Every ticket as the root of its provenance graph — goal, patch, approval, runs and (as the ticket-class build lands) mandate, manifestations, guarantees, journey, artifacts and children. Rendered through the <<<TicketGraphList>>> tag.
forge_evolution.py docs: my fabricated 'omni covers 11%' propagated into a merged tool (3 places). Measured by OBJECT it is 74% (193/261 tips) — my by-name join was broken. Conclusion unchanged (omni stays black-boxed); docs only.
goal
Correct a FALSE MEASUREMENT that I (the dispatch agent) fabricated and then wrote into
ticket #452's body, from which it propagated into a merged tool's documentation. Docs-only. The
tool's behaviour is CORRECT and must not change.
THE FALSEHOOD, now in checkouts/current/congruency/tools/forge_evolution.py at 3 places:
line 74 : "264 forge bare repos vs 145 repos in omni, of which only 29 overlap — omni covers 11%"
line 240 : "...it covers only 11%..."
line 2317: "...it covers 11% of the forge..."
WHY IT IS WRONG: I compared omni's repos to forge's BY NAME. The naming conventions do not align —
omni's refs are `refs/gitlab/<slug>` while forge's bare repos are `gitlab__<slug>.git` — so the join
matched almost nothing and I reported 11%. That number is an artifact of my broken comparison.
THE MEASURED TRUTH (name-independent, by OBJECT — re-runnable): for each forge bare repo take its
resolved tip and ask `git -C omni.git cat-file -e <tip>`:
forge repos with a resolvable tip : 261
tips omni DOES contain : 193 (74%)
tips omni does NOT contain : 68 (26%)
repos with no refs : 3
omni covers 74% of forge tips, NOT 11%. (Separately: the "~86%" the ticketer cites is BLOB coverage
in `occ`, not repo coverage — three different numbers, do not conflate them.)
THE CONCLUSION IS UNCHANGED — only my reasoning for it was wrong. omni STAYS BLACK-BOXED, on the two
legs that are true:
* it is NOT a git — every ordinary command fails (rev-parse HEAD prints the literal string "HEAD";
ls-tree/log are fatal). It is a DEDUPLICATION DEVICE and it was a TEST, not an archive.
* it is STALE and incomplete — missing agent_tools_suite's tip 888d3447c6 and 67 others.
So `_guard_store()` and every refusal STAY EXACTLY AS THEY ARE. This ticket does not relax the guard;
it removes a false justification for it.
DELIVERABLE: checkouts/current/congruency/tools/forge_evolution.py — DOCS ONLY.
1. Replace the 11%/145/29 claims with the measured 74% (193 of 261 tips by object), and say the
coverage was measured BY OBJECT because a by-name join does not work (the ref-shape mismatch
above). Anyone re-deriving it must not repeat my mistake.
2. Keep every black-box statement and every guard. Restate the two TRUE reasons (not a git; stale).
3. Change NO behaviour: no logic edits, no new/removed guards, no signature changes.
ACCEPTANCE: `grep -c '11%'` on the file is 0; the 74%/193/261 figures appear with the by-object
method named; `--selftest` still 119/119; `_guard_store()` still refuses omni.git and archives/;
a live run still prints "omni.git is BLACK-BOXED and never opened"; git diff shows COMMENTS ONLY.
RUN ON CLAUDE.
patch
none
approval
unapproved
runs
no runs recorded
mandate (clauses)
not yet recorded — lands with the ticket-class build
manifestations
not yet recorded — lands with the ticket-class build
guarantees
not yet recorded — lands with the ticket-class build
journey (blunders & successes)
not yet recorded — lands with the ticket-class build