Brummel
862882bc04
Architect drift review (bb90c42..481172a): no contract drift. C9 (render reads structure + label()/params()/output() only, never eval), C8 (sink = zero-output leaf, no special case), C12 (blueprint = root composite is render-structural only; types stay distinct), C23 (consumer-prefix mirrors the := binding), #38 (composites stay opaque) — all hold. Three fidelity items, resolved: - [high, fixed] spec acceptance over-claimed: the SimBroker slot-stub box read as in-scope though the work is deferred. Spec amended — the stub acceptance is annotated against #49, the met boxes checked. - [medium, recorded] no docs/plans/0023: this cycle ran spike→refactor under user steering instead of planner→implement, because the edge-label-vs- consumer-side fork needed an empirical probe (ascii-dag silently drops edge labels on collision) before a placeholder-free plan was possible. A plan RECORD (not a task plan) now holds the 0023 slot so the counter pairing stays intact and the next cycle takes 0024. - [low, carry-on] multi_output_field_name's leaf-producer fallback is latent and untested (no multi-output leaf in the corpus). Carried as debt. Regression scripts: none configured (no-op). Build/test/clippy green. Deferred stub work is tracked in #49 (user-signed). closes #48
2.7 KiB
Unify the Blueprint Main-Graph Render — Implementation Record
Parent spec:
docs/specs/0023-unify-blueprint-render.mdThis is NOT a task-by-task plan. Cycle 0023 was implemented through an exploratory spike→refactor path under direct user steering, not the standard
planner → implementdecomposition. This file exists to keep the spec/plan counter pairing intact (naming.counter_dirs: [docs/specs, docs/plans]) and to record why the standard plan artefact is absent, so the next cycle takes slot 0024 and a future reader does not mistake the gap for a lost file.
Goal: one shared render path for the blueprint main graph and the composite
where: interior — the blueprint is the root composite.
Why no task-by-task plan
The spec named the direction but left one detail to "the planner/implementer call against ascii-dag's actual capability": where a multi-output producer's selected field surfaces. Resolving it needed an empirical answer, not a paper decision:
- The obvious default (an ascii-dag per-edge label) turned out to be unreliable —
ascii-dag 0.9.1's
can_place_labelsilently drops a label it cannot place without collision (verified by throwaway probe against the real macd topology). The Plain path drops it with no legend; the colored path shifts it to a legend. - So the field had to move to the consumer node label
(
[histogram → Exposure(scale)]), which also removed therender_flatsignature change entirely (no edge-label slot).
A placeholder-free plan could not be written before that probe, so the work ran as
a spike (validate the notation visually with the user) → refactor (collapse the
duplication into one shared render_graph over borrowed slices — LeafFactory
is not Clone, so an owned root-composite adapter is impossible).
What shipped (commits)
5dd9503— spec: 0023 unify blueprint main-graph render481172a— feat(aura-cli): unify blueprint main-graph render through one shared core (render_graph;ParamNames{Aliases, Factory};EntryunifyingRole/SourceSpec; deleted the duplicatedblueprint_leaf_label)
Production change confined to crates/aura-cli/src/graph.rs; goldens re-captured
in crates/aura-cli/src/main.rs. No engine change. Build/test/clippy green.
Deferred / debt (filed, not lost)
- Issue #49 — top-level blueprint multi-input leaves render without
#-slot stubs (stub_ctx: None);fan_in_identifiers/signature_ofstays composite-coupled. User-signed deferral. - Latent, untested —
multi_output_field_name's leaf-producer fallback (from_field > 0→ index string) is unexercised (no multi-output leaf in the current corpus); carried as latent code, no anchoring test this cycle.