feat(v3): add bootstrap authority carrier

[briansrls]

Summary

• Add src/v3/std/bootstrap_authority.dag as the generalized bootstrap-authority membership carrier for row 82 prerequisites.
• Model membership as BootstrapAuthoritySet = Map<String, BootstrapAuthority> so paths are duplicate-free map keys, with each value carrying only the root classification (StdAuthority, V3StdAuthority, V3SpecAuthority, CompilerAuthority, or ExtdepsFixtureAuthority) and no duplicate path payload.
• Add a focused full-bootstrap snapshot ratchet proving bootstrap_authority is present and matches the generated full bootstrap source-file membership.

Verification will consume the bootstrap_authority map for diagnostics_empty_after_bootstrap row 82; this PR intentionally does not implement that Verification TestClaim.

Validation

• CARGO_TARGET_DIR=/tmp/gunbc-pr1554-target cargo test -p v3-compiler --test integration pb1_bootstrap_full_snapshot_test::bootstrap_authority_rows_match_full_bootstrap_source_files -- --exact --nocapture
• CARGO_TARGET_DIR=/tmp/gunbc-pr1554-target cargo test -p v3-compiler --test integration pb1_bootstrap_full_snapshot_test::generated_full_bootstrap_snapshots_have_no_diagnostics -- --exact --nocapture
• CARGO_TARGET_DIR=/tmp/gunbc-pr1554-target cargo test -p v3-compiler refresh_handwritten_parse_snapshot_manifest -- --ignored
• cargo fmt --check
• git diff --check
• cargo run -p v3-compiler --features bootstrap-regen-fresh --bin regen_bootstrap -- --verify

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 9d69b87d · Trigger: schedule
• Thinking: 282s wall

---

BLOCKING (3)

Root Cause

• src/v3/std/bootstrap_authority.dag bootstrap authority categories were added as a flat coproduct without a dissolution receipt → add the required classification/ledger or dissolve the categories into structural path-root facts.
• src/v3/std/bootstrap_authority.dag bootstrap membership is modeled as untyped path strings plus a separate category fact → encode the root/category in the path type or variant payload so mismatches are unrepresentable.
• src/v3/std/bootstrap_authority.dag set membership was represented with a list because it is convenient to literalize → use the std Set/Map shape or document a bounded scaffold with a dissolution trigger.

Non-blocking — Strengths

• src/v3/compiler/tests/integration/pb1_bootstrap_full_snapshot_test.rs The added snapshot test is useful drift coverage for the current generated bootstrap rows.

⚠️ The authority model is the right direction, but the new substrate shape needs the coproduct receipt and stronger type-level representation before it lands.

[briansrls]

BLOCKING: BootstrapAuthority makes kind and path independent fields even though the test treats kind-to-prefix as an invariant, so illegal category/path combinations remain representable instead of being type-enforced.

[briansrls]

Blocking on #1554: the carrier shape is the right one, but the new authority is not actually present in the generated full bootstrap snapshot yet.

Verified on 7e00ae2d5:

CARGO_TARGET_DIR=/tmp/gunbc-pr1554-target cargo test -p v3-compiler --test integration pb1_bootstrap_full_snapshot_test::bootstrap_authority_rows_match_full_bootstrap_source_files -- --exact

fails with:

full bootstrap loads bootstrap_authority

Static check matches that failure: src/v3/std/bootstrap_authority.dag is added, but neither src/v3/compiler/src/bootstrap_generated.rs nor src/v3/compiler/src/bootstrap_generated_without_parse_surface.rs contains bootstrap_authority / src/v3/std/bootstrap_authority.dag. As written, Verification cannot consume bootstrap_authority.authorities from generated_full_bootstrap_dag().

Please regenerate/check in the bootstrap snapshots after adding the new std authority, then rerun the focused snapshot test. Also replace the placeholder PR title/body with the actual row-82 carrier scope and validation before ready.

— sent from jolly-ram-908 (inbox #1130); reply at #1130

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 1d7dc44e · Trigger: schedule
• Thinking: 179s wall

---

BLOCKING (1)

Root Cause

• src/v3/std/bootstrap_authority.dag Bootstrap source roots are modeled as labels over unrefined strings → define root-specific FilePath refinements for each authority variant and use those in the authority carrier.

⚠️ One substrate invariant remains convention-checked by the Rust ratchet instead of represented in the model.

[briansrls]

BLOCKING: BootstrapAuthority variants still carry raw String paths, so root/prefix mismatches remain constructible instead of being enforced by FilePath/root-specific refinements, violating P2/API-level enforcement and M9 grounding.

[briansrls]

Review metadata

• Provider / model: cursor / composer-2
• Commit: a45a238a · Trigger: schedule
• Comparison: origin/main @ d2e72d26 ... review/pr-1554-a45a238a @ a45a238a
• Thinking: 46s wall

---

Findings

None that rise to rubric violations. The new coproduct is explicitly classified 🟢 terminal with rationale (bootstrap_authority.dag lines 14–18), duplicate key/path membership is acknowledged with a named dissolution trigger (bootstrap_authority.dag lines 32–37), and the integration test ties the declared map to the committed snapshot’s declaration source paths (pb1_bootstrap_full_snapshot_test.rs lines 103–121), which matches P2 (single authority / facts carried forward) rather than undermining it. verification.dag is unchanged in this diff, so nothing here establishes a second competing membership table alongside this carrier—the file header scopes this as the substrate prerequisite for later Verification consumption (bootstrap_authority.dag lines 4–12).

Verdict

APPROVE — The change is coherent: a substrate bootstrap_authority map plus bootstrap regen and a structural membership ratchet against generated_full_bootstrap_dag(). No concrete clash with INVARIANTS, modeling-discipline, CODING, or TESTING surfaced from the diff.

Exploratory (optional)

• pb1_bootstrap_full_snapshot_test.rs line 120: the failure message refers to `bootstrap_authority.authorities`, but the declaration under test is the top-level bootstrap_authority map body—worth tightening the string so failures match the actual shape (cosmetic only).

[briansrls]

Review metadata

• Provider / model: codex / codex-default
• Commit: 2bb88855 · Trigger: schedule
• Comparison: origin/main @ d2e72d26 ... review/pr-1554-2bb88855 @ 2bb88855
• Thinking: 44s wall

---

Verdict: APPROVE

The diff adds a typed bootstrap authority carrier with an explicit scope, root-kind classification, key/payload consistency ratchet, and a documented dissolution trigger for the temporary duplicated path payload. I don’t see a concrete violation of the pinned invariants, modeling discipline, coding guide, or testing guide in the changed lines.

I did not run builds or tests, per instruction.

[briansrls]

Review metadata

• Provider / model: codex / codex-default
• Commit: 890cf556 · Trigger: schedule
• Comparison: origin/main @ dfc61af7 ... review/pr-1554-890cf556 @ 890cf556
• Thinking: 51s wall

---

Findings:

• src/v3/std/bootstrap_authority.dag:26 — BLOCKING, violates modeling-discipline “illegal states unrepresentable” / P2 API-level enforcement. The new substrate carrier encodes authority root as the BootstrapAuthority variant, but each variant payload is only a wrapper around unconstrained FilePath (lines 19-23). That means StdAuthority { path: { path: "src/v3/spec/go.dag" } } is still representable and only caught by the Rust ratchet at pb1_bootstrap_full_snapshot_test.rs:123. For a substrate authority carrier, the root/path consistency needs to be structural, not convention/test-enforced.

Verdict: REQUEST_CHANGES. The PR is otherwise narrowly scoped and the key/value duplication is documented with a dissolution trigger, but the new authority model still admits the exact drift it says it prevents. I did not run builds or tests, per instruction.