P104 helper reordering, paid interior bases and independent split composition

[aliromman]

FINDING / SCOPE UPDATE — the planned composition is superseded by Joel Pulikkan's PR #5684, head d0f6fa2555410a1e23fe4191424c2097ab3d5d97, published during replay. I will not open a duplicate PR or count this result.

The smaller composition is retained at 39c90dc4c2bce215f0c9a1c8aeacab31df0e811f. Normal and explicit optimized registered replay pass, including 2,916 native maps and all30,096 common-frame split contexts. Its conditional .00310819637 is below PR5684's .00310871312. This is local author evidence, not independent acceptance of PR5684.

The narrow continuation is to check whether PR5684's changed word creates a new independent helper cut, including distinct physical cut and boundary roles. No R22 reordering or native-frame search is reserved. Discussion5645's native-frame work, Discussion5600's retired-input binding and Discussion5125's coded caller remain separate. A failed cut screen will be released and not counted as a constructive contribution. Expiry 2026-10-26T07:00:00Z.

Credit Joel Pulikkan with Claude assistance, PR5678, PR5660, PR5599 and every retained notice. Ali Romman with substantial OpenAI Codex assistance; no subagents. No selected-frontier change.

Original claim

CLAIM — compose P104 helper-start reordering with paid interior bases, then test new independent splits.

Base: canonical main df0137e after a fresh fetch. Priority revision 7 and selected conditional kappa .00310045 are unchanged. The live cache has complete pagination over 366 open PRs and 4,993 Discussion records through 2026-10-26T05:08:00Z. PR5678's current body and the active Discussion5125, 5600 and 5645 scopes were read.

Exact inputs: PR5666 at 693a220, by Joel Pulikkan / GamingPuzzled with Claude assistance; and PR5678 at af814f6, by Ali Romman with Codex assistance. PR5678 explicitly does not consume or claim additivity with PR5666. Its four paid interior changes concern different fields from helper473's endpoint and its two contact reorderings. Literal source matching and full replay must decide compatibility.

I will emit their exact composition and then test all independent helper cuts on that changed word, including a different physical cut and boundary role. A new input can exchange with a cut role at one common frame; the final output exchange moves both the full array value and its frame header, with positive ordinary routes. All original dirty erasures stay on their original owned fields. No paid native conjugation is assumed free. A positive cut must preserve every independent input, every exact native phase and the complete physical caller.

Success is a complete changed kappa bill, not an addition of published decimals. Recheck actual helper targets, entry and exit maps, every paid request, precision, compiler fees and all outer conditions. If the composition or the extended cut fails, report that scope without counting a negative result. Continue compatible positive cuts until this strategy has no further legal step. The current repeated-binding claim in Discussion5600 explicitly adds no independent target; native-frame regions in5645 and the coded caller in5125 stay separate.

Credit PR5666 and PR5678, PR5660, zxv PR5643/5650, PR5644, PR5627, PR5599's independent-split construction, and the retained P104/native/controller/compiler lineage. Ali Romman with substantial OpenAI Codex assistance; no subagents. About two hours. No merge or selected-frontier change.

[aliromman]

DONE / RELEASED — PR5684 at d0f6fa2 has zero independent-helper pairing in all30,090 eligible common-frame contexts across45 eligible boundary groups (47 groups considered). This exact rational check permits distinct physical boundary and cut roles. Evidence is local-only in /tmp/kappa-ten3-r05-post5684-cuts.json; no new construction or kappa is claimed. It does not exclude changed scalar words or paid transports. No continuation is reserved. The smaller composition at39c90dc4c2bce215f0c9a1c8aeacab31df0e811f is superseded and is not counted as a contribution. Credit Joel Pulikkan PR5684, PR5678, PR5599 and retained source authors. Ali Romman with substantial OpenAI Codex assistance.

[Clear1nsight]

HANDOFF — request for an existing coverage artifact, to avoid repeating the released cut screen.

Research discussion / PR / artifact: Your released PR #5684 cut-screen result.

Exact commit: d0f6fa2555410a1e23fe4191424c2097ab3d5d97, PR #5684.

What is established: Your checkpoint reports zero independent-helper pairings over 30,090 eligible common-frame contexts and 45 eligible boundary groups, with 47 groups considered and distinct cut/boundary roles allowed. We retain this as author evidence and are not treating the known failed domain as unexplored.

What remains uncertain: The exact eligibility predicate, identities of the 45 eligible and two excluded groups, and the boundary/cut coverage represented by the local JSON. We have not independently inspected that artifact or inferred a universal split obstruction.

Smallest useful next task: If an existing immutable export, script plus manifest, or group/cut table is available, please link it here; an explicit not-public/open status also suffices. We are asking for the already completed coverage, not a new scan, replay or construction. That will let a later proposal name a genuine coverage distinction before any lookup.

Reproduction command and expected runtime: Source/pointer review only; no new execution is requested.

Predecessor credit: Ali Romman's released screen and PR #5599 independent-input criterion; Joel Pulikkan's PR #5684, PR #5678, and the retained P104/native/helper/compiler lineage. Prepared for Clear1nsight with substantial OpenAI Codex assistance. No new candidate, price, κ claim or scope reservation is made.

[aliromman]

HANDOFF / EXISTING COVERAGE — the exact JSON receipt remains local-only. There is no public immutable export of that JSON, and I do not present it as an independently replayed certificate. The checker itself is already public at cuts.py. Its companion source bootstrap is in the same superseded package. This answers the artifact request without a new scan.

Input: PR5684 d0f6fa2, research/p104-reframe-r22/WORD.json.gz,743 public fields,9 owned fields and104 live main ports. A helper is any public field outside live and live+112. Group helpers by the exact ordered pair(initial frame,final frame). A cut frame is eligible only when its enumerated span strictly contains the entrance span and is strictly contained in the exit span. The bank includes all public and owned roles at that actual frame; no same-role boundary/cut pivot is required. Rational prefix and suffix elimination removes all input/output coordinates outside the helper group. The recorded predicate tests whether any pure prefix functional c and pure suffix direction d have c*d nonzero. The later ab=cd normalization and dyadic synthesis are not reached when all pairings vanish.

Correction to the earlier short description: the30090 count is the number of checked operation-boundary bank contexts, not a deduplicated count of distinct matrix states. The scan includes the entrance and both affected banks at every MOVE/INIT/CLEAR and the affected common bank at each scalar event. It is not a search over unlisted frames, changed native words or different boundary groups.

There are47 boundary groups. The excluded groups are(477,127), one helper741, and(267,127), one helper742: neither has an eligible intermediate frame in this word catalogue. All45 eligible group rows are below as(entry,exit,helper count,checked contexts); each recorded zero nonzero pairings.

3,127,412,8959
3,232,4,240
3,233,1,374
3,234,3,210
3,235,4,231
3,237,3,220
3,238,1,371
3,240,3,259
3,241,1,493
3,242,4,240
3,243,4,266
3,244,4,237
3,245,4,2788
3,246,8,2781
3,252,1,23
3,267,1,2208
3,306,1,188
3,391,1,30
3,424,1,23
3,462,1,23
3,477,1,2210
3,516,1,190
3,559,1,950
3,626,1,149
114,127,4,209
115,127,4,212
116,127,4,255
117,127,4,262
118,127,4,229
120,127,5,223
252,127,1,533
306,127,1,192
391,127,1,478
424,127,1,486
462,127,1,472
516,127,1,239
603,127,1,761
1027,127,4,235
1031,127,4,260
1048,127,4,252
1049,127,5,206
1050,127,4,210
1051,127,5,242
1052,127,4,228
1053,127,5,243

The public checker plus pinned word can reproduce the domain; its exact source environment is retained in the superseded package. The local script invoked run(word) after loading its frozen local module. No fresh runtime estimate or rerun is asserted here. Credit PR5599/5609 and Joel Pulikkan PR5684, PR5678 and the retained native/caller sources. Ali Romman with substantial OpenAI Codex assistance. No new construction or kappa is claimed.