Post-#332 follow-ups (June 2026)

Punch list captured after the foundational-refactors PR landed on main. Items 1–4 are tracked separately by Marko + the other agent; items 5–9 are open here for future sessions.

Tracked elsewhere (Marko + other agent)

1. 2D-debug helper / orbital math extraction

Extract Kepler propagation, spacecraft position, camera position, and screen projection into pure fns under $lib/orbital/. Render a minimal black-and- white 2D view of route / planet / ship / camera so the math is verifiable without the 3D stack. The 2D view consumes the same fns the 3D scene will eventually call — sanity check: if 2D and 3D disagree about Curiosity at MET=100 days, the bug is in the renderer, not the math. Doubles as a production helper for debug overlays + future test harnesses.

2. #327 — data.ts split + DAL architecture review

Phase 0 ADR answering the 7 architectural questions captured in the issue body (per-domain vs shared cache, query-key pattern vs lazy singletons, locale-overlay helper extraction, barrel-or-sweep, provenance-stays-or-moves, cache invalidation policy, foot-gun-proof cache-import shape). Phase 1 is the mechanical 10–11 module split, barrel-keep. RemoteData<E,T> (commit 3 of #332) is the consumer shape the split should adopt.

3. /fly effect-ladder consolidation (#330 C.3)

Deferred from #332 commit 12. /fly has 8+ scattered $effect blocks reacting to overlapping state (mission load, arc changes, view-mode switches, phase markers, cinematic state). Consolidate into 2-3 coordinated effects with internal switch(reason) dispatch. Needs careful cinematic- timing analysis so consolidation doesn't break the iconic-shot / flyby choreography landed on main during the #332 rebase window. The lifecycle + animate-loop foundation from #332 commit 8 is in place to build on.

4. /fly Apollo 11 phase-marker-jump test flake

tests/e2e/fly-apollo11-phase-markers.spec.ts:242 — pre-existing on main, not introduced by #332 or the post-rebase CSS fix. force: true click on the phase-marker dot somehow navigates the page to /missions (snapshot shows LaunchesBanner). The button has type="button" + plain onclick= {onJump} so the navigation is unexpected. Likely a hit-test / event bubbling issue with overlay z-stacking after the recent /fly cinematic work. Track + reproduce locally with headed Playwright.

Open here (deferred from #332)

5. Mission image rework — missing + garbage + hero selection

Scope: combined pass over static/images/missions/<id>.jpg (card cover) + static/images/missions/thumbnails/<id>.png (trajectory thumbnail) + static/images/missions/<id>/... (per-mission gallery).

• Missing entirely (4 missions, allowlisted in tests/e2e/_helpers/console-errors.ts):inspiration4, polaris-dawn, otv-6, otv-7. Slice B/C shipped index entries without art; cards render with cover-missing graceful degradation.
• Garbage that passed scoring: Marko reports the existing algorithmic scoring lets through low-quality images, especially in the hero slot. Hero is the most visible asset and the worst hit.
• Hero selection: which image becomes the "hero" per mission needs a tighter selection rule, not just a score threshold.

Next-step framing in progress with Marko (see "Image pass — framing" section in chat). Will fold into this section once the spec is set.

6. createSceneContext + createPickHandler

Deferred from #329 commit 4. Existing scene-renderer.ts + bindCanvasInputs cover the 5-of-7-route happy path. /iss + /tiangong's WebGL-availability + quality-tier-aware setup should drive the createSceneContext shape — easier to land alongside the first new consumer (e.g. a #325 raf + Three.js test harness) than to over-design in isolation.

7. Object3DUserData adoption sweep

The createUserDataTag<T> wrapper landed in #329 commit 4 with 11 tests but no callers migrated. ~27 as unknown as clusters across iss-proxy-model.ts, tiangong-proxy-model.ts, and station-assembly-anim.ts are the audit-identified targets. Mechanical sweep once #325 (raf + Three.js test harness) lands, since the wrapper makes pick-handler tests trivially easier.

8. RemoteData migration on /iss, /tiangong, /plan

#332 commit 10 migrated /missions + /fleet. The remaining routes have complex IO state shapes:

• /iss, /tiangong: two parallel fetches (modules + visitors) share a single loadFailed flag. Each becomes its own RemoteData<Error, IssModule[]> and a derived loadFailed = isError(modulesRequest) || isError(visitorsRequest).
• /plan: multiplexes computing / progress / loadFailed over a porkchop-grid compute + load path. RemoteData's three-state shape doesn't naturally absorb progress; needs a parameterised variant or separate progress field.

9. useUrlParam sweep beyond SurfaceScene hotspots

#332 commit 11 migrated SurfaceScene's ?hotspots= to useUrlParam<T> as the worked example. Other audit-flagged sites have awkward shapes:

• Panorama ?pano + ?yaw + ?pitch: 3 correlated fields syncing together; useUrlParam's single-field shape doesn't fit. Needs a panorama-specific wrapper or a multi-field rune variant.
• Station ?view + ?module: already abstracted via syncStationUrl; migration refactors the helper rather than the route.
• /missions filter set: overlaps #332 commit 9's typed filterState bag pattern — adopting useUrlParam would mix two structures in the same file. Bag stays for filters; useUrlParam is the right tool for isolated single-field contracts.
• /explore ?focus= / ?paths= / ?id=: read-only deep-links with no write-back; existing $effect pattern is already minimal.

Could fold into #327's follow-up once data.ts decisions settle.

---

Captured June 2026 after #332 (foundational-refactors PR) shipped to main at commit 0d284611a.