Skip to content

. - #124

Merged
briansrls merged 9 commits into
mainfrom
organization
Mar 8, 2026
Merged

.#124
briansrls merged 9 commits into
mainfrom
organization

Conversation

@briansrls

Copy link
Copy Markdown
Contributor

No description provided.

@briansrls
briansrls merged commit 364e09c into main Mar 8, 2026
1 check failed
briansrls added a commit that referenced this pull request May 13, 2026
… patterns + Part A (i) split into 3 separate greps to avoid alternation (openai-pro REQUEST_CHANGES on PR #2955)

Addresses openai-pro REQUEST_CHANGES (verdict on sha aeba066 at 2026-05-13T15:55:16Z) — finding valid:

The previous grep patterns contained `|` characters that broke the markdown table cell rendering. Two distinct pipe-character uses needed disambiguation:

1. **Regex literal-pipe escape** `\|` (matches the actual `|` character in the .dag variant-arm syntax): displayed via `\|` → renders as `\|` (backslash + pipe) without splitting the table cell. When user copies the rendered text into shell, `|` becomes `|`, yielding the actual regex `\|`.

2. **Alternation operator** `|` (was in Part A (i) `(PolyLogCost|ExponentialCost|FactorialCost)`): eliminated entirely by splitting (i) into three separate greps (i.a / i.b / i.c). No alternation operators remain in any Part A pattern.

Final pattern shape (rendered):
```
git grep -nE '^[[:space:]]*\| PolyLogCost' src/v3/std/algebra.dag
git grep -nE '^[[:space:]]*\| ExponentialCost' src/v3/std/algebra.dag
git grep -nE '^[[:space:]]*\| FactorialCost' src/v3/std/algebra.dag
git grep -nE '^[[:space:]]*\| PolynomialCost \{ var: SizeVariable, degree: Rational where nonzero \}' src/v3/std/algebra.dag
git grep -nE '^[[:space:]]*\| LinearCost' src/v3/std/algebra.dag
```

Source uses `\|` for each pipe; markdown table cell now renders without column-structure damage.

Receipt-on-failure remains: any of (i.a), (i.b), (i.c) returning 0 lines fails closed; (ii) returning 0 lines fails closed; (iii) returning more than 0 lines fails closed.

Authority:
- openai-pro REQUEST_CHANGES on PR #2955 sha aeba066 at 2026-05-13T15:55:16Z
- GitHub markdown table escape conventions
- INVARIANTS P5 (checkable receipt) — predicate must reliably check carrier shape AND render in authoritative doc

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls added a commit that referenced this pull request May 13, 2026
 merge) (#2955)

* docs(r3): §1.8 row #105 status DECLARED → CANVAS_RATIFIED (post-PR-#2828 merge)

PR #2828 (gate #105 SymbolicCost Tier 1 carrier-extension canvas — Substrate Mgr) merged at 2026-05-13T12:06:20Z (squash `fe1dd99c`). Row #105 status flips DECLARED → CANVAS_RATIFIED per the row's own Status progression line ("DECLARED → CANVAS_RATIFIED requires Director ratification of Substrate Mgr canvas covering Q1-Q5").

Extended ratification scope captured in the row:
- Q1-α (Director msg_676ad4e7): use `Field.compare` for Rational dominance lattice — no parallel Int-tuple representation
- Q2-Y (Director msg_7d51b699): `LinearCost ≡ PolyCost(degree=1)` Practice-4 collapse (same-variant); LinearCost removed as distinct variant
- Q6 Option B (Director msg_b80bcaa8): `PolynomialCost { degree: Rational where nonzero }` carrier refinement (Practice-2; cross-variant redundancy with ConstantCost dissolved via type-level state-space tightening; new Practice-2 vs Practice-4 disambiguation rule: same-variant→P4 collapse, cross-variant→P2 carrier refinement)
- Q7: SymbolicCost preserves full expression (Big-O = derived projection)
- 12 anti-patterns enumerated (canvas §10 + worker brief §11)
- Sign-orthogonal degree admission unblocks operator inverse-exponent request 2026-05-13 ("cubed -> quarter -> quintet roots ... also inverse applies"; n^-x exponential expressible via PolyCost(_, Rational(-x)))

Status progression remains: CANVAS_RATIFIED → CONSUMER_LANDED (Phase 1 worker brief carrier implementation) → PASSING (Part A predicate + Part B algebra tests both green).

Authority chain:
- PR #2828 merged 2026-05-13T12:06:20Z (squash `fe1dd99c`)
- Director msg_ad5e934d (Path A + Tier 1 IN-R3 ratification)
- Director msg_676ad4e7 (Q1-α)
- Director msg_7d51b699 (Q2-Y collapse + 7 anti-patterns added)
- Director msg_b80bcaa8 (Q6 Option B carrier refinement + Practice-2 vs Practice-4 rule)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* docs(r3): §1.8 row #105 — reconcile carrier-landed predicate + CONSUMER_LANDED criteria with ratified Q2-Y collapse + Q6 Option B nonzero refinement (openai-pro REQUEST_CHANGES on PR #2955)

Addresses openai-pro REQUEST_CHANGES (verdict on sha 5a6c678 at 2026-05-13T13:43:38Z) — finding valid:

The row previously had:
- "Path A — rational-degree polynomial promotion + 3 new named variants; net 7→11 variants"
- Part A: grep for "4 new variants" + "PolynomialCost { degree: Rational }" shape

These accept the OLD carrier shape and don't require Q2-Y LinearCost dissolution or Q6 Option B nonzero refinement. A faithful worker could pass the written predicate while:
- Keeping LinearCost as separate variant (violating Q2-Y Practice-4 collapse)
- Allowing `PolyCost(d=0)` (violating Q6 Option B cross-variant Practice-2 carrier refinement; collides with ConstantCost)

Fix updates the row in 3 places (all docs/r3-program-plan.md:333):

1. **Tier 1 ratified carrier extension** — explicitly cites Q2-Y LinearCost dissolution (1b) + Q6 Option B `degree: Rational where nonzero` refinement (1); removes stale "net 7→11 variants" pre-computation (deferred to worker brief execution to avoid drift); cites signed-degree admission for inverse exponents.

2. **Part A predicate** — grep now requires:
   - 3 new variants present (not 4)
   - `PolynomialCost { var: SizeVariable, degree: Rational where nonzero }` shape (refinement-explicit)
   - `LinearCost` absent from file (dissolution receipt)

3. **Part B predicate** — adds receipts for:
   - Dominance lattice via `Field.compare` (Q1-α; no parallel Int-tuple)
   - LinearCost → PolyCost(d=1) normalization
   - `PolyCost(_, Rational(0))` type-rejection negative test (Q6 Option B bootstrap ratchet)

4. **Status progression CONSUMER_LANDED criteria** — adds LinearCost dissolution + nonzero refinement + dominance lattice as required worker brief deliverables.

Authority:
- openai-pro REQUEST_CHANGES on PR #2955 sha 5a6c678 at 2026-05-13T13:43:38Z
- Director msg_7d51b699 (Q2-Y collapse)
- Director msg_b80bcaa8 (Q6 Option B nonzero refinement)
- Director msg_676ad4e7 (Q1-α Field.compare)
- INVARIANTS P2 (Boundary Discipline — single authority) + Practice 2 (illegal states unrepresentable)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* docs(r3): §1.8 row #105 — Part A predicate uses actual .dag variant-arm syntax (codex BLOCKING on PR #2955)

Addresses codex BLOCKING (verdict on sha 64f2ad1 at 2026-05-13T14:42:59Z) — finding valid:

Previous Part A grep `^type (PolynomialCost|PolyLogCost|...)` targets top-level `type` declarations, but `SymbolicCost` variants are `| Name(...)` / `| Name { ... }` arms inside the `type SymbolicCost inhabits Semiring<SymbolicCost>` body at `src/v3/std/algebra.dag:190`. Old grep would never match real variant arms; predicate was design-intent text, not actually verifiable against substrate.

New Part A predicate is three explicit grep checks targeting variant-arm syntax `^\s*\| Name`:

(i) 3 new variants present: `git grep -nE '^\s*\| (PolyLogCost|ExponentialCost|FactorialCost)' src/v3/std/algebra.dag`

(ii) PolynomialCost shape refinement-explicit (exactly one matching line): `git grep -nE '^\s*\| PolynomialCost \{ var: SizeVariable, degree: Rational where nonzero \}' src/v3/std/algebra.dag`

(iii) LinearCost dissolved (empty result): `git grep -nE '^\s*\| LinearCost' src/v3/std/algebra.dag`

Receipt-on-failure: if (ii) returns 0 lines or (iii) returns ≥1 line, Part A fails closed.

Sanity-checked against current algebra.dag (pre-implementation state): greps return expected pre-Q6/pre-Q2-Y matches. Post-Phase-1-worker-brief implementation, the predicates flip per ratified shape.

Also adds explicit framing reference to `src/v3/std/algebra.dag:190` SymbolicCost coproduct body location.

Authority:
- codex BLOCKING on PR #2955 sha 64f2ad1 at 2026-05-13T14:42:59Z
- Director msg_b80bcaa8 (Q6 Option B `where nonzero` refinement)
- Director msg_7d51b699 (Q2-Y LinearCost ≡ PolyCost(d=1) Practice-4 collapse)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* docs(r3): §1.8 row #105 — Part A grep predicates use POSIX-portable [[:space:]] instead of \s (codex BLOCKING on PR #2955)

Addresses codex BLOCKING (verdict on sha bf58a85 at 2026-05-13T15:33:41Z) — finding valid for portability discipline:

POSIX ERE does not define `\s`; GNU grep supports it as an extension, but git's bundled ERE engine across platforms may not. To guarantee Part A predicates fire reliably across the engines git ships with, the three grep checks now use POSIX-standard `[[:space:]]` character class:

```bash
# (i) 3 new variants present
git grep -nE '^[[:space:]]*\| (PolyLogCost|ExponentialCost|FactorialCost)' src/v3/std/algebra.dag

# (ii) PolynomialCost refinement-explicit (exactly one matching line)
git grep -nE '^[[:space:]]*\| PolynomialCost \{ var: SizeVariable, degree: Rational where nonzero \}' src/v3/std/algebra.dag

# (iii) LinearCost dissolved (empty result)
git grep -nE '^[[:space:]]*\| LinearCost' src/v3/std/algebra.dag
```

Sanity-checked: locally git grep accepts both `\s` and `[[:space:]]` (GNU extension), but `[[:space:]]` is the portable form per INVARIANTS P5 (checkable receipt) + P2 (boundary discipline — predicate must reliably check the carrier shape).

Also adds explicit framing note about git's ERE engine portability rationale.

Authority:
- codex BLOCKING on PR #2955 sha bf58a85 at 2026-05-13T15:33:41Z
- POSIX.1-2017 §9.3.5 (character classes)
- INVARIANTS P5 / P2

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* docs(r3): §1.8 row #105 — HTML entity &#124; for literal-pipe in grep patterns + Part A (i) split into 3 separate greps to avoid alternation (openai-pro REQUEST_CHANGES on PR #2955)

Addresses openai-pro REQUEST_CHANGES (verdict on sha aeba066 at 2026-05-13T15:55:16Z) — finding valid:

The previous grep patterns contained `|` characters that broke the markdown table cell rendering. Two distinct pipe-character uses needed disambiguation:

1. **Regex literal-pipe escape** `\|` (matches the actual `|` character in the .dag variant-arm syntax): displayed via `\&#124;` → renders as `\|` (backslash + pipe) without splitting the table cell. When user copies the rendered text into shell, `&#124;` becomes `|`, yielding the actual regex `\|`.

2. **Alternation operator** `|` (was in Part A (i) `(PolyLogCost|ExponentialCost|FactorialCost)`): eliminated entirely by splitting (i) into three separate greps (i.a / i.b / i.c). No alternation operators remain in any Part A pattern.

Final pattern shape (rendered):
```
git grep -nE '^[[:space:]]*\| PolyLogCost' src/v3/std/algebra.dag
git grep -nE '^[[:space:]]*\| ExponentialCost' src/v3/std/algebra.dag
git grep -nE '^[[:space:]]*\| FactorialCost' src/v3/std/algebra.dag
git grep -nE '^[[:space:]]*\| PolynomialCost \{ var: SizeVariable, degree: Rational where nonzero \}' src/v3/std/algebra.dag
git grep -nE '^[[:space:]]*\| LinearCost' src/v3/std/algebra.dag
```

Source uses `\&#124;` for each pipe; markdown table cell now renders without column-structure damage.

Receipt-on-failure remains: any of (i.a), (i.b), (i.c) returning 0 lines fails closed; (ii) returning 0 lines fails closed; (iii) returning more than 0 lines fails closed.

Authority:
- openai-pro REQUEST_CHANGES on PR #2955 sha aeba066 at 2026-05-13T15:55:16Z
- GitHub markdown table escape conventions
- INVARIANTS P5 (checkable receipt) — predicate must reliably check carrier shape AND render in authoritative doc

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@briansrls
briansrls deleted the organization branch June 1, 2026 18:41
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant