# Reproducibility record for Paper 8 ## Artifact and status boundary This record accompanies the internally accepted manuscript "Bidirectional Schema Mappings and Safe Update Propagation in CatDB." The paper artifact is accepted; its CatDB operational status remains `OBSTRUCTED`. The paper presents an evidence-backed design and falsification contract. It does not report a CatDB write-profile implementation, putback interpreter, authority evaluator, transaction compiler, P8 fixture corpus, P8 Haskell oracle, P8 Lean theorem, or P8 benchmark. The current bounded finite `\Delta_F` implementation is read-side migration evidence only. It is not evidence of safe write-back. The manuscript was authored only from: - `research/evidence-dossier.md` in this paper directory; and - shared repository research and coordination ledgers named by that dossier. No runtime source, fixture result, database execution, or benchmark was promoted into a P8 result. ## Source artifacts The manuscript source set is: ```text papers/08-bidirectional-mappings/paper.tex papers/08-bidirectional-mappings/references.bib papers/08-bidirectional-mappings/reproducibility.md papers/08-bidirectional-mappings/humanizer-scope.md papers/08-bidirectional-mappings/research/evidence-dossier.md ``` The binding shared ledgers include: ```text research/claim-register.md research/terminology.md research/paper-manifest.md research/paper-dependency-graph.md research/prior-art-map.md research/primary-source-bibliography.md research/formalization-roadmap.md research/implementation-roadmap.md research/benchmark-plan.md coordination/research-ledger.md coordination/verification-status.md coordination/open-questions.md coordination/review-conflicts.md coordination/agent-decisions.md ``` ## Isolated manuscript build The checked build does not place a PDF, cover, or LaTeX auxiliary files in the repository. From the repository root: ```bash build_dir=$(mktemp -d) cp papers/08-bidirectional-mappings/paper.tex "$build_dir/" cp papers/08-bidirectional-mappings/references.bib "$build_dir/" cd "$build_dir" pdflatex -interaction=nonstopmode -halt-on-error paper.tex bibtex paper pdflatex -interaction=nonstopmode -halt-on-error paper.tex pdflatex -interaction=nonstopmode -halt-on-error paper.tex pdflatex -interaction=nonstopmode -halt-on-error paper.tex pdfinfo paper.pdf ``` The expected acceptance checks for this provisional build are: - the initial `pdflatex`, `bibtex`, and three resolving `pdflatex` passes all exit zero; - `pdfinfo` reports at least 20 pages; - the final log contains no undefined references or citations; - the final log contains no LaTeX errors; - representative rendered pages show no clipping or malformed tables; and - no generated PDF or cover is copied into the repository. Warnings must be reviewed rather than hidden. A warning about an intentionally long URL is not equivalent to an undefined citation. An overfull box or broken longtable is a manuscript defect and should be corrected. ## Evidence needed to reproduce future P8 claims The paper preregisters, but does not yet execute, the following gates: 1. Freeze a closed `write-profile-v1` schema, update language, equality model, diagnostic registry, and source-capability vocabulary. 2. Create positive, negative, unsupported, conflict, and reconciliation-dependent fixtures with hand-computed source post-states. 3. Implement independent Haskell and Rust evaluators and compare normalized decisions for every closed fixture. 4. Represent selected pure finite definitions and laws in Lean with no unreported `sorry` or axioms. 5. Compile supported local cases to SQLite and PostgreSQL transactions. 6. Recompute the canonical view and diff the full declared source footprint after each accepted write. 7. Inject stale versions, hidden trigger effects, statement failures, unknown remote outcomes, duplicate delivery, capability drift, and partial commit. 8. Compare only equal-result, equal-isolation, equal-constraint, and equal-failure-semantics baselines. 9. Retain raw results and minimized failures as immutable evidence. 10. Obtain independent code review for relied-on code and AGY scientific review for the exact manuscript digest. Until those artifacts exist and pass, the claim table in the manuscript remains `OPEN` or `OBSTRUCTED`. ## Planned result directories The dossier reserves the following run layout: ```text benchmarks/runs//p08-environment.json benchmarks/runs//p08-preregistration.md benchmarks/runs//p08-workload-manifest.json benchmarks/runs//p08-classification.jsonl benchmarks/runs//p08-decisions.jsonl benchmarks/runs//p08-physical-plans/ benchmarks/runs//p08-source-before/ benchmarks/runs//p08-source-after/ benchmarks/runs//p08-view-recompute.jsonl benchmarks/runs//p08-conflicts.jsonl benchmarks/runs//p08-failures.jsonl benchmarks/runs//p08-provenance.jsonl benchmarks/runs//p08-metrics.parquet benchmarks/runs//p08-analysis.md ``` These paths are planned artifacts, not evidence that a P8 run exists. ## Review status AGY round 1 reviewed combined source digest `853e57ae220dabeb35002fbbb798004e05485af450a00af7d6276fed9767e3fe`. It found no blocker or major issue and requested one minor notation correction: the PutPut law had used an undeclared bare \(p\) instead of the paper's `\putstate` macro. Round 2 reviewed the corrected combined digest `b87bc237d76be9d86ca0cd3fd6aba5b9aec1210e8b8ede58f5fec57c7d38f542`, verified the correction, found no remaining blocker, major, or minor issue, and ended `VERDICT: ACCEPT`. ## Final accepted-artifact observation The final author source and an independent clean rebuild on 23 July 2026 produced the following bounded result: | Check | Result | |---|---:| | `paper.tex` SHA-256 | `6e511e130b32fcbccdc1c89f691f73a003b5d7cdf08e6165a61c191c345cd696` | | `references.bib` SHA-256 | `b92380c6f4d5ee113424a4966690a85d10a4d81e4a1355e997c966c5d20f06ed` | | combined source digest | `b87bc237d76be9d86ca0cd3fd6aba5b9aec1210e8b8ede58f5fec57c7d38f542` | | repository `paper.pdf` SHA-256 | `76269a0e835202451de81770793d1eb2f0bb4ca55d601bde9f412031cece6c3f` | | repository `paper.log` SHA-256 | `25a15d7a1883deb0c880f5d7dea992cd28ee91ec0d503a2453ae683cff039745` | | repository 300-DPI `cover.png` SHA-256 | `30bb74af47ccee2ebbb7959b9e54264ec6c8c9a734647414b650fc7383f797ce` | | canonical AGY review SHA-256 | `1a765d8822b6803cb39339e09b4e14046f2ded4aa9b6b31ac8b7abf38d4a03f7` | | Pages | 45 | | Bibliography entries | 26 | | LaTeX errors | 0 | | Undefined citations or references | 0 | | BibTeX warnings | 0 | | Overfull boxes | 0 | | Duplicate PDF destinations | 0 | The build ran in a fresh temporary directory. The repository retains only the accepted PDF, inspected log, and 300-DPI first-page cover, not auxiliary TeX files. The final verifier passed every source, freshness, page-count, LaTeX-log, cover-DPI, review-digest, and terminal-verdict check. The paper's acceptance does not promote any P8 runtime, theorem, or experiment.