docs(handoff): tier-0 orchestration position — S1 closed, coherence made honest, S2 parked green at the M1 wall #189

Manually merged
mdoorn merged 4 commits from feat/189-docs-handoff-tier-0-orchestration-position-s1 into master 2026-08-06 10:04:43 +02:00
Owner

docs(handoff): tier-0 orchestration position — S1 closed, coherence made honest, S2 parked green at the M1 wall

Along the way:

  • docs(ruling): #49 — M1 refused as malformed; the real answer is temporaries placement, and it leads the queue
  • docs(ruling): #50 — the raise cell is a declared record in its unit's plex, ruled (a) with the per-unit home made binding
  • docs(compress): CLAUDE.md to strict law, questions.md to its rulings — owner directive 2026-08-15

Touches 3 files: CLAUDE.md, handoff.md, questions.md.

docs(handoff): tier-0 orchestration position — S1 closed, coherence made honest, S2 parked green at the M1 wall Along the way: - docs(ruling): #49 — M1 refused as malformed; the real answer is temporaries placement, and it leads the queue - docs(ruling): #50 — the raise cell is a declared record in its unit's plex, ruled (a) with the per-unit home made binding - docs(compress): CLAUDE.md to strict law, questions.md to its rulings — owner directive 2026-08-15 Touches 3 files: `CLAUDE.md`, `handoff.md`, `questions.md`.
Author
Owner

Split. The decision in boundary is a pure function now, with its own tests in handoff.md.

Split. The decision in `boundary` is a pure function now, with its own tests in `handoff.md`.
Author
Owner

Comment added on exactly in questions.md. It was obvious when I wrote it and it will not be in a month.

Comment added on `exactly` in `questions.md`. It was obvious when I wrote it and it will not be in a month.
mdoorn approved these changes 2026-08-06 08:30:00 +02:00
mdoorn left a comment

Good. Small enough to read in one sitting, and handoff.md is easier to follow than it was, at #189.

Good. Small enough to read in one sitting, and `handoff.md` is easier to follow than it was, at #189.
@ -44,0 +50,4 @@
impl.** S2's new check caught ml_othello laundering `List<Disc>` through
`impl Concat<List<Int,…>>`: coherence keying wildcarded EVERY type
argument, not just the impl's own generics — `++` silently retyped Disc
elements to their i64 repr. The fn boundary always refused this; the
Owner

boundary reads, decides and writes in one function. The deciding part is the piece I would want to test on its own. Worth splitting?

`boundary` reads, decides and writes in one function. The deciding part is the piece I would want to test on its own. Worth splitting?
@ -297,0 +144,4 @@
delegation, 2026-08-15): the code STANDS as the fail-CLOSED guard — an
un-seated forwarder's placeholder is a zero address, so deletion trades
a refusal for a wild read; it retires when placement covers everything,
exactly as its named trajectory says — never a zombie code.
Owner

I had to read exactly twice. Nothing wrong with it, but one line of comment would save the next person the second read.

I had to read `exactly` twice. Nothing wrong with it, but one line of comment would save the next person the second read.
The night closes with two commits landed and verified on slice-t0-soundness
(unmerged; one test-rotate + merge at phase close), one owner question open
(M1: condemned site 392→393 for ml_othello's honest Concat), S2 finished and
parked, and the S8 IndexOutOfRange ruling recorded for slice I. S5 excluded —
the stack agent owns it.
The owner's standing law (written into CLAUDE.md the same day): no
shortcuts, no workarounds; memory reserved at start, runtime never
allocs, provable; less Rust, more CkLang. Applied to M1: the +1, the Int
rewrite, and the park were all workarounds around one defect — the
compiler does not yet compute the result slot of temporaries. questions.md
carries the full ruling; handoff.md reorders the queue with placement at
the front. S2 landed on slice-t0-soundness as dc5c4003 by extraction;
slice-s2-parked deleted; slice-s2-commit2-parked kept as the
placement-era reference.
Asked by the stack-phase orchestrator, amended against the MANDATORY RULE,
ruled by the watcher under the owner's standing delegation. The record law
arrives for the raise boundary: declared in .ck, placed by the one Σ-offset
law, read by ordinary field law; ck_lower::raise_cell::SYMBOL and the
handed-in cell parameters are deleted (Rust shrinks). Home is the unit's
own record, never one global program field — per-seat later per #46(1).
CLAUDE.md: the MANDATORY RULE leads; the owner's same-day additions absorbed
(memory fixed from startup · Arena is a boundary, never bump/slice/merge ·
the pre-reserved runtime stack with List[i] = X knowing its position ·
concat reuses or takes a reservation, all at bootstrap · Bootstrap.ck /
TestBootstrap.ck boot with no Rust · agents never circumvent). Section
numbers 0-7 preserved so every §-citation elsewhere stays valid; the tells,
probe recipe, capability law, seat-model size formula, and pipeline survive
verbatim in substance.

questions.md: second compression under the 2026-08-14 protocol — every
ruling number keeps its binding essence; deliberations live at this commit's
parent. Recorded in the same pass: the standing delegation in the protocol
header; the #47-tail ruling (E0508 stands as the fail-closed guard, retires
with placement); #12 restated with its plan_sizes §4 justification ("for
now" struck); #17's "allocated" corrected to RESERVED; #46(2)'s
app-modules-only trail guard flagged as debt against the MANDATORY RULE;
#46(6)'s open action named (scheduling own-the-frame).
mdoorn manually merged commit 31684ac789 into master 2026-08-06 10:04:43 +02:00
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
2 participants
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
northgale/cklang-bootstrap!189
No description provided.