Pinky UUUU: one window truth for the lineage graph (v1.103.0) #138

Merged
ClaudiuSchuster merged 1 commit from oss-oo/pinky-uuuu-graphwin into main 2026-10-17 05:26:44 +00:00
Owner

Mission U1 — the PPPP-F1 cure, "ein Blick, eine Wahrheit" for the graph too.

The finding (PPPP validator F1 + the RV regressions-audit family): the lineage-graph module carried a pre-PPPP module-local window twin (LG_WIN_MS / lgAnchor / lgWindowKeeps) whose keep-set exempted only running+favorite — a pinned aged-out row stayed in the TABLES and the census but silently dropped from the GRAPH.

The cure

  • windowKeepsCore(s, win, nowMs, view) — the ONE keep-set core, parameterized by window/now/exemptions; the tables' windowKeeps closes over it and the graph's lgWindowKeeps is a thin pass-through (a DIRECT call — no module twin left to drift; the old twin is structurally gone).
  • The graph rides the tables' exemptions verbatim: running / pinned / favorited / docked / open-panel + the two 🧠 toggles (the live view bundle lgWindowView()); the render signature carries a view fingerprint so a flip re-renders the forest between polls.
  • The all-cap decision (a): all = full history in the graph, exactly like the tables. The honest audit finding: the lineage graph NEVER capped all — the "graph falls back to 7 d" language was the USAGE chart's rule (round V era, usageHours()) misread onto the lineage graph; the header/◷ tooltips now scope the 7 d fallback to the usage chart BY NAME. The graph's readability answer is the aggregate fold, never a hidden cap.

Pins

  • tests/test_round_uuuu.py — THE cross-pin: a node harness slices the tables' windowKeeps AND the lineage module from the same dashboard.html and drives both over one fixture matrix (every exemption × window × toggle arm — 160 agreement checks), plus wiring tripwires and the opt-in vault specimen leg.
  • tests/test_lineage_graph.py — the module harness now runs against the sliced authority (WIN_MS + windowAnchor + windowKeepsCore prepended verbatim); a new keep-set group pins every exemption + the full-history all.
  • Vault specimen tests/rigs/uuuu_graphwin_cdp.py — 14 legs live: pinned/favorite aged-out stay in BOTH surfaces at 24h/3d/7d/all, unpinned aged-out drops from BOTH, the 🧠 toggle rides, chip switches flip the graph live, exact graph==tables agreement at every step. make rig-uuuu-graphwin.
  • README: the lineage section, the v1.42.0 window passage and the timeline carry the decision.

Suite: 1752 tests green (baseline 1741/5skip + 11 legs + 1 opt-in skip). Super-Linter green on the final tree (six lanes).

Mission U1 — the PPPP-F1 cure, "ein Blick, eine Wahrheit" for the graph too. **The finding** (PPPP validator F1 + the RV regressions-audit family): the lineage-graph module carried a pre-PPPP module-local window twin (`LG_WIN_MS` / `lgAnchor` / `lgWindowKeeps`) whose keep-set exempted only running+favorite — a pinned aged-out row stayed in the TABLES and the census but silently dropped from the GRAPH. **The cure** - `windowKeepsCore(s, win, nowMs, view)` — the ONE keep-set core, parameterized by window/now/exemptions; the tables' `windowKeeps` closes over it and the graph's `lgWindowKeeps` is a thin pass-through (a DIRECT call — no module twin left to drift; the old twin is structurally gone). - The graph rides the tables' exemptions verbatim: running / pinned / favorited / docked / open-panel + the two 🧠 toggles (the live view bundle `lgWindowView()`); the render signature carries a view fingerprint so a flip re-renders the forest between polls. - **The all-cap decision (a)**: `all` = full history in the graph, exactly like the tables. The honest audit finding: the lineage graph NEVER capped `all` — the "graph falls back to 7 d" language was the USAGE chart's rule (round V era, `usageHours()`) misread onto the lineage graph; the header/◷ tooltips now scope the 7 d fallback to the usage chart BY NAME. The graph's readability answer is the aggregate fold, never a hidden cap. **Pins** - `tests/test_round_uuuu.py` — THE cross-pin: a node harness slices the tables' `windowKeeps` AND the lineage module from the same dashboard.html and drives both over one fixture matrix (every exemption × window × toggle arm — 160 agreement checks), plus wiring tripwires and the opt-in vault specimen leg. - `tests/test_lineage_graph.py` — the module harness now runs against the sliced authority (WIN_MS + windowAnchor + windowKeepsCore prepended verbatim); a new keep-set group pins every exemption + the full-history `all`. - Vault specimen `tests/rigs/uuuu_graphwin_cdp.py` — 14 legs live: pinned/favorite aged-out stay in BOTH surfaces at 24h/3d/7d/all, unpinned aged-out drops from BOTH, the 🧠 toggle rides, chip switches flip the graph live, exact graph==tables agreement at every step. `make rig-uuuu-graphwin`. - README: the lineage section, the v1.42.0 window passage and the timeline carry the decision. Suite: 1752 tests green (baseline 1741/5skip + 11 legs + 1 opt-in skip). Super-Linter green on the final tree (six lanes).
ClaudiuSchuster deleted branch oss-oo/pinky-uuuu-graphwin 2026-10-17 05:26:44 +00:00
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
OSS-Singularity/zcode-dispatch-observatory!138
No description provided.