]> dreyeck.freedombox.rocks Git - topic-maps.git/blob - docs/slice-19b-validation.md
docs(ctm): preserve Template Tail grammar investigation and evidence
[topic-maps.git] / docs / slice-19b-validation.md
1 # Slice 19B validation
2
3 Base: `9ce2a2b5116edfd921a5240efcb2f604f9523693`.
4 Branch: `codex/ctm-slice-19b`.
5 Pinned corpus: `cd165b150176c1d6842b37be9cc9051b40d4996b`.
6
7 Reproduce the full retained regression and new CTM Mergemap tests:
8
9 ```sh
10 cd /Users/rgb/workspace/topic-maps
11 PYTHONDONTWRITEBYTECODE=1 python3 scripts/validate-slice-19b.py \
12 --sbcl /nix/store/hlsiblxpxr1hax3dw9x8znd6zy097ddj-sbcl-2.5.4/bin/sbcl
13 ```
14
15 The validator runs `scripts/test.sh` in a fresh SBCL 2.5.4 process under macOS
16 sandbox-exec with network access denied. It retains the established source
17 registry/output translation settings without cache, relocation or additional
18 runtime work. The documented Lisp inspection is executed separately, and a
19 base lexer counterprobe verifies its former unsupported-directive response.
20
21 `reports/slice-19b/execution.json` records UTC execution times, exact commands,
22 exit status, all phase totals and source/artifact hashes. `tests.log` is the
23 actual complete run; `mergemap.json` separates original positives/negatives from
24 its focused checks. Four actual CXTM documents are preserved in `actual/`.
25 An independent Python checker compares original inputs, model identifier sets,
26 resource keys/IRIs and negative source positions. The existing CXTM checker
27 compares every executed positive baseline byte-for-byte and retains the two
28 DateTime mismatch records.
29
30 Only seven imported statuses may change: the four requested positive Mergemap
31 cases and three requested invalid cases. Expected aggregate:
32 **194 PASS (169 positives, 25 negatives), 2 known FAIL, 0 UNSUPPORTED, 77 NOT-RUN**.
33 Every other imported case record, the previous 167 CXTM records and all retained
34 model observations must match Slice 19A. All pinned sources, baselines,
35 references, TMDM and CXTM source files must remain unchanged.
36
37 Two older tests used unknown notation `urn:ctm` but expected all Mergemap syntax
38 to be unsupported. They now assert the specific unknown-notation rejection.
39 The Reference boundary label remains unchanged; the Template check is renamed
40 to `diagnostic:unknown-mergemap-notation`. Their phase totals and original corpus
41 contracts remain intact. The validator allows exactly this one report-label
42 replacement and no other retained-report drift.
43
44 Identity tests cover shared SIs, Topic IIs, SI/II overlap, SLs, dependent item
45 merging and retained target handles. Observations use the actual import's object
46 references. Additional tests cover resource resolution, namespace isolation,
47 Wildcard counters, repeated/nested imports, execution order, strict bytes/BOM,
48 resolver policy, diagnostic chains, metadata and deterministic pure CXTM.
49 Core, Reifier, Variant and identifier conflicts leave both graphs and their
50 CXTM unchanged at the import boundary. Non-Topic identifiers in synthetic model
51 tests are explicit TMDM API additions, not extra CTM syntax or corpus passes.
52
53 Results derive from execution evidence, not these expected counts. No persistent
54 Lisp image, HyperDoc, network resolver or server is used. Include and XTM
55 Mergemap contracts remain NOT-RUN, and no Slice 19C/19D work is included.