Curriculum tranche 002 — 8 units → 9 lesson flips (G1 forecast)

[theislampill]

Curriculum tranche 002

Recomputed from repo ledgers at main@c8b5db5 after owner-approved G1 (2026-08-05) by tools/select_tranche.py --write. This issue is a public mirror of the committed ledgers; on any conflict the ledgers win. Evidence in comments must cite repo paths, PR numbers, or blocker ids — never private-workspace packet contents or server topology.

What this tranche is

This is one unit-closure tranche of the 226-lesson absorption schedule (protocol: docs/FUSHA-TRANCHE-PLAN-2026-08-04.md; selector: tools/select_tranche.py). The current selector forecast closes 8 canonical units, which flips 9 lessons to fully_operationalized.

Units to close: cu-grapheme-inventory-and-confusables, cu-orthographic-connectivity-classes, cu-short-vowel-diacritics-and-vocalization-state, u-s06, u-s01, cu-nunation-support-orthography, u-n01, cu-definite-article-assimilation

Lessons that flip when they close: L1.M1.01, L1.M1.02, L1.M1.03, L1.M1.04, L1.M1.05, L1.M1.06, L1.M1.07, L1.M2.01, L1.M2.10

No earlier-unit closure is assumed in this forecast: G1 reopened fixture-only units until real behavioral grading is proven.

Impact on transclusion & the sarf/naḥw lattice projection

Canonical units are the clean-room knowledge plane that the meta-transclusive sarf & naḥw lattice
projects from: every rich-colour segmentation and rich-hover gloss the tutor or the qamus renderer
emits must trace to certified, unit-grounded analyses — never to surface-match reuse (og-1/og-2).
Closing these units upgrades what the projection lattice can honestly serve for every construction
they govern, and their occurrence-grounding rows (Line 3) bind curriculum knowledge to concrete
quran:S:A:W occurrences — the same occurrence nodes the qamus VN/P issues project onto pages.

Impact on the flywheel & repo

• machine packs green under the runtime consumer (or explicit instructional-only disposition);
• misconception clusters → runtime KC coverage + realized drill keys (dr-* rows);
• error sections → red-first fixtures in the owning skill's eval bank (+ registry rows);
• ledgers regenerate green; FUSHA-BENCH runs at assembly and its axis deltas land in the tranche PR.

Closure conditions

• Tranche manifest committed by tools/select_tranche.py --write (membership finalized)
• Line 1 — machine packs exist and are green for every machine-eligible unit (or explicit instructional-only disposition)
• Line 2 — misconception clusters have runtime KC coverage + realized drill-key rows
• Line 4 — error-section rows landed as red-first fixtures in the owning skill's eval bank
• Behavioral coverage — every required fixture compared against a real consumer verdict; behavioral_coverage.satisfied=true
• Binding rows emitted (unit and lesson); all ledgers regenerate green
• Line 3 — occurrence grounding landed from committed evidence or explicitly parked (evidence-blocked state; never stalls the tranche)
• Every listed lesson fully_operationalized (or explicit blocked/not_applicable with cause) in lesson-burndown.json
• One integration train: full harness green + FUSHA-BENCH run, deltas in the tranche PR description

Update & closure discipline

• Membership above is the deterministic forecast at main@c8b5db5 after PR fix: separate fixture tracing from behavioral closure #220 introduced the separate behavioral-coverage gate. It changed from 4 units / 11 lesson flips to 8 units / 9 lesson flips. Membership refreshes again when tranche 002 actually starts after prior-tranche closure.
• The executor updates checkboxes as production lines land, each with PR + ledger citations.
• This issue closes when curriculum/l1l6/reports/lesson-burndown.json shows every listed lesson
  fully_operationalized (or an explicit blocked/not_applicable disposition with cause — no
  weaker state may borrow a stronger name). The closing comment must state the realized impact:
  lessons flipped, units closed, FUSHA-BENCH axis deltas, fixtures/KCs/packs returned to the skills,
  and any unlocks for other issues (name them).

[theislampill]

Membership refresh: tools/select_tranche.py --write at main@98b242d selected u-s06, u-s01, u-n01, u-s08 and the same 11 lesson flips. Diff from the main@3396d79 forecast: none. Ledger: curriculum/l1l6/reports/lesson-burndown.json.

[theislampill]

Cross-issue measurement from proposed PR #220: G1 reopens five fixture-only tranche-001 units for real behavioral grading. On the PR head, the selector forecast for tranche 002 changes from 4 units / 11 lesson flips to 8 units / 9 lesson flips. Main remains authoritative until merge; membership will be refreshed again after G1/G2 establish the actual prior-tranche closure. Ledger: curriculum/l1l6/reports/lesson-burndown.json.

[theislampill]

Post-merge forecast refresh: python tools/select_tranche.py --write at main@c8b5db5 was byte-identical. The tranche-002 forecast remains 8 units / 9 lesson flips; membership refreshes again only after tranche 001 actually closes, per steer §2.1.