docs(repo): #415 git-weight audit — Phase 1 already closed (#416), Phase 2 jsboige-gated - #621
Merged
Merged
Conversation
…ase 2 jsboige-gated Idle of ai-01 deep-queue supersede (msg-...370u0q): #415 .git weight analysis. Read-only audit + reproducible tool. 0 write Cards/. Stale-issue refresh (4th stale-dispatch this session, after #609/#618/#619): #415 body (measured on 0bf7785, 2026-06-01) states "~1,4 GB tracked at HEAD" + proposes Phase 1 (gitignore + git rm --cached). Phase 1 has SINCE been executed and merged: - PR #416 (94b4371) "stop tracking 968 MB regenerable artifacts" - PR #501 (ff03147) "#415 untrack 10.5 MB 2sxc install module" Audit on master 18b4d02 (tools/git-weight-audit.py, reproducible): - size-pack = 2.05 GiB (matches issue; Phase 1 doesn't reduce pack). - Regenerable zones = 0 at HEAD (untracked), survive only in history: Published/ .NET builds 1.2 GB (42 paths) DNN Downloads zips 206 MB (12) 2sxc/DNN .resources 152 MB (197) -> Phase 1 is CLOSED. The 2 GB is 100% historical. Phase 2 (filter-repo) = the ONLY remaining weight-reducer (~1.5-1.6 GB reclaimable -> <200 MB clone). DESTRUCTIVE, jsboige-gated (rewrites all SHAs, force-push, all machines re-clone). Not autonomous. Sources to PRESERVE (Phase 3, LFS/external, jsboige decision): Cards/Packaging *.ai/pdf/svg 97 MB, Sketch .sketch 45 MB, card PNG 76 MB. Note: Mindmap SVGs (48 MB) still tracked at HEAD though regenerable — not covered by #416; byte-stable deliverables (#565); left conservative. Interim onboarding mitigation (non-destructive, actionable NOW): document `git clone --filter=blob:none` (no partialclone default currently set). This is the one concrete release-safe action recommended. Scope: tools/git-weight-audit.py + docs/repo/415-git-weight-audit.md only. 0 write Cards/, 0 AssetConverter code change. Base 18b4d02. Relates to #415, #416, #501, #134. Co-Authored-By: Claude-Code <noreply@anthropic.com>
clusterManager-Myia
left a comment
clusterManager-Myia
left a comment
Collaborator
There was a problem hiding this comment.
[NanoClaw]
LGTM (audit-tier, lecture complète + tool lu + claims vérifiés contre l'état live). C'est exactement le genre d'audit repo-poids qu'il faut : mesuré, reproductible, honnête sur ce qui est fait vs reste.
Vérifié contre l'état live du dépôt :
- Claim « Phase 1 déjà fermée via #416 » ✓ TRUE. #416 = « stop tracking 968 MB of regenerable build artifacts (#415 Phase 1) », closed 2026-06-01.
- Claim PR #501 second volet Phase 1 ✓ TRUE. #501 merged au commit
ff031470— correspond byte-pour-byte auff031470cité dans l'audit. Le corps du #415 est bien stale sur ce point. - Référence #134 (releases comme home des distributables) ✓ #134 open, cité à bon escient.
Le tool tools/git-weight-audit.py est read-only ✓ — vérifié ligne par ligne : seules opérations git = count-objects -vH, rev-list --objects --all, cat-file --batch-check, ls-tree -r -l HEAD. Zéro rm/add/push/commit/checkout/filter-repo. Le claim « 0 write » est exact. Le script déduplique bien les blobs d'historique par path (garde la plus grosse version), agrège par zone avec flag régénérabilité — reproductible, pas de chiffres boîte noire.
Le raisonnement opérationnel est correct et prudent : Phase 1 (gitignore + git rm --cached) ne réduit pas le pack — ça n'arrête que la ré-addition au HEAD — donc le size-pack 2,05 GiB inchangé est attendu et bien expliqué. Phase 2 (filter-repo) est à juste titre gated sur Emerjesse : réécriture destructive de tous les SHA → backup + force-push coordonné + re-clone de toutes les machines + invalidation des refs de PR ouvertes. La distinction régénérable (Published/, Downloads, .resources = purgables) vs à préserver (sources design Cards/Packaging/*.ai, Sketch = LFS candidates) est nette. Et la mitigation intérimaire non-destructive (git clone --filter=blob:none) débloque l'onboarding indépendamment de la décision Phase 2 — c'est la bonne recommandation actionnable maintenant.
Net : audit sain, claims factuels vérifiés, tool read-only, decision-gating correct sur Phase 2/3. Prêt à merger.
jsboige
added a commit
that referenced
this pull request
Jul 1, 2026
…abrication (#626) * docs(taxonomy): #141 AIF Stage-1 full-scale results — 1232 nodes, 0 fabrication PRIMAIRE of ai-01 deep-queue supersede #3 (msg-...v95b6l): the full-scale AIF Stage-1 candidate sidecar + coverage/WARN report. The generator was shipped in #623; this is the run output (1232/1232 non-card nodes, gpt-5.5 assist, dry-run, 0 write Cards/). Headline — anti-fab closed-set HOLDS at full scale: - 1232 nodes covered (100% of Fallacies non-cards, the #609 gap). - 3850 crossLinks (3.12/node, 100% nodes) + 2310 AIF refs (1.88/node, 96% nodes). - confidence mean 0.73, median 0.72 (1301 high >=0.8 / 2460 mid / 89 low). - FABRICATION = 0. invalid targets = 0. invalid verbs = 0. The closed-set design (targets from enumerated real nodes, AIF tokens from the 60-token observed Walton vocab) contains LLM fabrication completely across 1232 nodes. crossLink verb mix: IsRelatedTo 42% / Leverages 33% / Mirrors 15% / Allows 7% / Inverts 2% / Opposes 1% / PredatesOn 1% / Denounces <1%. Honest improvement vs pilot: Opposes/PredatesOn/Denounces were 0 in the 28-node pilot (flagged limitation) -> non-zero at scale (sample was too small to surface them); still sparse, conservative-but-incomplete. Top Walton DirectRef schemes well-represented: Bias_Inference 262, Sign_Inference 119, Ethotic_Inference 117, EvidenceToHypothesis 94. WARN = 87 nodes (7%), ALL bad_map:* (mappingType outside the observed {broad,close,narrow}Match set: relatedMatch 78, none 5, exactMatch 2, noMatch 2). These are LEGIT SKOS predicates -> a schema-extension decision for the expert gate, NOT fabrications. none/noMatch = hedge candidates to DROP during ratification. Sidecars: 141-aif-candidates-fullscale.csv (1.1 MB, flat, reviewable, committed) + report. The structured JSON (1.4 MB) is regenerable via --finalize from the checkpoint and is NOT committed (repo already 2.05 GiB per #415/#621 — limit additive weight). 1 node (6.3.1.2.3.1.4 Hypocrisie morale) failed once on an API read- timeout, recovered on a 1-node resume (0 content/parse errors). Scope: docs/taxonomy/141-aif-candidates-fullscale.csv + report only. 0 write Cards/, 0 AssetConverter code change. Base 0e39425. Next (gated post-release): Stage 2 dry-run diff, Stage 3 expert/jsboige ratification gate (decide relatedMatch/exactMatch schema extension, drop none/noMatch hedges, prioritize >0.8 candidates), Stage 4 -> #130 OWL / #136 2sxc. Relates to #141, #609, #620, #130, #136, #192. Co-Authored-By: Claude-Code <noreply@anthropic.com> * docs(taxonomy): #141 AIF Stage-2 dry-run diff — 9 conflict / 2 confirm / 5 silent Read-only diff of the Stage-1 sidecar (#626) against the 12 non-card nodes that already carry an AIF value on the taxonomy (0.97% — the other 1220 are entirely net-new, no conflict possible). High-signal prep for the expert gate: - 2 CONFIRM: gpt-5.5 independently re-derived the exact expert token (Ignorance_Inference, VagueVerbalClassification_Inference) — closed-set produces real Walton mappings. - 9 CONFLICT: 1 field-swap/churn node (Pente glissante) + 5 frank token disagreements → gate must adjudicate, not auto-apply. - 5 SILENT: generator stayed silent on a filled field → gate must preserve existing values. Adds 141-aif-stage2-diff.py (read-only, 0 Cards/ write), 16-row diff.csv, and a Stage-2 section to the report. Same scope/honors pre-tag freeze. Co-Authored-By: Claude-Code <noreply@anthropic.com> --------- Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 1, 2026
…tion) (#628) Interprets the Phase-1 audit (#621, git-weight-audit.py). Quantifies the reduction: ~1.55 GB of fully-regenerable build artifacts in history (Published/ .NET builds 1.2 GB + DNN Downloads zips 205 MB + .resources 150 MB) -> repo 2.05 GiB to ~0.45-0.5 GiB (~75%). Includes a costed git filter-repo plan + 6 explicit risks (force-push to master, 8 open PRs break, tag/release integrity, SHA-shift cascades, filter-repo remote-removal foot-gun, mindmap SVG provenance). Recommendation: DEFER to post-v0.9.0-tag (rewrite during the freeze with 8 open PRs + pending tag = max risk, zero release benefit). Go/no-go = jsboige. Flags design sources (Packaging 97 MB, Sketch 45 MB) as preserve/LFS, out of scope. PROPOSAL ONLY — no clone, no force-push, no history mutation. Reproducible via python tools/git-weight-audit.py. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
This was referenced Jul 3, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Idle of ai-01 deep-queue supersede (
msg-…370u0q): the #415.gitweight analysis. Read-only audit + reproducible tool. 0 writeCards/.Finding — Phase 1 of #415 is already CLOSED (stale-issue refresh)
The #415 body (measured
0bf77852, 2026-06-01) states "~1,4 GB de binaires encore trackés au HEAD" and proposes Phase 1 (gitignore +git rm --cached). Phase 1 has since been executed and merged:94b43712— stop tracking 968 MB of regenerable build artifactsff031470— untrack 10.5 MB 2sxc install moduleThe audit confirms it on
18b4d023: regenerable zones are 0 bytes at HEAD (untracked), surviving only in history.Published/.NET buildsDNN Downloads zips2sxc/DNN .resourcessize-pack= 2.05 GiB — unchanged from the issue (Phase 1 doesn't reduce the pack; it stops re-adding at HEAD). The weight is now 100 % historical.What remains
git filter-repo) is the only remaining weight-reducer — ~1.5–1.6 GB reclaimable → projected clone < 200 MB. DESTRUCTIVE, jsboige-gated (rewrites all SHAs → backup + coordinated force-push + all machines re-clone). Not autonomous.*.ai/pdf/svg97 MB, Sketch 45 MB, card PNG 76 MB) → Git LFS or external; jsboige decision. Must NOT be touched by a blindfilter-repo.One concrete release-safe action recommended
Document the partial-clone mitigation for onboarding NOW — it unblocks new machines independent of the Phase-2 decision:
No
partialclone.filterdefault is currently set, so new clones still fetch all 2 GB. Surfacing this command (README onboarding) is non-destructive and directly addresses #415's original pain ("clone is very long").Reproducibility
tools/git-weight-audit.py—python tools/git-weight-audit.py. Measures pack size, top history blobs (deduped), top HEAD files, zone aggregation with regenerability flags + proof-of-preservation.Note
Mindmap SVGs (48 MB) are still tracked at HEAD though regenerable — not covered by #416; they are byte-stable deliverables (#565); left conservative rather than untracked blindly.
Safety / scope
tools/git-weight-audit.py(read-only git plumbing) +docs/repo/415-git-weight-audit.md. 184 insertions.Cards/, 0 AssetConverter code change (pre-tag safe). Base18b4d023.Relates to #415, #416, #501, #134. 4th stale-dispatch refresh this session (after #609/#618/#619).