Files
ThothII/docs/testing/evidence-restructuring-manual.md
T

155 lines
8.8 KiB
Markdown

# Evidence restructuring: owner migration gate
Status: **PENDING OWNER AUTHORIZATION**. This guide records the manual work that must
occur only after the owner authorizes a migration window and exact PSD target branch.
The automated runner is hermetic: it uses a fake restructurer and temporary inputs.
The owner-gate inventory below is a separately authorized, read-only PSD snapshot;
it did not write, stage, branch, commit, migrate, activate, push, or read secrets.
## Recorded automated boundary
Run from ThothII:
```bash
bash scripts/evidence-restructuring-acceptance.sh
```
It proves the local contracts with an intentionally badly structured fixture: typed
splitting, review-item blocking, one-source membership, Git-visible proposals and
recoverability, no-op reruns, dirty-state refusal, pipeline-version refusal plus full
`--upgrade`, and orphan blocking. It also runs the hermetic authoring, canonical-kind,
formula, chunking, candidate-evaluation, hybrid-query/fail-closed, and pinned-Qdrant
L0 suites. The runner supplies only a `mktemp` workspace and asserts the ThothII
worktree is unchanged; consequently it performs no external PSD write.
The real Pi invocation is deliberately not automated here. A reviewer must run it once
per changed source after the authorization gate and examine every proposed curated file.
## Owner-gate package (issue #46)
Read-only snapshot collected 2026-08-25:
- PSD repository: `/Users/mp/projects/tht-workspace-psd`, clean before and after the
inspection; immutable pre-migration commit
`47516f85b4db4a67cfa8a86cea4cb2e7b98c5813`.
- proposed PSD branch name: `codex/evidence-restructuring-psd` (proposal only; no
external branch has been created);
- rollback command for the owner to use only if the later authorized migration must be
undone:
```bash
git -C /Users/mp/projects/tht-workspace-psd reset --hard 47516f85b4db4a67cfa8a86cea4cb2e7b98c5813
```
- exact source inventory (36 files; `evidence/README.md` is not a source):
```text
psd-clinical/evidence/00-glossario/coorti-universi-pazienti.md
psd-clinical/evidence/00-glossario/glossario-termini-analitici.md
psd-clinical/evidence/00-glossario/glossario-termini-clinici.md
psd-clinical/evidence/00-glossario/glossario-termini-dwh.md
psd-clinical/evidence/00-glossario/note-di-lettura.md
psd-clinical/evidence/00-glossario/tassonomia-eventi.md
psd-clinical/evidence/10-domini-clinici/ablazione.md
psd-clinical/evidence/10-domini-clinici/anagrafica-paziente.md
psd-clinical/evidence/10-domini-clinici/cardioversione-elettrica.md
psd-clinical/evidence/10-domini-clinici/chiusura-auricola.md
psd-clinical/evidence/10-domini-clinici/documenti-clinici.md
psd-clinical/evidence/10-domini-clinici/genetica-clinica.md
psd-clinical/evidence/10-domini-clinici/icd.md
psd-clinical/evidence/10-domini-clinici/ilr.md
psd-clinical/evidence/10-domini-clinici/pacemaker.md
psd-clinical/evidence/10-domini-clinici/pm-icd-altro.md
psd-clinical/evidence/10-domini-clinici/visite-cardiologiche-genetiche.md
psd-clinical/evidence/20-valori-enum/enum-flag-booleani.md
psd-clinical/evidence/20-valori-enum/enum-flag-note-sn.md
psd-clinical/evidence/20-valori-enum/enum-innesto.md
psd-clinical/evidence/20-valori-enum/enum-isteresi.md
psd-clinical/evidence/20-valori-enum/enum-tipo-intervento.md
psd-clinical/evidence/30-esempi-nlq/nlq-ablazione.md
psd-clinical/evidence/30-esempi-nlq/nlq-cardioversione.md
psd-clinical/evidence/30-esempi-nlq/nlq-device-pacemaker-icd.md
psd-clinical/evidence/30-esempi-nlq/nlq-percorso-paziente.md
psd-clinical/evidence/30-esempi-nlq/nlq-studio-elettrofisiologico.md
psd-clinical/evidence/40-mapping-semantico/catena-staging-integration-dwh.md
psd-clinical/evidence/40-mapping-semantico/matrice-viewpoint-dominio.md
psd-clinical/evidence/40-mapping-semantico/registry-testo-clinico-fact-clinical-event-text.md
psd-clinical/evidence/40-mapping-semantico/regole-classificazione-fact-dim-bridge.md
psd-clinical/evidence/40-mapping-semantico/trasformazioni-valori-mapping-colonne.md
psd-clinical/evidence/50-metadati-normalizzazione/normalizzatori-valori-dwh.md
psd-clinical/evidence/50-metadati-normalizzazione/normalizzazione-codici-paziente-medici.md
psd-clinical/evidence/50-metadati-normalizzazione/normalizzazione-device-cied.md
```
The read-only Qdrant observation for the `psd-clinical` collection on the legacy PSD
bind `127.0.0.1:6333` was 163 `schema_table`, 2,275 `schema_column`, 2 `memory`, and
1 `solved_question` point. Representative IDs are:
| Kind | Sample IDs |
| --- | --- |
| `schema_table` | `01bc2535-24d6-5722-a58b-64a122b90b36`, `024ddf60-c80d-5223-ac06-9247e7de7027`, `086e00c8-b4a9-5063-b82d-e5a67add29cd` |
| `schema_column` | `00126cc1-7564-521a-a084-c2d670263258`, `00365200-2c55-5bb4-86bb-87dd2d1bb529`, `004304bc-b543-5b2d-8b40-f18da8e82af7` |
| `memory` | `8d5cd772-563a-5e22-b764-2ca76cf6efca`, `db74457a-3de8-5b95-9a31-d28a1ecf8141` |
| `solved_question` | `2b6bb7d2-1a35-5f49-bd8d-b0cdcb98459a` |
`tht ... workspace vector inspect --json` was attempted read-only but was blocked by
the local maintenance image's missing production `auth.yaml` / `AUTH_MODE=upstream`.
The counts above therefore come from the Qdrant read-only scroll endpoint and must be
repeated through the successful `vector inspect` command immediately before the future
authorized preprocessing action. No secret was read to bypass that guard.
### Reproducible no-write record
The inspection used only these read operations; `git status --porcelain` was empty both
before and after, and `git diff --quiet` succeeded after:
```bash
git -C /Users/mp/projects/tht-workspace-psd rev-parse HEAD
git -C /Users/mp/projects/tht-workspace-psd status --porcelain
git -C /Users/mp/projects/tht-workspace-psd ls-tree -r --name-only HEAD -- psd-clinical/evidence
node -e 'fetch("http://127.0.0.1:6333/collections/psd-clinical/points/scroll", {method:"POST", headers:{"content-type":"application/json"}, body:JSON.stringify({limit:4096,with_payload:true,with_vector:false})}).then(r => r.json()).then(x => console.log(x.result.points.length))'
git -C /Users/mp/projects/tht-workspace-psd diff --quiet
```
Before any PSD write, provide the owner this inventory, SHA, rollback command, baseline
counts/IDs, clean ThothII commit, and exact local-gate output. Issue #47 still owns all
external mutations and manual acceptance.
## Manual acceptance after authorization (issue #47)
Record a separate PASS/FAIL and evidence for each item; never substitute an automated
test for a human Git review.
1. **Authoring and Git review.** In the authorized PSD clone, move only the approved
36 source paths to `evidence/source/`, run `tht evidence prepare`, verify exactly
one no-tool/no-session Pi call per changed source, inspect the Git diff, correct
every `review_item`, run `tht evidence validate`, and obtain the normal human Git
review. Confirm source renames/reclassifications retain IDs, semantic splits receive
new IDs, IDs use `evidence:<slug>`, and no orphan is deleted automatically.
2. **Additive BM25 schema upgrade.** Before preprocessing, run vector inspect and save
configuration, counts, and IDs. Run `workspace preprocess evidence`; verify the
unnamed dense vector remains and only `bm25` with IDF is added. Do not accept a
destructive rebuild, vector rename, or fallback engine.
3. **Schema and Memory non-regression.** Compare before/after counts and the saved
representative IDs for `schema_table`, `schema_column`, `memory`, and
`solved_question`; repeat dense Schema and Memory searches and attach the results.
4. **Preprocessing publication.** Confirm a validated corpus builds an inactive
candidate, evaluates that exact generation, and switches active generation only
after every evaluation query has an expected ID in the first ten fused hits. Attach
dense, BM25, and fused ranks for lexical, semantic, and mixed queries.
5. **Hybrid and formula retrieval.** Check dense and BM25 receive the identical NFC /
newline / outer-trim-only query text. Confirm Formula Evidence accepts a PostgreSQL
expression but rejects a full query, and retrieve one approved formula by its typed
Evidence path.
6. **Empty versus blocking unavailable.** Record one available empty retrieval and one
controlled Qdrant failure. The first may continue; the second must block the stage
without stale generation or purpose fallback.
7. **Complete session behavior.** Walk through `clarification`, `rewriting`,
`schema_linking`, `cte`, and `final_sql`; confirm independently persisted minimal
receipts. Confirm neither `memory` nor `synthesis` invokes Evidence search and that
session formula proposals remain unpublished.
Write the reviewer identity, UTC time, commit IDs, command output locations, and one
final `manual acceptance: PASS` or `manual acceptance: FAIL` line when (and only when)
the authorized walkthrough is complete.