Skip to content

PR B0: production stage-origin carrier design note (design only) - #7798

Merged
briansrls merged 7 commits into
mainfrom
session/wise-ram-22-prb0
Aug 4, 2026
Merged

briansrls merged 7 commits into
mainfrom
session/wise-ram-22-prb0

Conversation

@gunbai-bot

@gunbai-bot gunbai-bot Bot commented Aug 4, 2026

Copy link
Copy Markdown
Contributor

Design only. No implementation follows from this note. Per operator ruling 2026-08-04:
write the carrier law now; wait for a closed constructor/consumer denominator before
claiming or implementing class-wide closure.

Two files: the note and its HandAuthoredDocBind row. #7778 is untouched.

The class

FixtureOriginCanMintProductionQualification. Measured on main by execution:

Input to mint_provenanced_rust_artifact Result
direct_rust_door_inferred_tree (hand-authored) Accepted
infer(assemble(direct_rust_door_ingest())) (real pipeline) Rejected

The fixture did not bypass a check. It used a supported API whose signature had already
discarded the epistemic distinction — mint_provenanced_rust_artifact accepts a bare
InferredTree, and nothing in that type says where the tree came from.

What the note locks

  1. Production qualification consumes origin-bearing carriers, never a raw InferredTree,
    raw emitted source, a module-path string, or a fixture receipt.
  2. Subject continuity is part of the type — each rung consumes the previous rung, so a
    fixture cannot skip into the chain.
  3. Origin is derived, never a caller-authored field; the constructor is owned by the
    stage that performed the operation.
  4. UnknownOrigin refuses. It never disappears — that fail-open shape is what P0: fixture-origin production qualification origin probe #7786's
    early drafts carried.
  5. Local construction prevention is not global closure; the certificate separates
    them.
  6. The law is written over an abstract derived population, not today's three known sites,
    so it stays stable whether the census finds three or three hundred.

Claimed rung, deliberately narrow

Mechanically preventable for resolved record-literal construction sites inside the
currently recognized file-identity scope.

The sole_constructor audit (executed) found one call site guarding record literals only,
casts unguarded, a file-identity scope test that contradicts DESIGN section 3 and breaks
under module-identity work, and a fail-open lookup miss. The note carries a ceiling table
and states that the claimed rung is its minimum row. The mechanism is not a wall and
this note does not call it one.

Gates

B0  design only                        <- this PR
B1  covered-form construction admission     after B0 sign
B2  call-occurrence population closure      consumes #7786 successor
B3  production API migration
B4  bypass closure (casts, module identity, fail-open lookup)

Also carries a 10-item mutation contract required before class closure may be claimed.

🤖 Generated with Claude Code

Design only, no implementation. Carrier algebra, subject continuity, origin
resolution states, the executed sole_constructor audit, the exact current
ceiling, closure certificate, mutation contract, and B1-B4 gates.

Claimed rung is the MINIMUM of the ceiling table: mechanically preventable
for resolved record-literal construction sites inside the currently
recognized file-identity scope.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@gunbai-bot
gunbai-bot Bot marked this pull request as ready for review August 4, 2026 18:20
Brian Searls and others added 6 commits August 4, 2026 19:06
doc_graph_has_no_dangling_links and doc_graph_is_clean red because the note
linked witness-evidence-lifecycle-design.md, which is on an unmerged branch.
Naming it by registered slug is also the DESIGN section 3 form - a path link
is a positional reference to something the bind registry already names.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The note asserted witness-evidence-lifecycle-design is registered on
gunbc.doc_graph_roots. Verified false on this PR head and on main: the slug
exists only on the unmerged #7778 branch. Now stated as a planned sibling
with its actual location named, and the condition under which it becomes a
registered citation.

This is the DESIGN section 3 stale-citation class in the change that invokes
section 3 to justify itself.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…erge-order-independent

P0: ExecutionMeasured was listed as a fourth ORIGIN variant. A fixture-derived
artifact can also be execution-measured - that IS the dangerous direct-door
state - so a sibling variant forces the model to discard one fact and reopens
laundering under better-typed vocabulary. Origin and grounding are now
orthogonal axes; measurement carries an origin, never replaces one.

gunbc#7786 inherited this error from this note.

P1: the sibling-registration paragraph asserted a fact that becomes false the
moment #7778 merges. Now merge-order-independent.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…ferenceAdmission and bound what it does not close
@briansrls
briansrls merged commit a44e79f into main Aug 4, 2026
2 of 3 checks passed
@briansrls
briansrls deleted the session/wise-ram-22-prb0 branch August 4, 2026 23:50
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