Date: 2026-03-03
Status: Execution checklist (feature/gate based only)
Purpose: Define the historical closure program that was required before starting the egui_glow -> egui_wgpu cut. The UI-backend cut has since landed; this checklist remains useful only as gate archaeology for deeper WebRender/runtime bridge work.

Canonical alignment:

• testing/wr-wgpu_notes/2026-03-01_webrender_wgpu_renderer_implementation_plan.md
• research/2026-03-01_webrender_wgpu_renderer_research.md
• subsystem_ux_semantics/2026-03-01_ux_execution_control_plane.md
• subsystem_ux_semantics/2026-02-28_ux_contract_register.md
• viewer/2026-02-26_composited_viewer_pass_contract.md
• 2026-03-02_scaffold_registry.md
• PLANNING_REGISTER.md (§0.12 / §0.13 deferral posture, §1C lanes and done gates)

---

This checklist is not time-based.
Progress is measured only by feature closure + validation gate evidence.

A gate is closed only when:

1. feature semantics are implemented end-to-end,
2. diagnostics evidence exists for the critical contract,
3. scenario coverage exists for the critical path,
4. spec/code parity artifacts are updated.

If any one item is missing, gate remains open.

---

Feature objective:

• Core UX baseline from the UX control-plane is coherent and executable (interaction correctness, viewer correctness, lifecycle/routing correctness, degradation clarity, spec/code parity).

Primary issue domains:

• #292-#301 (UX integration deliverables and parity)
• #302 parity baseline maintenance

Validation gate:

• Control-plane and matrix artifacts agree on status:

  • subsystem_ux_semantics/2026-03-01_ux_execution_control_plane.md
  • 2026-03-01_ux_migration_feature_spec_coverage_matrix.md
  • 2026-03-01_ux_migration_lifecycle_audit_register.md

• No contradictory claim between docs and runtime behavior for active milestone slices.

Status: open

---

Feature objective:

• Graph camera/navigation/selection interactions are deterministic under normal and focus-churn paths.

Primary issue domains:

• #173, #104, #101, #103, #271, #185, #102

Validation gate:

• Critical interaction scenarios pass in harness:

  • camera pan/zoom/reset/fit
  • lasso + selection semantics under modifier combinations
  • keyboard and pointer interaction coexistence

• Diagnostics channels show no authority/routing ambiguity for camera dispatch in critical journeys.

• User-visible regressions in lane:stabilization bug register are closed with scenario evidence.

Status: partial

---

Feature objective:

• Open/close/split/focus handoff and first-render activation are deterministic for Workbench/Frame/Tile flows.

Primary issue domains:

• #174, #186, #187 (primary)
• #118, #119 (enabling decomposition)

Validation gate:

• Scenario coverage for:

  • spawn/activate first render
  • close-pane successor focus handoff
  • deterministic focus owner transitions across graph pane and node/tool panes

• No blank viewport/focus-race repros in active stabilization register.

• Coordinator-boundary invariants remain green for touched files.

Status: partial

---

Feature objective:

• Content-opening actions always route through Graphshell semantic ownership (no legacy bypass).

Primary issue domains:

• #175

Validation gate:

• All open-in-new-view paths create/route through canonical Graphshell node/tile semantics.
• No legacy context-menu bypass path can create pane state outside declared intent flow.
• Routing diagnostics include explicit reason/path for open decision.

Status: open

---

Feature objective:

• F2/global command surface, contextual command surface, and radial/menu semantics share one command model.

Primary issue domains:

• #176, #106, #107, #108, #178, #270

Validation gate:

• Command naming/scope reflects canonical semantics (no misleading surface labels).
• Command invocation parity exists across keyboard/pointer entry points.
• Disabled-state policy and context category mapping are explicit and test-covered.

Status: open

---

Feature objective:

• Settings/history/control surfaces behave as first-class panes with stable apply and return-path behavior.

Primary issue domains:

• #109, #110, #177, #189

Validation gate:

• Settings changes with persistence semantics have scenario tests (including restart persistence where applicable).
• Tool pane entry/exit return-target behavior is deterministic.
• Control-surface state changes emit diagnostics or observable state transitions.
• Internal control-surface routing uses the canonical verso:// namespace (with legacy graphshell:// compatibility only as an alias), and workbench authority does not depend on hand-built address strings.

Status: partial

---

Feature objective:

• Viewer render mode behavior and fallback/degraded states are explicit, deterministic, and diagnostics-visible.

Primary issue domains:

• #162, #188, #111, #112
• related lanes: lane:stabilization, lane:viewer-platform, lane:spec-code-parity

Validation gate:

• Surface Composition Contract behavior is provable in runtime evidence:

  • pass ordering for composited mode
  • documented native-overlay affordance limitations
  • fallback reason visibility

• TileRenderMode-driven behavior is test-covered where applicable.

• Viewer/fallback docs match runtime reality.

Current status note:

• Fallback/degraded-state clarity itself is already landed: TileRenderMode::Placeholder vs CompositedTexture degraded-state semantics are covered by the committed pre_wgpu_critical_path.rs scenarios, and viewer diagnostics/affordance language is canonical under #162 + #188.
• The remaining G6 work before a backend switch is narrower: compositor pass-contract proof, native-overlay limitation receipts, and any adapter-level bridge invariants that still need runtime evidence.

Status: partial

---

Feature objective:

• Reliability is enforced through scenario + diagnostics + semantic probes, not ad hoc manual repro loops.

• #94, #251, #257, #261, #269, #272, #273

• Critical-path UxHarness scenarios exist and run as a required gate for affected areas.
• UxTree/Probe invariants catch ownership/focus/routing regressions before merge.
• Diagnostics evidence is present for all core authority boundaries.
• Address-routing coverage includes both system/workbench authority (verso://...) and domain-record handoff (notes://..., plus any active graph://... / node://... paths), so content identity and workbench placement failures are caught before merge.

Current status note:

• Critical-path UxHarness coverage is already closed by #273; the committed suite in pre_wgpu_critical_path.rs is the current baseline for migration-adjacent UX flows.
• The remaining blocker is turning that committed coverage into structural snapshot baseline/diff CI and merge-blocking automation (#257), alongside the existing probe/diagnostics contracts.

Status: partial

---

• No pre-wgpu-critical feature remains in scaffold state for core UX closure paths.

Primary scaffold markers:

• [SCAFFOLD:view-dimension-ui-wiring]
• [SCAFFOLD:verse-protocol-handler]
• [SCAFFOLD:wasm-mod-loader-runtime]

Validation gate:

• For each marker still present in the active scaffold registry and impacting core UX closure lanes: closure criteria met and marker removed or explicitly de-scoped from pre-wgpu closure.
• No “partial but merged” slice in core path without explicit scaffold marker and closure gate.

Current status note:

• The active scaffold registry is narrower than this checklist's original snapshot: [SCAFFOLD:viewer-wry-runtime-registration] is already closed, and [SCAFFOLD:divergent-layout-commit] is no longer an active registry marker.
• The remaining G8 question is whether the still-active markers are either closed or explicitly de-scoped from pre-wgpu UX closure, rather than silently treated as background debt.

Status: open

---

Feature objective:

• Only begin renderer backend migration once application readiness is true by feature evidence.

Primary source:

• testing/wr-wgpu_notes/2026-03-01_webrender_wgpu_renderer_implementation_plan.md
• research/2026-03-01_webrender_wgpu_renderer_research.md
• PLANNING_REGISTER.md (§0.12 / §0.13)

Validation gate:

• Application readiness conditions are all true:

  1. authority boundaries explicit/enforced,
  2. core UX flows usable without constant regressions,
  3. app semantics complete enough for practical use,
  4. current-stack bugs are not dominating development,
  5. migration can be evaluated as upgrade, not desperation.

• Runtime viewer bridge precondition (#180) has current tracker-linked viability evidence; technical-definition-only or archived receipts are insufficient for authorization.

Current status note:

• Active canon now treats #180 / #183 as upstream-first, non-milestone renderer work while Glow remains the milestone runtime composition policy.
• G9 therefore authorizes backend-migration start only when the app-side feature gates are evidence-complete and #180 is backed by current viability evidence rather than problem-definition or historical spike receipts alone.

Status: blocked (by G0-G8 + #180)

---

1. No net-new UX feature expansion while any of G1, G2, G3, or G6 is open.
2. Any PR touching a gate-owned area must include:
   • contract-level scenario evidence,
   • diagnostics evidence,
   • parity doc delta where behavior changed.
3. If a regression reopens a closed gate, that gate reverts to open immediately.
4. G9 cannot be manually overridden by schedule pressure.

---

• The next move is not “more planning.” It is strict gate closure on partially implemented pre-wgpu features.
• The dominant risk is not missing ideas; it is accepting partially closed interaction contracts.
• Success criterion is not velocity of merged slices; it is number of gates moved from open/partial to closed with evidence.

---

1. Close G7's remaining automation work by landing #257 snapshot baseline/diff CI on top of the committed critical-path suite.
2. Close the remaining G6 compositor/adapter evidence for render-mode fallback and affordance behavior.
3. Continue G1 residual closure where camera/selection determinism still lacks durable evidence.
4. Continue G2 / G3 residual closure only where a slice directly discharges a listed residual or a regression proves the contract is not actually stable; do not re-plan already-canonical semantics.
5. Re-evaluate G9 only after G0-G8 status is evidence-complete.