acdream/docs/research/2026-08-28-owner-closed-issue-validity-audit.md

# Owner-closed issue validity audit — 2026-08-28

## Scope and verdicts

This is a read-only companion to the
[open-entry validity audit](2026-08-28-open-issue-validity-audit.md). It checks
the exact **57 unique issue IDs** in the owner's 2026-08-28 closure request.
Fifty-six entries were newly marked closed; `#167` was already `DONE` before
that request. This audit changes no issue status and fixes no product code.

- **OWNER-CONFIRMED CLOSED DESPITE REMAINING EVIDENCE** means current source or
  still-applicable recorded evidence supports unresolved work, but the owner
  explicitly chose on 2026-08-28 to keep the entry closed. This records a
  scope/status decision, not technical proof that the symptom is fixed.
- **VERIFY BEFORE KEEPING CLOSED** means the historical report was credible,
  but later architectural changes or an implemented fix make a current live
  gate necessary before deciding whether the closure is correct.
- **SAFE TO REMAIN CLOSED** means current code/tests and later project history
  show the issue fixed, superseded, disproved, owner-accepted, or not a valid
  product defect in the first place.

The audit inspected current source and focused tests, each issue's complete
evidence trail, later campaign/user-gate records, and the current architecture.
The supporting focused Release tests completed with **321 passed, 1 skipped,
0 failed**. The companion open-entry audit's clean Release build completed with
**0 warnings and 0 errors**.

## Owner-confirmed closed despite remaining evidence — 3

| ID | Remaining technical evidence |
|---:|---|
| #64 | The local pickup transaction sends/accepts the pickup, but the current local-player animation path still has no pickup one-shot trigger and no admitted server-motion equivalent. The original missing local bend/grab animation remains an unimplemented path. |
| #235 | The live animation presenter still publishes part poses at admitted object-clock frames. Root presentation is smoothed, but no render-time interpolation exists between animated part poses, which is the exact capped/RDP cadence alias described by the issue. |
| #263 | The Drudge Scrying Orb residual was user-confirmed after the general translucency fix and explicitly deferred. No later orb-specific ClipMap/order fix or clearing visual gate exists. |

## Verify before keeping closed — 11

| ID | What is known and what should be gated |
|---:|---|
| #41 | The remote-root-motion correction was implemented, but the entry never received its final two-client sub-decimeter-blip visual gate. Observe another player starting/stopping/turning on the current binary. |
| #94 | The original held-light report became testable only after child parenting landed. Current lighting distinguishes authored point and spot lights, but there is no post-parenting side-by-side reproduction or clearance of the wall spotlight symptom. |
| #115 | The cramped-interior camera feel report was credible, but camera collision, viewer stepping, display pacing, and the renderer changed substantially afterward. Repeat the same pressed-wall turns against retail. |
| #129 | The long-distance door/doorway leak fix shipped, but the entry retained an explicit distant-view visual gate. Revisit the named over-one-landblock views before treating it as conclusively closed. |
| #165 | Remote wall penetration was observed, while the leading physics explanations were later ruled out and the remaining presentation hypothesis was never proven. A current observer/client reproduction decides it. |
| #181 | The Facility Hub artifact survived its original seven-fix ladder, but its early attribution was partly retracted and the camera/render pipeline later changed. Repeat the exact pressed-camera pose without capture overlays. |
| #208 | The peaceful-login combat-bar correction is present, but the issue's final fresh-login visual check was never recorded. Gate a fresh peaceful login and one relog. |
| #225 | The particle/translucent-object ordering implementation and stress tests pass, but the issue explicitly retained its final lifestone/candle visual matrix. Run that matrix; `#263` remains a separate orb-specific residual. |
| #337 | The Neftet rock behavior is covered by the `#333` broadphase fix, but the exact rock-plateau live acceptance was never recorded. Re-test that fixture before relying only on the general mechanism. |
| #409 | The client-wide tooltip system is code-complete and one live path was proven, but the remaining connected hover paths in its acceptance matrix were still owed. Complete that small UI gate. |
| #418 | The original 27 s/32-blocks-per-second limiter was fixed and reduced dramatically, so the title is stale. Cold-login readiness was still above the retail-feel target at the last measured checkpoint, and later PAK/runtime work was not re-measured with the same probe; run one current cold-login timing gate. |

## Safe to remain closed — 43

| ID | Why the closure is safe |
|---:|---|
| #102 | Fixed: `PortalVisibilityBuilder` now uses retail's in-place `update_count` watermark/fixpoint propagation; the old per-cell cap is gone and focused convergence tests pass. |
| #104 | Fixed: VFX instances carry owner-cell identity and `ParticleVisibilityController` gates them against the current PView visible-cell set. |
| #114 | Superseded by the current indoor renderer: the old `gl_ClipDistance` shell-chop mechanism named by the issue was deliberately deleted in favor of whole-shell depth/order handling. |
| #116 | The second divergence was fixed; the remaining first shape was traced to a simplified fixture while the real Path-5/Path-6 behavior is retail-exact. No confirmed engine defect remains. |
| #120 | Fixed: the reciprocal/in-place convergence case is pinned by the current portal-visibility replay tests. |
| #121 | Fixed and superseded by later portal campaigns/gates; world-portal VFX are routed through the current scene-particle path rather than the broken visibility path described here. |
| #139 | Fixed: retained buttons and menus now implement the authored normal, rollover, pressed, highlight, and open-state transitions with regression coverage. |
| #144 | Not an established defect: the empty-slot release is a harmless no-op and the entry never established that retail suppresses the click notification differently. |
| #146 | Non-actionable polish list: the capacity-bar visuals were already accepted and the entry asks for unspecified additional polish rather than describing a remaining defect. |
| #148 | Fixed: the authored backpack button toggles the inventory panel and projects open/closed state back to Highlight/Normal, with tests. |
| #151 | Fixed for the reported defect: the Arwic/city wall collision was user-verified. The residual terrain note was a question, not a demonstrated mismatch. |
| #158 | Not a bug by its own description; it is an old, vague character-window polish list against presentation paths that were subsequently replaced. |
| #166 | Fixed and user-gated as part of the closed Campaign P landing-momentum/bounce family. |
| #167 | Already `DONE` before the cleanup request; the retail leash behavior and constants were ported and covered. |
| #172 | Fixed and covered by the user-accepted Campaign P portal-platform/step-up gate. |
| #173 | Fixed and covered by the user-accepted Campaign P remote ceiling-response gate. |
| #174 | Fixed: door-use animation-link queue handling shipped and the later physics/door gate was accepted. |
| #175 | Fixed: door collision now follows the closed motion-table pose; the later physics/door gate was accepted. |
| #176 | The original purple flashing/lighting defect was fixed and verified. Its distinct pressed-camera residual was split into `#181`, so keeping this parent closed is correct. |
| #182 | Fixed by the retail collision-velocity rebuild and covered by the accepted crowded-movement Campaign P matrix. |
| #191 | Fixed by the root-motion presentation path; current coverage asserts that a brief forward edge cannot glide when the animation contributes no root delta. |
| #193 | Fixed and measurement-verified: the retained-resource/OOM correction survived the extended-play resource gates. |
| #202 | Fixed: the current `WeenieErrorMessages` table and routing replace the old missing lookup and cover interpolation plus `UseDone` behavior. |
| #209 | Fixed and subsequently live-gated: the jump/combat power-bar modes and captions were accepted in the later keyboard/powerbar rounds. |
| #226 | Fixed and connected-visual verified: building/EnvCell detail-texture overlay is present in the modern renderer. |
| #253 | Superseded/disproved by later byte-exact retail-DAT research: the shared row template authors a flush-left 20×20 icon cell, and the current sprite draw preserves authored texture alpha. The earlier guessed centering offset was not retail ground truth. |
| #264 | Research note, not a confirmed client defect: it explicitly records no known `WATER_CONTACT_TS` divergence and asks for xref/swim investigation. Any discovered mismatch should be filed as a concrete new issue. |
| #273 | Fixed: the exact captured Holtburg gap fixture now blocks traversal, requires both real geometry participants, and matches graph/flat collision paths in focused tests. |
| #319 | Fixed and user-gated in placement Campaign C4 route 7: player-parented children receive canonical-cell propagation through the unified owner. |
| #333 | Fixed: the broadphase no longer rejects an off-centre BSP part by measuring reach only from the part origin; regression coverage pins the case. |
| #348 | Fixed: cursor objects are process-lifetime cached instead of being recreated on every alternation; later long connected sessions did not reproduce handle exhaustion. |
| #394 | Fixed: Configure Keyboard row captions use the authored serif DAT font, with current layout tests and later keyboard gates. |
| #395 | Fixed: key captions resolve through retail's DAT/localized key-name pipeline rather than raw enum spellings. |
| #414 | Fixed: logout teardown no longer leaves the cursor in raw-captured fly mode. |
| #415 | Fixed: deferred UI-probe wait verbs are bound independently of the optional artifact-directory setting. |
| #416 | Fixed: roster rollover state clears when the pointer leaves a row. |
| #417 | Fixed: ambience ownership is reset on world logout and does not continue/re-fire at character selection. |
| #419 | Fixed and explicitly owner-accepted: the portal rim/ring artifact disappeared after matching retail's view-plane exit animation. |
| #420 | Fixed: `UiButton` safely handles the character-select media-state case that previously produced the null-key crash. |
| #424 | Fixed: exclusive-fullscreen alt-tab no longer sends a zero-area extent into render-pack activation validation. |
| #426 | Fixed: NO_POS_UVS/solid-colour faces are extracted and rendered instead of being discarded as having no positive face. |
| #432 | Disproved as a production performance bug: the allocation/frame-time spike came from diagnostics being enabled, not the normal client path. |
| #443 | Fixed and owner-passed: paperdoll/private-viewport ownership now hydrates on first open without the intermittent missing/delayed doll. |

## Recommended action

1. Keep `#64`, `#235`, and `#263` closed per the owner's explicit post-audit
   decision. If one becomes a priority later, reopen it or file a fresh issue
   with a current reproduction.
2. Run one compact current-binary gate for the 11 verification items; keep each
   closed if the named route passes, otherwise reopen it with the new evidence.
3. Leave the remaining 43 closed. Do not resurrect their stale titles merely
   because their original status text once said `OPEN`.