

• The lab script lives outside the distributed package and has no stable API guarantees. It is intentionally loose so experiments remain easy.

• The documentation now points to the script so new contributors know where to start when exploring decoded scenes.

• Future tooling (for example, JSON emitters or visualisers) can reuse the same traversal strategy. The first packaged helpers derived from this work are rmfiles.scene_to_json() and rmfiles.scene_to_svg(), which produce structured JSON and SVG output for SceneTree objects.