Newer
Older
openstack-caracal-dc-dc / docs / audit / h1-token-map-20260719.md

H1 token migration map -- sweep Batch 2 item 2.5 (2026-07-19)

Per the ratified stage-status vocabulary (Option A + amendment H1, docs/audit/ga-rulings.md): the extractor's status scan (record-audit.py section 6, Batch 0.4) listed EVERY distinct status token; this map records each token's disposition so none survives unmapped. The legal set (OPEN/BLOCKED/HELD/CLOSED) is used ONLY in CURRENT-STATE.md.

Scan input: 12 distinct tokens, 163 flagged counted occurrences (pre-migration run, 2026-07-19). Post-migration verification at the foot.

Token Class Disposition
OPEN legal retained; CURRENT-STATE rows only as a status; elsewhere prose/history
BLOCKED legal retained; G9 migrated TO this from STOPPED (blocker named in the row)
HELD legal retained (vr1-dc1 G12 etc.)
CLOSED legal retained
STOPPED illegal -> BLOCKED with the blocker named (G9 + section-1 prose); commit-message quotes cited by hash, not re-quoted
DEPLOY-READY illegal -> removed (per Option A: readiness is not a state); readiness-doc banner demoted (GA-F02); historical mentions lowercased
READY illegal -> removed as status; no live row carried it (prose mentions lowercased where flagged)
COMPLETE illegal -> CLOSED where a gate/stage row carried it (none did on CURRENT-STATE after Batch-1 edits); prose/dated events lowercased
COMPLETED illegal same as COMPLETE (extractor normalizes)
DONE illegal -> dated lowercase event wording on live surfaces (workflow doc, runbooks, readiness); close-out instructions RETARGETED to write status into CURRENT-STATE.md, not the workflow tracker
TO RUN illegal -> OPEN with in-flight evidence (per H1's explicit mapping); the one live row (G4) already CLOSED in Batch 1
RUNNING illegal as a STAGE status -> OPEN with in-flight evidence; every flagged occurrence was factual VM-run-state prose -> lowercased
VOID / REOPENED / SUPERSEDED event words legitimate as transition history in prose; the one status-position occurrence (workflow component table) lowercased into dated event wording

History-class surfaces deliberately NOT token-edited (their disposition is structural, not lexical): docs/session-ledger.md (GA-R4 rule-6 one-time rotation, Batch 4), docs/changelog-*.md + docs/v1-redeploy-changelog.md (changelog class, GA-R2 consolidation, Batch 4), docs/design-decisions.md (append-only register, KEPT per GA-R3). Verbatim quotes (operator utterances, commit messages) are never reworded; where a quote carried a token, the citation was switched to hash form.

Post-migration verification (extractor re-run, 2026-07-19): zero flagged occurrences on live non-history surfaces (CURRENT-STATE, workflow doc, runbooks, readiness doc); remaining flags sit only on the history-class files listed above, awaiting their Batch-4 rotation. Gate rows carry GA-R6/E1 type tags ([V]/[R]) as of this batch.