]> dreyeck.freedombox.rocks Git - topic-maps.git/blob - docs/slice-20e-empty2-registration.md
Implement noncyclic XTM mergeMap content import (Slice 22E)
[topic-maps.git] / docs / slice-20e-empty2-registration.md
1 # Slice 20E: register the original empty2 contract
2
3 This slice registers only ctm/in/empty2.ctm after all retained 20A/20B/20C
4 stages, reusing the Slice 20A exercise-case function and the same discovered
5 Case Record. No CTM reader, TMDM or CXTM implementation changes are required.
6
7 The original source is exactly "%version 1.0" followed by LF (13 bytes).
8 Its baseline is "<topicMap>" LF "</topicMap>" LF (23 bytes).
9 CTM 3.5 [44]/[45] and 3.7 [48] permit this prolog without later constructs:
10 https://www.isotopicmaps.org/ctm/ctm.html .
11 The pinned corpus retains its historical 2010-03-31 draft qualification.
12
13 The existing original-execution function deserializes the unchanged file with
14 the discovered Document IRI. Its shared empty-map witness checks zero Topics
15 and Associations. It compares canonical octets with the original baseline,
16 checks raw model state and retained handles across canonicalization, checks
17 determinism and unchanged source bytes, and records PASS only for an actual
18 byte match. The new stage additionally checks the precise 13/23-byte vectors,
19 Case Record identity, stable previous observations and exactly one status
20 transition. ASDF topic-maps/ctm/empty2-tests is invoked unconditionally by
21 scripts/fresh-sbcl.lisp, after the historical stage reports are written.
22
23 Reproduce the complete offline acceptance run with an existing SBCL 2.5.4:
24
25 python3 scripts/validate-slice-20e.py --sbcl /absolute/path/to/sbcl --output /private/tmp/new-slice-20e-report
26
27 The output directory must be new. The validator uses verified local Store
28 sources and the existing XML patches without downloads. Fresh SBCL processes
29 run with network denied. It runs TMDM/CTM/CXTM, XML qualification, standalone
30 XTM, XTM Mergemap, Slice 20A/20B/20C and the new original registration, plus the
31 existing Python tests and NFC oracle. It verifies the pinned corpus and hashes
32 every historical report and all protected repository files. Historical stage
33 reports are regenerated separately and checked against their prior provenance.
34
35 The run's actual results, the original comparison, source hashes and commands
36 are in reports/slice-20e/execution.json; full process output is retained in
37 the adjacent logs. current.registration.json records the executed checks,
38 current.cxtm.json contains actual and expected streams, and
39 current.validation.json records the final corpus statuses. These files are
40 written only by the successful execution; this document does not predeclare
41 the expected 256 PASS / 4 FAIL / 13 NOT-RUN as an observed result.
42
43 The default tool sandbox could not start because the configured writable root
44 /Users/rgb/.nix-profile is a symlink. Execution used the regular approved
45 require_escalated path; neither that symlink nor sandbox configuration was
46 changed. No persistent Lisp image, HyperDoc content, original corpus source,
47 baseline or historical report is modified. No push or deployment is performed.