Assessment 223: independent stores without common codes

12 October 2026. Written native construction 76, with two readout contracts. No root admission or external priority claim.

Contents

Result and dependent extension

The complete argument begins with two independently prepared stores and closes two directly dependent extensions. There may be unboundedly many stores globally if each source mentions only a fixed number. That per-source bound also disappears when ordinary target prefixes pay routing to a single store. The latter clause changes global target identity: identical local suffixes in distinct store namespaces are distinct targets.

Each source/store lane retains its own committed union and pending row. The actual local alias services only that lane’s deferred work. This avoids charging a long old operation to a shorter alias from another store. Literal directory lookups are paid by actual store-name prefixes before local alias or target service begins. Local code equality is never exported across stores.

The producer checks once-only contiguous store groups, learns local widths independently and checks each group’s ready density gate when leaving it. The number of names, groups and retained versions is unbounded. No external heap, supplied index or free ready handle is assumed. The native language consequence concerns checked Horn SAT in the stated input orientation; PEG recognition applies to its reversal.

Independent adversarial review

The reviewer checked both full-language clauses, actual encoded lengths, group transitions, independent initialization, source/store directory routing, fixed capsule nesting, short-name termination, malformed-width behavior, per-lane service and final output. The complete manuscript was approved relative to the inherited native trie, strict FIFO, coherent-capsule and simultaneous-record primitives.

Review clarified that the dependency gate counts all distinct mentioned stores, including absent local aliases; unknown left stores reject. It also required all ordinary words to be nonempty, removing an ambiguity between the manuscript’s final suffix syntax and the executable parser. The routed clause was reviewed separately from the overlapping-support clause, so it does not silently rename an arbitrary overlapping instance.

Independent finite evidence

The standalone join audit imports the separately audited local producer from construction 75’s checker. The join oracle independently compares against direct literal relation composition. It does not reimplement or claim independent verification of every local producer primitive.

Scope Completed checks
Overlapping-support histories 60
Routed histories 3
Right records 4,800
Left records 5,262
Overlapping-support negative queries 249,600
Routed negative queries 11,016
Retained join snapshots 240
Cross-store code collisions 1,348
Long pending jobs preserved across shorter other-lane aliases 3,381
Unchanged-other-lane checks 9,647
Gate, grouping, width, unknown-store and namespace rejections 192
Malformed prefix checks 5
Maximum dependencies for one routed source 17

The routed source exceeds the first clause’s tested cap of three. Host maps, sets, codes and snapshots are finite semantic oracles, not native execution or timing measurements. The universal resource claim is the written schedule.

Failed routes, controls and remaining frontier

Cross-store code comparison is invalid because independent pools have different interpretations. A single pending slot cannot justify service across heterogeneous alias widths. Both failures have explicit alternatives in this construction.

A restricted opaque-bit argument explains the remaining fan-in issue: without a prepared aggregate or a route excluding components, an exact all-negative OR answer must inspect every independent component bit. This is a separated-reader contract bound, not a native-language lower bound. An input-built index or aggregate is an allowed escape.

Replay, affine-fragment joins, persistent causal sharing and coherent unary algebra retain their original hypotheses. Entry 129 retains its growing alphabet and conditional state/port bounds. Construction 76 uses stronger literal readout and charged routing; it does not transfer those bounds to unrestricted native storage. No new primary-proof review is counted; the last dedicated comparison remains assessment 220.

Next decisive test: Allow unbounded per-source dependence on stores with overlapping global targets, using an input-built route to relevant lanes or a paid aggregate. Begin with targets belonging to at most two stores and arbitrary ordinary names. Charge ownership discovery, source filtering, updates and final-query routing; no incidence directory is supplied. External-heap import, variable widths within a store and right-group reopening remain separate.

Integration and accounting

One maintained GitLab source owns the proof, failure qualifications, checker, checkpoint and reader synthesis. Styled HTML/KaTeX readings and highlighted source pages with exact raw downloads are generated projections. The existing homepage program entry records substantive progress without changing publication ranking or root admission. Build, scoped rendering and source/deployment parity receipts are in integration state. Browser viewport inspection is unavailable under the supported workflow and is not claimed.

Scheduled 127 to 128 and mathematical 138 to 139 count once. Window five after reassessment 134; native construction 76. Next 129 is research, primary review 130 and editorial 132. Installed legacy reverse-writer retirement remains separate and unverified.