Context
The Approve+Follow-Up disposition from PR #16726 (merged @ e4b36b45f3, resolving #16507): the continuity guard's red-control test stages a deliberate whole-body clear (dock host held transparent across 10 rAF frames mid-screencast) and proves the guard convicts it. Its restore assertion then checks that the TARGETED surface recovered — getComputedStyle(host).opacity === '1' — and nothing else. The reviewer dispositioned this as follow-up, not Required Action (cycle-2, pullrequestreview-4889442963): the exposure is bounded (a single CSS property with no state feedback path), but the assertion class is the known-weak one.
Filed by the fixture's author per that disposition. Live latest-open sweep: latest 20 checked at filing, no equivalent; A2A claim scan clean.
The Problem
A plausible-but-wrong restore would pass the current assertion. The precedent is #15178 / PR #15180: a data-layer-correct re-materialization rendered a false surface (same name, same position, different identity, lost instance state) past a "verified safe" review — the transferable law being that a projection which re-renders into a plausible-but-wrong state is invisible to any assertion that only checks the targeted row. Applied here: if the staged clear's restore left any untargeted workspace state degraded (a pane re-rendered wrong, content lost while the host's own opacity reads '1'), the fixture would report green while having damaged the very surface whose continuity it exists to guard.
The Architectural Reality
test/playwright/e2e/workstation/WorkstationFiveBeatNL.spec.mjs — the red-control test (film-take self-excluded); captureWorkspaceContinuity + assertWorkspaceContinuity are the shipped instruments, already capable of the needed measurement.- The healthy population is measured: baselines ≈5.30–5.35; the staged-clear minimum measured 1.4576 against floor ≈3.447 in both the author run and the reviewer's independent falsifier run.
The Fix
After the staged clear's restore, run a second continuity capture over a short settle window and assert the workspace measures baseline-class again (minEntropy ≥ baselineEntropy * 0.65 via the existing helper, expectedCleared: false) — the untargeted-surface discipline: the restore is proven by the same instrument that proved the clear, not by reading back the property the fixture itself wrote.
Acceptance Criteria
Out of Scope
- The guard's floor provenance and red-control conviction (shipped in PR
#16726).
- Pane-scoped or semantic continuity witnesses (
#16498-family surfaces).
Decision Record impact
none.
Related
#16507 / PR #16726 (the shipped red-control + the reviewer disposition), #15178 / PR #15180 (the plausible-but-wrong re-render precedent), #15252 (film epic).
Origin Session ID: 7e8a0e84-6733-474e-865e-1757feb4b5f8
Retrieval Hint: query_raw_memories("red-control restore assertion untargeted surface baseline-class post-restore continuity capture")
Context
The Approve+Follow-Up disposition from PR
#16726(merged @e4b36b45f3, resolving#16507): the continuity guard's red-control test stages a deliberate whole-body clear (dock host held transparent across 10 rAF frames mid-screencast) and proves the guard convicts it. Its restore assertion then checks that the TARGETED surface recovered —getComputedStyle(host).opacity === '1'— and nothing else. The reviewer dispositioned this as follow-up, not Required Action (cycle-2, pullrequestreview-4889442963): the exposure is bounded (a single CSS property with no state feedback path), but the assertion class is the known-weak one.Filed by the fixture's author per that disposition. Live latest-open sweep: latest 20 checked at filing, no equivalent; A2A claim scan clean.
The Problem
A plausible-but-wrong restore would pass the current assertion. The precedent is
#15178/ PR#15180: a data-layer-correct re-materialization rendered a false surface (same name, same position, different identity, lost instance state) past a "verified safe" review — the transferable law being that a projection which re-renders into a plausible-but-wrong state is invisible to any assertion that only checks the targeted row. Applied here: if the staged clear's restore left any untargeted workspace state degraded (a pane re-rendered wrong, content lost while the host's own opacity reads'1'), the fixture would report green while having damaged the very surface whose continuity it exists to guard.The Architectural Reality
test/playwright/e2e/workstation/WorkstationFiveBeatNL.spec.mjs— the red-control test (film-take self-excluded);captureWorkspaceContinuity+assertWorkspaceContinuityare the shipped instruments, already capable of the needed measurement.The Fix
After the staged clear's restore, run a second continuity capture over a short settle window and assert the workspace measures baseline-class again (
minEntropy ≥ baselineEntropy * 0.65via the existing helper,expectedCleared: false) — the untargeted-surface discipline: the restore is proven by the same instrument that proved the clear, not by reading back the property the fixture itself wrote.Acceptance Criteria
Out of Scope
#16726).#16498-family surfaces).Decision Record impact
none.
Related
#16507/ PR#16726(the shipped red-control + the reviewer disposition),#15178/ PR#15180(the plausible-but-wrong re-render precedent),#15252(film epic).Origin Session ID: 7e8a0e84-6733-474e-865e-1757feb4b5f8
Retrieval Hint:
query_raw_memories("red-control restore assertion untargeted surface baseline-class post-restore continuity capture")