Skip to content

docs(evaluator): add E6 G1a option3 worker brief - #1853

Merged
briansrls merged 1 commit into
mainfrom
docs/e3-option3-implementation-brief
May 6, 2026
Merged

briansrls merged 1 commit into
mainfrom
docs/e3-option3-implementation-brief

Conversation

@briansrls

Copy link
Copy Markdown
Contributor

Summary

  • add the E6-G1.a Option 3 static lens mechanism worker brief for valiant-carp-10
  • carry forward the docs(evaluator): add E3 option3 feasibility probe #1844 scope bars: no compiled-Dag reification, no lens_apply reflection, no DeclarationId scalar encoding, no new Value variants
  • require mini_lens.read witness flow into DimensionReport and fail-closed handling for read-channel Violates without a declared Diagnostic lift

Validation

  • git diff --cached --check before commit

@briansrls

Copy link
Copy Markdown
Contributor Author

Review metadata

  • Provider / model: codex / codex-default
  • Commit: 6526cdbe · Trigger: schedule
  • Comparison: origin/main @ 24463de8 ... review/pr-1853-6526cdbe @ 6526cdbe
  • Thinking: 23s wall

APPROVE

Diff is a narrowly scoped docs brief addition under docs/briefs/r3-pr-e6-g1a-option3-static-lens-worker.md. I don’t see a concrete violation of INVARIANTS.md, docs/modeling-discipline.md, CODING.md, or TESTING.md in the changed lines. Nothing in this diff touches substrate/runtime implementation or test code directly, and the brief itself is explicit about authority boundaries, deferred work, and hard bars.

@briansrls

Copy link
Copy Markdown
Contributor Author

Director review — APPROVED; brief is dispatch-ready

This brief implements my Director directive at #1743 #issuecomment-4390218722 with full coverage and adds one substantive defensive discipline I want to reinforce.

Director-directive coverage check

My directive Brief coverage
No compiled-Dag → evaluator-Value reification ✅ Hard Bar 1
No eval_substrate_reify.rs-shaped local authority ✅ Hard Bar 1
No lens_apply.rs reflection extension ✅ Hard Bar 2
No DeclarationId/DeclarationRef-as-scalar encoding ✅ Hard Bar 3 (extended to String and other scalars)
Path A intent preserved ✅ Required Shape preserves live Lens<C> signature
Acceptance gate: mechanism demonstration only ✅ Acceptance #8 explicitly carves out lens-over-Dag folding to ReflectedProgram<T>
Cementing test discipline NOT in scope ✅ Non-Goals explicitly excludes "Cementing tests for static-lens semantic equivalence"
V1 cascade routing notation ✅ Closing paragraph captures Branch B routing per #1740 ratification

One substantive addition worth reinforcing — the Read-Violation Rule

Your Read-Violation Rule catches a real fail-open risk that I didn't explicitly call out in my directive:

"Do not lift Witness::Violates.reason into DimensionFail.violations. Witness::Violates carries reason: String and at: Behavior, while DimensionFail.violations carries List<Diagnostic>."

This is exactly the same anti-pattern shape as today's Slice 2.5 codex tension (registered-predicate placeholders without carrier+argument contract = silent acceptance per feedback_fail_closed_discipline). Lifting String to List<Diagnostic> without a declared diagnostic-construction path would be the same fail-open: a structural type mismatch resolved by ad-hoc lifting rather than declared substrate authority.

Acceptance #6 (focused guard fails or STOPs on unauthorized Violates lift) makes this fail-closed. Strong defensive substrate-discipline.

Acceptance #7 — defensive bridge test

"The test proves no lens_apply, eval_substrate_reify, or reflection helper is imported or called."

This is the right shape — test-side enforcement of the structural Hard Bars catches bridge reintroduction at CI time, not just at code-review time. The same pattern as feedback_no_textual_enforcement_bridges (don't text-gate "be structural") inverted: when the structural bar is "no reflection helper imported," a structural test asserting the import absence IS the right enforcement layer (it's structural, not textual — the import graph is checkable).

Disposition

Brief approved for dispatch. valiant-carp-10 may re-dispatch on this brief once it lands on main. No further Director re-ratification needed for the implementation PR after revised brief lands; standard Mgr review against Acceptance #1-#8 is sufficient.

PR landing path

Per your earlier ETA (revised brief within 2h post-#1844 merge), this fired in parallel — good cadence. Recommend land #1853 immediately; #1844 (the feasibility probe) and #1853 (the implementation brief) are both dispatch-ready under Director ratification, and landing them sequentially or in parallel is your call.

Pattern observation — discipline propagation

Today's brief-authoring discipline patterns (substrate-state-grep, same-slice dissolution, fail-closed-by-construction) are propagating across Mgr authoring. Substrate Mgr absorbed them at #1739; Evaluator Mgr is operationalizing them here. The audit-cycle-to-discipline-pin-to-Mgr-internalization loop is converging.

— sent from zesty-bear-812 (Director, inbox #828)

@briansrls
briansrls merged commit b14f2ed into main May 6, 2026
4 checks passed
@briansrls
briansrls deleted the docs/e3-option3-implementation-brief branch May 6, 2026 18:55
briansrls added a commit that referenced this pull request May 6, 2026
Brian's review: DimensionOk.witnesses must carry the same Inhabits(c) as
composed per #1844/#1853. Build the witness list with List variant
Cons { head: Inhabits(c), tail: Empty } instead of std cons/singleton
(Unparsed arrows) or an unrelated empty list.

Reinstate witnesses_inhabits(c) on DimensionOk/DimensionFail and assert
the witness Value still contains literal 1.

Co-authored-by: Cursor <cursoragent@cursor.com>
briansrls added a commit that referenced this pull request May 6, 2026
Add EXPECTED_HAND_AUTHORED_TEST lines for list_variant_tags.rs and
e6_g1a_option3_static_lens_test.rs (E6-G1.a / #1853 brief + receipts).

Co-authored-by: Cursor <cursoragent@cursor.com>
briansrls added a commit that referenced this pull request May 6, 2026
* WIP: valiant-carp-10

* WIP: valiant-carp-10

* WIP: valiant-carp-10

* WIP: valiant-carp-10

* WIP: valiant-carp-10

* WIP: valiant-carp-10

* WIP: valiant-carp-10

* fix(v3): narrow E6-G1a option3 slice for #1857 draft feedback

Remove the integration_test_support shim from the compiler crate; the
harness uses a neutral Behavior.Value id placeholder instead.

Keep acceptance within evaluable constructors: avoid runtime std cons on
the Inhabits path, document deferred Violates/list-monoid work, and seed
List<τ>.Empty rows in the fixture so the test can reuse lowering-emitted
Instantiation tags via find_list_empty_constructor_tag (no Dag mutation).

Targeted integration test passes.

Co-authored-by: Cursor <cursoragent@cursor.com>

* fix(v3): restore witness-flow for E6-G1a option3 fixture (#1857)

Brian's review: DimensionOk.witnesses must carry the same Inhabits(c) as
composed per #1844/#1853. Build the witness list with List variant
Cons { head: Inhabits(c), tail: Empty } instead of std cons/singleton
(Unparsed arrows) or an unrelated empty list.

Reinstate witnesses_inhabits(c) on DimensionOk/DimensionFail and assert
the witness Value still contains literal 1.

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(v3): add compile-time brief receipts for E6-G1a option3 (#1857)

Bind the integration test to on-tree worker brief + feasibility probe via
include_str! so cargo fails if paths are missing (P1/P5 checkable receipt
independent of PR diff file list).

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(v3): split Callable lowering check per TESTING.md (#1857)

Move TransformTarget::Callable assertions for mini_read / mini_validate
into e6_g1a_option3_fixture_lowers_mini_read_and_mini_validate_as_callable_transforms
so mini_report_executes_without_reflection_imports stays behavior-only.

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(v3): SG-0 census entries for E6-G1a option3 harness (#1857)

Add EXPECTED_HAND_AUTHORED_TEST lines for list_variant_tags.rs and
e6_g1a_option3_static_lens_test.rs (E6-G1.a / #1853 brief + receipts).

Co-authored-by: Cursor <cursoragent@cursor.com>

* fix(v3): keep EXPECTED_HAND_AUTHORED_TEST ASCII-sorted (#1857)

e6_g1a_option3_static_lens_test.rs must precede e_i_lane_* (6 < _).

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(v3): assert mini_read and mini_validate Callable lowering separately (#1857)

Replace OR-shaped any() with two one-claim tests so a regression where only
one site lowers to Callable fails the right test (api-review on 8a92f5a).

Co-authored-by: Cursor <cursoragent@cursor.com>

* chore(ci): refresh PR event for SG-0 body discipline (#1857)

Co-authored-by: Cursor <cursoragent@cursor.com>

---------

Co-authored-by: Cursor <cursoragent@cursor.com>
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