Round 10: CR_013 classified evidence rows
Inventory and review boundary
Current requirements, test designs/executions, system/software elements, architecture/Draw.io exports, interfaces, dependency manifests, state machines, implementation/verification plans and source/review registers form the context. The scoped inventory identifies changed/accepted content separately from prior accepted or still-draft context; a hash alone is not evidence of review. Unrelated PRSA gates, later occurrence mode, arbitrary manual positioning and new release qualification are excluded without declaring them resolved.
Changed requirement paths: Position test evidence in c... (STV_SHR_012), Keep navigation positions s... (STV_SHR_054), Preserve semantic layout (STV_SYS_005), Place classified test evide... (STV_SWR_012), Honor resolved declared lev... (STV_SWR_037), Order controls using target... (STV_SWR_055), Stabilize committed navigat... (STV_SWR_058), Resolve configurable per-ob... (STV_SWR_061). Retained context includes Preserve requirement levels (STV_SHR_009), Position system architecture (STV_SHR_010), Position software architecture (STV_SHR_011), Use a replaceable input dat... (STV_SHR_014), Integrate the host data model (STV_SYS_002), Normalize replacement mappings (STV_SWR_014), Center parent boxes with co... (STV_SWR_056), Render complete edge bounds... (STV_SWR_059), Use natural ID order for fr... (STV_SWR_060). Component owners remain Host model normalization (STV_SWE_MODEL), Graph adapter and indexes (STV_SWE_GRAPH), Semantic rank layout (STV_SWE_LAYOUT), Accessible view and viewport (STV_SWE_RENDER). Test designs: Validate classified placeme... (STV_TC_SW_046), Resolve stable classificati... (STV_TC_SW_047), Align classified rows throu... (STV_TC_SW_048), Published nine-row evidence... (STV_TC_SY_061); prior legacy and PR013 cases remain regression context, with their historical executions preserved.
Findings and design resolutions
STV_R10_001 (requirements conflict): Mandatory right-side tests contradict nine target rows. Revised requirements distinguish classified ordinary rows from an explicitly configured compatibility role; ADR 009 partially supersedes ADR 008’s test-lane assumptions while retaining stable navigation.
STV_R10_002 (classification ownership): Neighbor/ID inference would move shared tests or depend on naming. Configurable tag mapping and complete-graph one-hop owner mapping resolve placement at build time, with additive object fields consumed by both layout and control ordering. No duplicate host types.
STV_R10_003 (exception policy): Strict failure was proposed; the owner selected fallback retention. A common labelled fallback below software architecture and above problems handles ambiguous/missing records. Invalid configuration remains an error. Hidden owners cannot leak or reclassify nodes.
STV_R10_004 (test completeness): Four concrete designs cover model paths, actual classification/hidden/mixed-owner cases, layout/control/stability and Firefox file/HTTP entry from software/cases/executions. Legacy regression and clean installed distribution are separate gates, not inferred passes.
Semantic and feasibility assessment
SYS 005 retains the observable row boundary; SWR 012/037 own browser projection. SWR 061 derives from SYS 002 and SHR 012, allocated to build-side MODEL/GRAPH, which already own normalization/snapshot publication. Existing containment is consistent for actual system parents. No artificial component or system requirement is introduced. The unchanged stack supports Python dictionary/set classification and JS resolved-level lookup. Indexed two-pass resolution avoids per-execution whole-graph scans; cyclic/chained data links use fallback under the one-hop contract, rather than recursively guessing classifications.
Designs distinguish missing classification data (retained fallback) from invalid configuration (actionable build error), specify row alignment versus child-span centering, and retain transactional scene/geometry ownership. No new browser state or timer is needed. The build snapshot action and software flow labels are refined in editable diagrams, preserving routes and component boundaries.
Review decision and handover
The clean scoped concept is accepted under the owner’s recorded delegation. Accepted internal requirements: Preserve semantic layout (STV_SYS_005), Place classified test evide... (STV_SWR_012), Honor resolved declared lev... (STV_SWR_037), Order controls using target... (STV_SWR_055), Stabilize committed navigat... (STV_SWR_058), Resolve configurable per-ob... (STV_SWR_061). Stakeholder revision Position test evidence in c... (STV_SHR_012) and the classified/legacy clarification of Keep navigation positions s... (STV_SHR_054) cite the explicit owner CR_013 implementation instruction and fallback decision, not inferred approval from a build. ADR 009 is accepted; ADR 008’s reciprocal partial supersession preserves all stable-navigation policy.
The four new cases above are accepted as designs, still not_run with no fabricated executions. Changed ordinary documents (interfaces, architecture, state actions, implementation/verification plans, technology and demo guidance) and the software/state Draw.io/SVG pairs are accepted against the concept-gate inventory (retained in the engineering checkout). Unchanged context retains its own prior accepted/draft states; no blanket reacceptance is claimed.
All four findings are closed for concept scope. Initial Sphinx validation exposed an ADR entry appended outside the toctree; it was moved into the index and the warnings-as-errors rebuild passed. Complete –require-test-coverage validation passed: 55 SHR, 10 SYS, 61 SWR, 109 cases, 600 executions. CSV paths were exported and reviewed for the revised paths. System/software/state SVGs and their embedded pages were visually inspected at reading scale; routes, arrowheads and labels are readable. No new boundary or missing core policy remains. Runtime evidence is explicitly not assessed at this gate.
Implementation handover: apply the model selectors and two-pass graph contract, share resolved-level lookup between layout and controls, label fallback nodes, configure the project’s nine rows, and execute the four new procedures plus legacy, PR013, packaging and capacity regression gates. No release is authorized.
Implementation feedback re-review
STV_R10_005 (closed): A malformed data link to another owner-derived record must not bypass the owner’s fallback policy. Such chains/cycles now use fallback (or the documented strict mode when fallback is absent), with no recursive inference. Invalid classifier configuration still fails early. Updated graph cases cover cycles, hidden/mixed owners and input order. The same scoped design is reaffirmed; no stakeholder outcome or default model changes.
A large graph case additionally resolves 30,000 classified case/execution records and 15,000 owner links. It asserts actual boundary assignments without inventing a performance budget. The revised SW_047 design is reaffirmed under the same review delegation; runtime outcome remains dependent on the actual runner report.
Final implementation disposition
The scoped implementation and feedback re-review are complete. Four accepted canonical procedures now have actual passing aggregate executions. The full 134-test development selection passed, including 30,000-record classification, real Firefox file/HTTP nine-row navigation, legacy/PR013 regression, clean installed packaging and rebuilt browser capacity. The five review findings are closed; no unresolved design/implementation blocker remains in CR_013. Verification-state updates are limited to the revised eight requirement paths. The normal embedded project demo uses the classified model and owner-selected fallback. Final publication/evidence validation is retained in CR_013 classified evidence rows: development verification; no release qualification is inferred.
The final inventory (retained in the engineering checkout) records current content after source ownership, fallback clarification and execution links. The original inventory preserves the concept-gate baseline. Unchanged context and unrelated draft security artifacts retain their earlier acceptance states.
Final checks passed: warnings-as-errors Sphinx; full –require-test-coverage validation with 55 SHR, 10 SYS, 61 SWR, 109 cases and 604 executions; 134 executable tests mapped without unknown IDs; 2/9/11 published diagram links. No acceptance-blocking finding remains in the scoped final inventory.