From cb2174532ef8e9483b8dd53e36bd3ebdfec75f27 Mon Sep 17 00:00:00 2001 From: Brian Searls Date: Tue, 6 Oct 2026 16:07:36 +0000 Subject: [PATCH 1/2] Repoint the v2.std.optional imports #13367 landed after #13388's re-home #13367 (96479e6103) merged first and added five imports of v2.std.optional; #13388 (a453be995a) then re-homed that module to std.optional, so main refuses in every generated job (CarrierRefused: unresolved import v2.std.optional, from dag/extdeps/llm/claude_code_stream_json.dag). This is #13388's own instrument, tools.source_reference_repoint repoint, run over the whole tree at a453be995a (TREE_EXACT): it rewrote exactly these five import paths. Present and Absent are declared in std.optional. String fixtures are untouched by design. Co-Authored-By: Claude Opus 5.5 (1M context) --- dag/extdeps/llm/claude_code_stream_json.dag | 2 +- dag/gunbc/claude_code_credential.dag | 2 +- dag/gunbc/claude_code_limit_standing.dag | 2 +- dag/test/claim/claude_code_dispatch_witness_test.dag | 2 +- .../roadmap/roadmap_belt_base_advance_wet_witness_test.dag | 2 +- 5 files changed, 5 insertions(+), 5 deletions(-) diff --git a/dag/extdeps/llm/claude_code_stream_json.dag b/dag/extdeps/llm/claude_code_stream_json.dag index e9c18fc3d4c..489451c25c8 100644 --- a/dag/extdeps/llm/claude_code_stream_json.dag +++ b/dag/extdeps/llm/claude_code_stream_json.dag @@ -6,7 +6,7 @@ import std.measure { BasisPoint, basis_point } import std.decimal { ExactDecimal, decimal_pow10 } import std.checked_arithmetic { checked_int_to_nat } import std.decl_ref { DeclarationRef, WholeDeclaration } -import v2.std.optional { Present, Absent } +import std.optional { Present, Absent } import extdeps.external_authority { ExternalAuthority, ExternalModelScope, ExternalSubjectRef } import extdeps.uri { Uri, Https } import extdeps.languages.json.parse { diff --git a/dag/gunbc/claude_code_credential.dag b/dag/gunbc/claude_code_credential.dag index f665f9818e3..d1796d7af55 100644 --- a/dag/gunbc/claude_code_credential.dag +++ b/dag/gunbc/claude_code_credential.dag @@ -1,7 +1,7 @@ module gunbc.claude_code_credential import std.types { String, List, NonEmptyStr } -import v2.std.optional { Present, Absent } +import std.optional { Present, Absent } import extdeps.cloud.gcp.secret_ref { SecretRef } import gunbc.secret_provision { fleet_secret_ref, secret_ref_pin_version } import extdeps.llm.claude_setup_token_cli { claude_code_oauth_token_env_var } diff --git a/dag/gunbc/claude_code_limit_standing.dag b/dag/gunbc/claude_code_limit_standing.dag index f6919b73357..dbbb4a553ab 100644 --- a/dag/gunbc/claude_code_limit_standing.dag +++ b/dag/gunbc/claude_code_limit_standing.dag @@ -3,7 +3,7 @@ module gunbc.claude_code_limit_standing import std.types { String, List, Bool, Int, NonEmptyStr } import std.measure { basis_point_count, basis_point_unity_count } import std.algebra { trim } -import v2.std.optional { Present, Absent } +import std.optional { Present, Absent } import extdeps.llm.claude_code_stream_json { ClaudeCodeStreamLine, ClaudeCodeRateLimitEventLine, diff --git a/dag/test/claim/claude_code_dispatch_witness_test.dag b/dag/test/claim/claude_code_dispatch_witness_test.dag index ee0c4cdd739..77b9d605a4d 100644 --- a/dag/test/claim/claude_code_dispatch_witness_test.dag +++ b/dag/test/claim/claude_code_dispatch_witness_test.dag @@ -4,7 +4,7 @@ import std.types { Bool, Int, List, NonEmptyStr, String, FilePath } import v2.std.live_tree { LiveTreeDisposition, SubstrateInputsOnly } import std.claim_evidence { RecordedFactId } import std.measure { basis_point_count } -import v2.std.optional { Present, Absent } +import std.optional { Present, Absent } import gunbc.roadmap_model { RoadmapNodeId } import extdeps.llm.claude_code_stream_json { ClaudeCodeStreamLine, diff --git a/dag/test/claim/roadmap/roadmap_belt_base_advance_wet_witness_test.dag b/dag/test/claim/roadmap/roadmap_belt_base_advance_wet_witness_test.dag index 39b75f486c6..d482f213005 100644 --- a/dag/test/claim/roadmap/roadmap_belt_base_advance_wet_witness_test.dag +++ b/dag/test/claim/roadmap/roadmap_belt_base_advance_wet_witness_test.dag @@ -13,7 +13,7 @@ module test.claim.roadmap.roadmap_belt_base_advance_wet_witness_test import std.logic { Bool } import std.types { String, FilePath, NonEmptyStr, List } -import v2.std.optional { Present, Absent } +import std.optional { Present, Absent } import v2.std.live_tree { LiveTreeDisposition, SubstrateInputsOnly } import extdeps.shell import extdeps.filesystem.filesystem_io { Filesystem } From f82b2e9f630a8574697c4258ebdba7850ca32191 Mon Sep 17 00:00:00 2001 From: Brian Searls Date: Tue, 6 Oct 2026 17:21:35 +0000 Subject: [PATCH 2/2] Declare the identity-cast route claim's eval-step cost drop (its fill-debt standing was retired by #13388) v2.test.claim.coercion.identity_cast_emission_route an_admitted_identity_cast_emits_through_the_closure_route_holds is a required_gate_authored_modules member, so it runs on every PR. It was carried as single-claim fill debt until #13388 retired that row as BecameSharedByDemand; nothing replaced the standing, and with main resolving again it now blocks every floor: verdict holds, 657,428 eval steps against the new-witness budget of 72,300 (runs 37486290875 and 37493248052). Own list floor_eval_step_cost_drop_identity_cast_route_rows plus the declared 4b(3) drop gunbc.rung_drop identity_cast_route_new_witness_eval_step_cost (MechanicallyPreventable -> Mitigatable; the claim still executes and a semantic red or wall crossing still blocks; trigger: the route claims restructured to supplied inputs, DESIGN section 3 witness rule, keeping one real-route inhabitance claim). Co-Authored-By: Claude Opus 5.5 (1M context) --- ..._cast_route_new_witness_eval_step_cost.dag | 43 +++++++++++++++++++ src/v2/workflow/floor_eval_step_cost_drop.dag | 16 ++++++- 2 files changed, 58 insertions(+), 1 deletion(-) create mode 100644 dag/gunbc/rung_drop/identity_cast_route_new_witness_eval_step_cost.dag diff --git a/dag/gunbc/rung_drop/identity_cast_route_new_witness_eval_step_cost.dag b/dag/gunbc/rung_drop/identity_cast_route_new_witness_eval_step_cost.dag new file mode 100644 index 00000000000..ea882475125 --- /dev/null +++ b/dag/gunbc/rung_drop/identity_cast_route_new_witness_eval_step_cost.dag @@ -0,0 +1,43 @@ +module gunbc.rung_drop.identity_cast_route_new_witness_eval_step_cost + +import std.types { String, List, NonEmptyStr } +import gunbc.rung_drop { RungDrop, Standing, TypedDeclaration, ReplacementStaged } +import gunbc.guarantee_rung { Mitigatable, MechanicallyPreventable } +import v2.workflow.floor_eval_step_cost_drop { floor_eval_step_cost_drop_identity_cast_route_rows } + +// DECLARED 4b(3) DROP (gunbc#13487, 2026-10-06; directed by sharp-raven-357 to restore main). The +// DECLARATION lives here; the POPULATION's authority is `v2.workflow.floor_eval_step_cost_drop` +// `floor_eval_step_cost_drop_identity_cast_route_rows`. It is its own list because growing another +// drop's list is a new drop. +// +// WHAT IS LOST IS ONE WALL, AND THE CLAIM IS UNTOUCHED. The identity stays on the gate: planned, +// executed and measured on every run, its eval_steps recorded; a semantic red and a wall-clock +// crossing still block. Only an eval-step overrun is reported under the declared drop rather than +// refusing. The budget is not raised. +// +// HOW THE STANDING WAS LOST. The claim was carried as single-claim fill debt; gunbc#13388 retired that +// row as BecameSharedByDemand, correctly, because its fixture producer is now demanded by several +// claims. Nothing replaced the standing, and while main could not resolve (the v2.std.optional +// conflict this PR repairs) the floor never reached the claim, so the loss was invisible until the +// repair made the corpus resolve again. +data identity_cast_route_new_witness_eval_step_cost_population: List = floor_eval_step_cost_drop_identity_cast_route_rows |> map(m => concat(concat(m.identity, ": over the new-witness eval-step budget, rostered in v2.workflow.floor_eval_step_cost_drop floor_eval_step_cost_drop_identity_cast_route_rows; measured by "), m.measured_by)) + +data identity_cast_route_new_witness_eval_step_cost: RungDrop = RungDrop { + identity: "identity_cast_route_new_witness_eval_step_cost" as NonEmptyStr, + + subject: "new-witness eval-step cost gate over the one real-route claim of the admitted identity cast: it still executes, eval_steps stay recorded, a semantic red and a wall-clock crossing still block; only the eval-step cost-gate rung is lowered", + + declared: "2026-10-06", + + standing: Standing, + + declaration: TypedDeclaration { + previous: MechanicallyPreventable, + temporary: Mitigatable, + reason: ReplacementStaged { + replacement: "the identity-cast route claims restructured to supplied inputs (DESIGN section 3, a witness discriminates at one interface), keeping exactly one inhabitance claim of the real route", + }, + population: identity_cast_route_new_witness_eval_step_cost_population, + restoration_trigger: "THE CAPABILITY: the identity-cast route claims restructured to supplied inputs (DESIGN section 3 witness rule), so that every claim over the route's verdict reads a supplied emission and exactly ONE inhabitance claim executes the real route. WHAT THAT MUST BE SUFFICIENT FOR: that one real-route identity, still emitting the admitted cast through the closure route and still asserting the route, measures under the NewWitnessTier budget that v2.workflow.required_floor claim_ceiling_eval_step_budget derives. Deleting the identity, supplying the emission to the last real-route claim (which removes the last execution of the real path), or relocating it off the required gate satisfies none of it.", + } +} diff --git a/src/v2/workflow/floor_eval_step_cost_drop.dag b/src/v2/workflow/floor_eval_step_cost_drop.dag index 0e282649911..23d58816127 100644 --- a/src/v2/workflow/floor_eval_step_cost_drop.dag +++ b/src/v2/workflow/floor_eval_step_cost_drop.dag @@ -36,6 +36,8 @@ import std.types { NonEmptyStr, List } // approval-request constructor (`approval_intent_digest_new_witness_eval_step_cost`). // - `floor_eval_step_cost_drop_dag_text_round_trip_rows`: the two nested-Bind text round trips // of the dag target (`dag_text_round_trip_nested_bind_new_witness_eval_step_cost`). +// - `floor_eval_step_cost_drop_identity_cast_route_rows`: the one real-route claim of the admitted +// identity cast (`identity_cast_route_new_witness_eval_step_cost`). // The loss is a WORKFLOW fact -- which // identities the eval-step ceiling does not refuse -- so it lives beside the other two rosters the // ceiling consults (`v2.workflow.floor_grandfathered_roster`, `v2.workflow.floor_cost_debt`) and @@ -419,8 +421,20 @@ data floor_eval_step_cost_drop_dag_text_round_trip_rows: List = [ + EvalStepCostDropMeasurement { + identity: "v2.test.claim.coercion.identity_cast_emission_route.an_admitted_identity_cast_emits_through_the_closure_route_holds" as NonEmptyStr, + measured_by: "required floor runs 37486290875 (branch fix/claude-code-std-optional at 675c2c68c2) and 37493248052 (gunbc#13487 at cb21745), both main a453be995a plus the v2.std.optional repoint: COMPLETED-OVER-COST-REQUIREMENT for this identity, verdict holds, 657,428 eval steps against the new-witness budget of 72,300; re-derived by the required floor on any pull request, since its module is a required_gate_authored_modules member. The billed work is the one real emission of the admitted identity cast through the closure route, which this claim is the only execution of" as NonEmptyStr, + }, +] + fn floor_eval_step_cost_drop_all_rows() -> List { - concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(floor_eval_step_cost_drop_rows, floor_eval_step_cost_drop_apply_render_rows), floor_eval_step_cost_drop_live_forecast_rows), floor_eval_step_cost_drop_page_style_rows), floor_eval_step_cost_drop_interpreted_crypto_rows), floor_eval_step_cost_drop_app_attest_verifier_rows), floor_eval_step_cost_drop_span_program_rows), floor_eval_step_cost_drop_python_to_typescript_inhabitance_rows), floor_eval_step_cost_drop_boot_matrix_rows), floor_eval_step_cost_drop_dark_install_render_rows), floor_eval_step_cost_drop_build_job_membership_rows), floor_eval_step_cost_drop_dag_emit_round_trip_rows), floor_eval_step_cost_drop_approval_intent_digest_rows), floor_eval_step_cost_drop_dag_text_round_trip_rows) + concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(concat(floor_eval_step_cost_drop_rows, floor_eval_step_cost_drop_apply_render_rows), floor_eval_step_cost_drop_live_forecast_rows), floor_eval_step_cost_drop_page_style_rows), floor_eval_step_cost_drop_interpreted_crypto_rows), floor_eval_step_cost_drop_app_attest_verifier_rows), floor_eval_step_cost_drop_span_program_rows), floor_eval_step_cost_drop_python_to_typescript_inhabitance_rows), floor_eval_step_cost_drop_boot_matrix_rows), floor_eval_step_cost_drop_dark_install_render_rows), floor_eval_step_cost_drop_build_job_membership_rows), floor_eval_step_cost_drop_dag_emit_round_trip_rows), floor_eval_step_cost_drop_approval_intent_digest_rows), floor_eval_step_cost_drop_dag_text_round_trip_rows), floor_eval_step_cost_drop_identity_cast_route_rows) } // THE PROJECTION THE SEED RUNNER READS AND THE MODEL TESTS AGAINST. The runner refuses an empty or