From 72bad21b2e311bbaf3ade92243c5314e7a8e7bab Mon Sep 17 00:00:00 2001 From: Brian Searls Date: Sat, 5 Sep 2026 20:32:42 +0000 Subject: [PATCH 1/2] Convert floor_cost_basis_boundedness_stall to the current trigger carrier MAIN DOES NOT RESOLVE TODAY. gunbc.guarantee_stall.roster refuses with 'type mismatch: expected Product(NextRungTrigger), got Primitive(String)' at floor_cost_basis_boundedness_stall. Every branch that merges main pays for it. THE ORDER, MEASURED IN BOTH DIRECTIONS BECAUSE MY FIRST READING HAD IT BACKWARDS. The carrier landed FIRST -- 86be37f355 (#10582) at 17:21:02 already declared next_rung_trigger: NextRungTrigger and supplied climbs_when. The row landed 74 minutes later, 45074b4e9c (#10281) at 18:35:47, spelling a bare String. git merge-base --is-ancestor answers YES carrier-before-row and NO the other way. So #10582 stranded nothing and this is NOT a migration that missed a caller: an old-format declaration landed onto a carrier that had already changed. THE INCIDENT IS DENOMINATOR LOSS, NOT A LYING GATE. The required-floor-disposition TSV (artifact 9975676354, run 33986383584 at 2f8819b891, checked against its published SHA-256) carries all ELEVEN identities of guarantee_stall_witness_test as declined_outside_gate_closure / not_executed. The floor reported honestly, per identity, that they did not run; the AGGREGATE reported SUCCESS over a population that had already excluded them. Discovering an identity so it can receive a disposition is not the same operation as preparing and executing its dependency closure. The narrowing is gunbc.recurring_failure_mode.preparation_membership_derived_from_the_execution_selector. THE CONVERSION IS MECHANICAL AND VERIFIED AS SUCH, not asserted: the capability text is byte-identical at 321 characters, and awaiting: [] discards nothing -- the original trigger carried no referent provenance. Before/after through the SAME compiler route and the same binary (md5 bf014f57, built 2026-09-04, which PREDATES this tree and is not claimed to be main's current compiler): the bad initializer refuses with the diagnostic above, the converted row resolves with zero resolve errors. Because both readings used one binary, its revision cannot explain the difference. LANDING THIS CLOSES NOTHING ABOUT ADMISSION. It restores resolution. The question it leaves open is sharper than the one I started with and no census at the carrier's own PR could have answered it: does admission validate an incoming declaration against the carrier current at placement? A clean merge and an active author do not establish compatibility with an intervening authority change. Co-Authored-By: Claude Opus 5 Claude-Session: https://claude.ai/code/session_01RvMEJFpYsy55NRdJ1Az4Vd --- .../floor_cost_basis_boundedness_stall.dag | 68 ++++++++++++++++++- 1 file changed, 66 insertions(+), 2 deletions(-) diff --git a/dag/gunbc/guarantee_stall/floor_cost_basis_boundedness_stall.dag b/dag/gunbc/guarantee_stall/floor_cost_basis_boundedness_stall.dag index d6fa6d98126..aeb8ebe350a 100644 --- a/dag/gunbc/guarantee_stall/floor_cost_basis_boundedness_stall.dag +++ b/dag/gunbc/guarantee_stall/floor_cost_basis_boundedness_stall.dag @@ -1,7 +1,7 @@ module gunbc.guarantee_stall.floor_cost_basis_boundedness_stall import gunbc.guarantee_rung { OutsideTheLadder, StructurallyGuaranteed } -import gunbc.guarantee_stall { GuaranteeStall, AwaitsOneGrounding, UncountedNotEnumerable } +import gunbc.guarantee_stall { GuaranteeStall, AwaitsOneGrounding, UncountedNotEnumerable, climbs_when } // NO MEASUREMENT CAMPAIGN CAN DISCHARGE THIS ARM, AND THAT IS WHY IT IS A STALL RATHER THAN UNBUILT // WORK. `gunbc.rung_drop.floor_cost_claim_qualification_unavailable` states that its restoration @@ -57,6 +57,70 @@ import gunbc.guarantee_stall { GuaranteeStall, AwaitsOneGrounding, UncountedNotE // drop's previous/temporary/reason fields have no answer here. The two rows are also different // facts: the drop states the standing unavailability, this states that one named restoration // conjunct is unreachable BY MEASUREMENT and what would reach it. +// THE CONVERSION BELOW IS A CARRIER REPAIR, NOT A CORRECTION TO THIS ROW'S CONTENT. Read this +// before drawing a conclusion about this row's author from the diff that added `climbs_when`. +// +// THE ORDER, MEASURED IN BOTH DIRECTIONS BECAUSE AN EARLIER DRAFT OF THIS COMMENT HAD IT +// BACKWARDS. The carrier landed FIRST: 86be37f355 (gunbc#10582) at 17:21:02 already declared +// `next_rung_trigger: NextRungTrigger` and supplied `climbs_when`. This row landed at 18:35:47, +// SEVENTY-FOUR MINUTES LATER, in 45074b4e9c (gunbc#10281), spelling the field as a bare `String`. +// `git merge-base --is-ancestor` answers YES in the carrier-before-row direction and NO in the +// other. So this is NOT a migration that forgot a caller, and gunbc#10582 stranded nothing: AN +// OLD-FORMAT DECLARATION LANDED ONTO A CARRIER THAT HAD ALREADY CHANGED. +// +// THE MEASURED CONSEQUENCE. 31 of the 33 rows in this directory use the constructor; this one did +// not, and the closure enumerating them refuses with `type mismatch: expected +// 'Product(NextRungTrigger)', got 'Primitive(String)'`. A branch merging main could not resolve +// that closure until this conversion was made. +// +// THE MECHANISM, IN ONE LINE: DISCOVERING AN IDENTITY SO IT CAN RECEIVE A DISPOSITION IS NOT THE +// SAME OPERATION AS PREPARING AND EXECUTING ITS DEPENDENCY CLOSURE. The witness DEPENDS ON the +// roster -- live folds reaching the trigger, not a dead import. The required run ADMITS that +// witness module into PREPARATION -- it does not. The first does not establish the second. +// +// THE INCIDENT, READ FROM THE ARTIFACT RATHER THAN INFERRED. The required-floor-disposition TSV +// (artifact 9975676354, run 33986383584 at 2f8819b891, checked against its published SHA-256) +// carries all ELEVEN identities of `test.claim.guarantee_stall_witness_test` as +// `disposition = declined_outside_gate_closure`, `outcome = not_executed` -- including +// `every_live_stall_names_a_next_rung_trigger` and `every_live_stall_trigger_is_admitted`. +// +// SO THE FLOOR DID NOT LIE, AND SAYING IT DID WOULD BE BOTH WRONG AND WEAKER. Every per-identity +// disposition is exactly right: the run reported, honestly and per identity, that these witnesses +// DID NOT RUN. What is wrong is that an AGGREGATE reported SUCCESS over a population that had +// already excluded them. A defender can rebut "the floor lied about eleven witnesses"; nobody can +// rebut "the floor said not_executed and the aggregate said success". +// +// THAT IS DENOMINATOR LOSS -- an aggregate verdict computed over an already-narrowed population, +// so success never establishes that the ADMITTED set survived. The same shape is carried as an +// open P0 finding at the install boundary; two boundaries exhibiting one shape is the fact worth +// recording, not two incidents. The narrowing itself is +// `gunbc.recurring_failure_mode.preparation_membership_derived_from_the_execution_selector`, which +// is the authority for it; this comment mints no second wording. +// +// WHAT THIS DOES NOT LICENSE, stated so nobody reaches for it: it is not an argument for compiling +// everything whenever the selector is uncertain. That is the absorbing fallback -- denominated in +// the corpus rather than the change, growing until the budget breaks instead of the build. A +// widening must be explicit, bounded, and credited only for the population it actually restores. +// +// TWO EXPLANATIONS WERE OFFERED FOR THE GREEN AND BOTH ARE WITHDRAWN, recorded because an +// unmeasured mechanism in a permanent comment is worse than no mechanism. The first was that the +// migration missed a caller -- refuted by the ordering above. The second was that the roster is +// enumerated somewhere the required floor does not compile -- refuted by the witnesses existing, +// consuming it through live folds, and being reported per identity. What is measured is that they +// were DISCOVERED and DECLINED, never prepared. +// +// THE QUESTION THIS OPENS is sharper than either withdrawn explanation, and no census at the +// carrier's own PR could have answered it: DOES ADMISSION VALIDATE AN INCOMING DECLARATION AGAINST +// THE CARRIER CURRENT AT PLACEMENT? A clean merge and an active author do not establish +// compatibility with an intervening authority change. The discriminating evidence for that work is +// an incoming old-format constructor under the new carrier, REFUSED BY AN EXECUTING REQUIRED +// CONSUMER, with the valid `climbs_when` construction still accepted; a roster-membership check +// establishes neither half. +// +// THE CONVERSION IS MECHANICAL AND VERIFIED AS SUCH: the capability text is byte-identical at 321 +// characters, and `awaiting: []` discards nothing -- the original trigger carried no referent +// provenance to lose. It restores resolution; IT CLOSES NOTHING about admission. + data floor_cost_basis_boundedness_stall: GuaranteeStall = GuaranteeStall { subject: "the BOUNDEDNESS-BY-CONSTRUCTION arm of conjunct (ii) of gunbc.rung_drop.floor_cost_claim_qualification_unavailable's restoration trigger: repeated execution measurement cannot establish it, because every observed maximum is a LOWER bound on the smallest valid bound rather than an upper bound over the admitted envelope", current: OutsideTheLadder, @@ -67,5 +131,5 @@ data floor_cost_basis_boundedness_stall: GuaranteeStall = GuaranteeStall { population: UncountedNotEnumerable { reason: "the subject is ONE ARM of conjunct (ii) of a restoration trigger authored as prose inside gunbc.rung_drop.floor_cost_claim_qualification_unavailable. The population this carrier reports is the EXPOSURE, not a roster of the source artifacts that carry or discuss the subject, and the exposed set here is every required identity for which the CPU deadline is armed, derived per run. That set is produced by execution rather than declared in source, so no authored list in this standing row would be true when written. It is finite, not mathematically infinite, and a run may well emit its own occurrence roster; what does not exist is a stable source-level authority this row could name. TWO EARLIER REVISIONS OF THIS FIELD WERE WRONG IN DIFFERENT WAYS, recorded because the second failure is the instructive one. The first list named gunbc.floor_cost_distribution.RunCost.eval_steps, a field that does not exist -- RunCost declares run and rows -- so its BoundedPopulation was not an identity join and the bound was never real. The second revision substituted a symbol that does exist, v2.workflow.required_floor.claim_safety_outcome, and was still wrong, because the defect was the SHAPE rather than the bad member: the list enumerated CARRIERS -- an instrument column, a deciding consumer, and the parent drop -- which is one-carrier-answers-for-the-whole-population, the move `gunbc.guarantee_stall.two_nat_authorities_stall` rejects under this same StallPopulation carrier even though its own declaration half and four direct importers are perfectly enumerable. THE FUTURE BOUNDARY IS NOT A TYPED SPLIT OF THE PARENT TRIGGER. Giving the arm a typed source identity would make the SUBJECT referable and may be worth doing on its own, but it neither enumerates the exposed required identities nor is needed to; naming that new arm identity here would repeat the same rejected move one level down. What would permit an honest BoundedPopulation is a typed PRODUCER of the exposed population at this grain -- a derivation joining a run's required-identity dispositions to the CpuDeadlineArmed policy, or another closed authority that enumerates every affected identity", }, - next_rung_trigger: "the construction-bound authority named in this row's grounding EXISTS and is CONSUMED by the required-floor verdict at the declared subject grain, with its over-bound control refused on the real acceptance path; no volume of execution sampling retires this row, and a widening of the measurement campaign does not move it" + next_rung_trigger: climbs_when(capability: "the construction-bound authority named in this row's grounding EXISTS and is CONSUMED by the required-floor verdict at the declared subject grain, with its over-bound control refused on the real acceptance path; no volume of execution sampling retires this row, and a widening of the measurement campaign does not move it"), } From 02605c5b35ee41e57413bc291d8363f3b5dbc044 Mon Sep 17 00:00:00 2001 From: Brian Searls Date: Sat, 5 Sep 2026 21:45:03 +0000 Subject: [PATCH 2/2] Withdraw the refuted explanation, keep the measured exclusion, drop the attribution Four wording repairs to this row's annotation. The conversion itself is unchanged and the capability text stays byte-identical. 1. A CONTRADICTION IS RESOLVED BY SPLITTING TWO CLAIMS AN EARLIER DRAFT COLLAPSED. WITHDRAWN: "the roster is enumerated somewhere the floor does not compile" -- inadequate, because the witnesses exist and consume the roster through live folds. NOT WITHDRAWN, because it is measured from the artifact: those eleven witnesses were declined before preparation and did not execute. The witnesses existing refutes only the claim that nothing reaches the roster; it does not refute their exclusion. 2. The aggregate accusation is replaced by the narrower measured statement. A run succeeding for its SELECTED work did not establish that this declaration satisfied the current carrier. DISCOVERY-ADMITTED, PREPARED and SELECTED-FOR-EXECUTION are three populations and must not silently become one. 3. "gunbc#10582 stranded nothing" is localised to "DID NOT STRAND THIS ROW, WHICH LANDED LATER" -- the ordering measurement covers this row and says nothing about whether every other constructor was migrated correctly. 4. The attribution heading is removed entirely. Commit order establishes neither fault nor its absence without branch and review context, and this repair needs no attribution investigation to be correct. Also: the edit is no longer described as "no semantic change". It is a representation and type-correctness repair preserving the row's substantive capability requirement -- an incompatible String initializer becoming a product construction is compiler-observable, which is the point of making it. Co-Authored-By: Claude Opus 5 Claude-Session: https://claude.ai/code/session_01RvMEJFpYsy55NRdJ1Az4Vd --- .../floor_cost_basis_boundedness_stall.dag | 88 +++++++++---------- 1 file changed, 41 insertions(+), 47 deletions(-) diff --git a/dag/gunbc/guarantee_stall/floor_cost_basis_boundedness_stall.dag b/dag/gunbc/guarantee_stall/floor_cost_basis_boundedness_stall.dag index aeb8ebe350a..da14e6c38b4 100644 --- a/dag/gunbc/guarantee_stall/floor_cost_basis_boundedness_stall.dag +++ b/dag/gunbc/guarantee_stall/floor_cost_basis_boundedness_stall.dag @@ -57,69 +57,63 @@ import gunbc.guarantee_stall { GuaranteeStall, AwaitsOneGrounding, UncountedNotE // drop's previous/temporary/reason fields have no answer here. The two rows are also different // facts: the drop states the standing unavailability, this states that one named restoration // conjunct is unreachable BY MEASUREMENT and what would reach it. -// THE CONVERSION BELOW IS A CARRIER REPAIR, NOT A CORRECTION TO THIS ROW'S CONTENT. Read this -// before drawing a conclusion about this row's author from the diff that added `climbs_when`. +// THE CONVERSION BELOW IS A CARRIER REPAIR, NOT A CORRECTION TO THIS ROW'S CONTENT. It makes NO +// CLAIM ABOUT ANYONE'S CONDUCT: commit order establishes neither fault nor its absence without +// branch and review context, and this repair needs no attribution investigation to be correct. // -// THE ORDER, MEASURED IN BOTH DIRECTIONS BECAUSE AN EARLIER DRAFT OF THIS COMMENT HAD IT -// BACKWARDS. The carrier landed FIRST: 86be37f355 (gunbc#10582) at 17:21:02 already declared +// WHAT IT IS, ACCURATELY. A REPRESENTATION AND TYPE-CORRECTNESS REPAIR PRESERVING THIS ROW'S +// SUBSTANTIVE CAPABILITY REQUIREMENT. It is NOT a no-semantic-change edit: an incompatible +// `String` initializer becoming a product construction is compiler-observable, and that is the +// entire point of making it. +// +// THE ORDER, MEASURED IN BOTH DIRECTIONS BECAUSE AN EARLIER DRAFT HAD IT BACKWARDS. The carrier +// landed FIRST: 86be37f355 (gunbc#10582) at 17:21:02 already declared // `next_rung_trigger: NextRungTrigger` and supplied `climbs_when`. This row landed at 18:35:47, // SEVENTY-FOUR MINUTES LATER, in 45074b4e9c (gunbc#10281), spelling the field as a bare `String`. -// `git merge-base --is-ancestor` answers YES in the carrier-before-row direction and NO in the -// other. So this is NOT a migration that forgot a caller, and gunbc#10582 stranded nothing: AN -// OLD-FORMAT DECLARATION LANDED ONTO A CARRIER THAT HAD ALREADY CHANGED. +// `git merge-base --is-ancestor` answers YES carrier-before-row and NO the other way. So gunbc#10582 +// DID NOT STRAND THIS ROW, WHICH LANDED LATER -- and that is all the ordering establishes; it says +// nothing about whether every OTHER constructor was migrated correctly. // // THE MEASURED CONSEQUENCE. 31 of the 33 rows in this directory use the constructor; this one did // not, and the closure enumerating them refuses with `type mismatch: expected // 'Product(NextRungTrigger)', got 'Primitive(String)'`. A branch merging main could not resolve // that closure until this conversion was made. // -// THE MECHANISM, IN ONE LINE: DISCOVERING AN IDENTITY SO IT CAN RECEIVE A DISPOSITION IS NOT THE -// SAME OPERATION AS PREPARING AND EXECUTING ITS DEPENDENCY CLOSURE. The witness DEPENDS ON the -// roster -- live folds reaching the trigger, not a dead import. The required run ADMITS that -// witness module into PREPARATION -- it does not. The first does not establish the second. -// -// THE INCIDENT, READ FROM THE ARTIFACT RATHER THAN INFERRED. The required-floor-disposition TSV -// (artifact 9975676354, run 33986383584 at 2f8819b891, checked against its published SHA-256) -// carries all ELEVEN identities of `test.claim.guarantee_stall_witness_test` as -// `disposition = declined_outside_gate_closure`, `outcome = not_executed` -- including -// `every_live_stall_names_a_next_rung_trigger` and `every_live_stall_trigger_is_admitted`. +// THE INCIDENT, AT ITS MEASURED GRAIN AND NO WIDER. After the product carrier landed, this +// old-format declaration was added to main. The examined required run succeeded for its selected +// work while all eleven roster-consuming witnesses were explicitly declined before preparation and +// not executed. THAT SUCCESS DID NOT ESTABLISH THAT THIS DECLARATION SATISFIED THE CURRENT +// CARRIER. This conversion repairs the initializer, not admission. // -// SO THE FLOOR DID NOT LIE, AND SAYING IT DID WOULD BE BOTH WRONG AND WEAKER. Every per-identity -// disposition is exactly right: the run reported, honestly and per identity, that these witnesses -// DID NOT RUN. What is wrong is that an AGGREGATE reported SUCCESS over a population that had -// already excluded them. A defender can rebut "the floor lied about eleven witnesses"; nobody can -// rebut "the floor said not_executed and the aggregate said success". -// -// THAT IS DENOMINATOR LOSS -- an aggregate verdict computed over an already-narrowed population, -// so success never establishes that the ADMITTED set survived. The same shape is carried as an -// open P0 finding at the install boundary; two boundaries exhibiting one shape is the fact worth -// recording, not two incidents. The narrowing itself is +// THE MECHANISM, IN ONE LINE: DISCOVERING AN IDENTITY SO IT CAN RECEIVE A DISPOSITION IS NOT THE +// SAME OPERATION AS PREPARING AND EXECUTING ITS DEPENDENCY CLOSURE. Read from the artifact rather +// than inferred: the required-floor-disposition TSV (artifact 9975676354, run 33986383584 at +// 2f8819b891, checked against its published SHA-256) carries all ELEVEN identities of +// `test.claim.guarantee_stall_witness_test` as `declined_outside_gate_closure` / `not_executed`. +// The narrowing itself is // `gunbc.recurring_failure_mode.preparation_membership_derived_from_the_execution_selector`, which -// is the authority for it; this comment mints no second wording. +// is the authority for that shape; this comment mints no second wording for it. // -// WHAT THIS DOES NOT LICENSE, stated so nobody reaches for it: it is not an argument for compiling -// everything whenever the selector is uncertain. That is the absorbing fallback -- denominated in -// the corpus rather than the change, growing until the budget breaks instead of the build. A -// widening must be explicit, bounded, and credited only for the population it actually restores. +// ONE EXPLANATION IS WITHDRAWN AND ONE DEMONSTRATED FACT IS NOT, and an earlier draft of this +// paragraph collapsed them. WITHDRAWN: "the roster is enumerated somewhere the floor does not +// compile" -- inadequate, because the witnesses exist and consume the roster through live folds. +// NOT WITHDRAWN, because it is measured: those witnesses were EXCLUDED FROM PREPARATION and did +// not execute. The witnesses existing does not refute their exclusion; it refutes only the claim +// that nothing reaches the roster. // -// TWO EXPLANATIONS WERE OFFERED FOR THE GREEN AND BOTH ARE WITHDRAWN, recorded because an -// unmeasured mechanism in a permanent comment is worse than no mechanism. The first was that the -// migration missed a caller -- refuted by the ordering above. The second was that the roster is -// enumerated somewhere the required floor does not compile -- refuted by the witnesses existing, -// consuming it through live folds, and being reported per identity. What is measured is that they -// were DISCOVERED and DECLINED, never prepared. +// THE BUILD LANE DOES NOT ESTABLISH THIS EITHER, which is why no CI green is cited above as +// evidence. Main's required-witnesses-build at 2f8819b891 reported SUCCESS with this exact type +// error present. Compiling the DECLARATION of `NextRungTrigger` does not establish that an +// INITIALIZER using it was checked, and the run's own identity census makes that distinction. // -// THE QUESTION THIS OPENS is sharper than either withdrawn explanation, and no census at the -// carrier's own PR could have answered it: DOES ADMISSION VALIDATE AN INCOMING DECLARATION AGAINST -// THE CARRIER CURRENT AT PLACEMENT? A clean merge and an active author do not establish -// compatibility with an intervening authority change. The discriminating evidence for that work is -// an incoming old-format constructor under the new carrier, REFUSED BY AN EXECUTING REQUIRED -// CONSUMER, with the valid `climbs_when` construction still accepted; a roster-membership check -// establishes neither half. +// WHAT THIS DOES NOT LICENSE: it is not an argument for compiling everything whenever the selector +// is uncertain. That is the absorbing fallback -- denominated in the corpus rather than the change, +// growing until the budget breaks instead of the build. A widening must be explicit, bounded, and +// credited only for the population it actually restores. // // THE CONVERSION IS MECHANICAL AND VERIFIED AS SUCH: the capability text is byte-identical at 321 -// characters, and `awaiting: []` discards nothing -- the original trigger carried no referent -// provenance to lose. It restores resolution; IT CLOSES NOTHING about admission. +// characters, and `awaiting: []` discards nothing -- the original trigger carried no +// delivering-lane provenance to lose. It restores resolution; IT CLOSES NOTHING about admission. data floor_cost_basis_boundedness_stall: GuaranteeStall = GuaranteeStall { subject: "the BOUNDEDNESS-BY-CONSTRUCTION arm of conjunct (ii) of gunbc.rung_drop.floor_cost_claim_qualification_unavailable's restoration trigger: repeated execution measurement cannot establish it, because every observed maximum is a LOWER bound on the smallest valid bound rather than an upper bound over the admitted envelope",