A deep Fable review of the real manifest corpus (READ-ONLY reference) showed the
manifests are an ENFORCEMENT PROGRAM, not an application inventory, and that
auto-seeding the Applications catalog from entry Type + Name was wrong:
- The catalog ALREADY tracks these apps from the classic-shopdb migration, with
version histories (PC - DMIS, UDC x11 versions, eMX / eDNC, CLM, CSF, Oracle
Database, FormTracePak). Seeding from manifest labels created DUPLICATES under
different names (PC-DMIS 2016 vs PC - DMIS; eDNC (bundles NTLARS) vs eMX / eDNC;
OpenText HostExplorer ShopFloor vs CSF). It also misclassified config drops
(eMxInfo.txt) as apps and could never match a PC's reported ARP name.
So the seed-applications command + service are removed. Properly linking
manifest entries to the EXISTING catalog is a curated feature, not label-scraping.
Two REAL bugs the review found are fixed and kept:
- Report status (R4): every healthy cycle runs Always/no-detection scripts the
engine counts as "installed", so keying self-heal off installed>0 marked the
common scope selfhealed forever and made 'ok' unreachable. Status now derives
from explicit per-entry self-heal flags only; the stored flag no longer infers
from action=='installed'; the client kit doc reflects it.
- PCTypesStrict (R5): the runtime engine has no strict handling (preinstall
runner only). filters.matches_pctype now applies strict only when phase ==
'preinstall'; simulate + parity thread the scope phase through; the strict test
uses a preinstall scope.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Consolidated fixes from a three-dimension adversarial review.
Data-loss (HIGH): the manifest entry editor stripped fields the form did not
expose, because PUT /entries is a full reset-then-apply. The form now captures
everything - InUseCheck processes as structured name/ExePath/timeout rows (not
just names), LogFile, and the three preinstall flags as checkboxes; the dead
payload-source control (never wired) is removed. New regression test proves an
edit preserves ExePath/timeout/LogFile/PreEnrollment/PCTypesStrict.
Update-entry crash (found by that regression test): replacing an entry's
one-to-one InUseCheck (unique entryid) collided with the old row mid-flush ->
IntegrityError -> 400. update_entry now frees the old InUseCheck (delete+flush)
before populate re-inserts it.
Export truncation (MEDIUM): export_scope_to_share used a plain truncating open,
so a failed/partial write left the live on-share manifest (every PC reads it)
empty. Now writes a temp file in the same dir and os.replace() atomically.
Report dedup case bug (MEDIUM, confirmed by scratch test): the iscurrent demote
matched hostname case-sensitively while the read path uses ilike, so a PC
reporting different casing left two iscurrent rows and double-counted. Demote is
now case-insensitive; regression test added.
Simulator fidelity (MEDIUM): PCTypesStrict was captured but ignored by the
filter mirror, so the simulator wrongly matched a collections-only strict entry
to a nocollections PC via the shared Standard alias group. matches_pctype now
honors PCTypesStrict (disables alias expansion); test added.
Hardening: removed the dead/unscoped GEENFORCE_API_KEY env fallback (never wired
into config; tokens are the only path); create/update entry return 400 on a
duplicate Name instead of 500; parity now asserts scope-level Version/Site; a
new test guards real-manifest field lengths against column limits (the DB-free
parity harness can't see truncation); error handling added to the previously
unguarded editor + reports API calls.
Full suite green; naming + frontend build green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First execution phases of docs/proposals/ge-enforce-plugin.md. The GE-Enforce
manifest becomes shopdb data.
P0 scaffold: new geenforce plugin (api_prefix /api/geenforce, default_enabled
false, core_version >=0.7.0). Registered in PLUGIN_TABLE_OWNERS (ADR-008); its
0001 baseline really creates the tables.
P1a model: one wide manifestentries table + entrytype discriminator (not STI,
not JSON blob), manifestscopes (UNIQUE scopename+phase), the three multi-value
filter child tables, inusechecks + processes, immutable manifestpublishedversions
(frozen rendered JSON), manifestpayloads (inline, capped), pctypealiases
(mirror of the engine lib's alias graph). regvalue stored as its raw JSON
literal so DWord typing survives.
P1c importer + exporter: parse common + gea-shopfloor-* + preinstall.json into
draft rows and rebuild the JSON verbatim from rows in sortorder.
P1d parity harness (GATE A): filters.py mirrors the engine's four filter
functions + alias graph; parity.py proves import+export is behaviorally lossless
(field-identical + same-entries-fire across 18 machine-profile fixtures) WITHOUT
byte-diffing. Verified PASS against all 11 real reference manifests (64 entries)
and a synthetic site-neutral fixture covering every type/filter (the CI gate).
First slice (gea-shopfloor-cmm shape): service layer (import/publish/rollback/
export-to-share), CLI (parity, import-share, publish, export-share), and the
client endpoint GET /api/geenforce/manifest serving the current published
snapshot (never the draft) with ETag/304. Split permissions
geenforce.manage/publish/fetch. Tests prove import->publish->serve, draft edits
never change served bytes, publish+rollback, and auth (401 unauth/wrong-scope).
Contract 0.11.0: added service_token_authorized(scope) to shopdb.api so plugin
service endpoints authorize a scoped managed token without importing core token
internals. Documented in PLUGIN-HOOKS.md.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
