Repository navigation
probe(dr-math): D-DRCA-3 DR #4 method routing falsifier - #1849
Merged
Merged
Conversation
…GIT reference predicates Records seven user decisions from 2026-10-28 in DECISION / SCOPE / BASIS / REVISIT WHEN form: - the three DT64 levels; - DR#3 Identity × Properties as the primary view; - WHEN×ON×DO per DR#3 cell; - DR64 = direct routing and DT64 = DTO bind store; - immutable classid → behaviour via hotplug; - global behaviour needs no classid; - OGIT read unpinned. Maps each decision onto existing code and plans with evidence states, and stages D-DT64-6-0..5: - OGIT reader reconciliation; - ogit:allowed expansion in lance-graph-ontology; - DT64 predicate registration; - consumer HotPlug access. Requests, but does not apply, an amendment to D-DT64-Q0 line 9. Docs only. No code, ABI, test, runtime or merge authorization. Board: entry, INTEGRATION_PLANS and STATUS_BOARD prepends, and the regenerated entries index. The supersession index regenerates unchanged. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
…number The citation-decay gate failed on nine grounding rows. The cited lines were correct, but each row's citation sat next to the following row's backticked text, which the gate took as its anchor. Line numbers into moving files are what the gate exists to retire, so every in-repo citation now names a symbol or section heading instead. Also replaces the two grounding rows that rested on grep output with claims from full reads: the three merged DT64 plans, and the qr_focus / qr_register lab tests (whose actual wording is now quoted correctly). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
…copy amortizes prefetch DR64 direct routing references existing identities and mints none. The global->local prefetch holds a copy of the register (identity references and local positions only, never values), reused by every read under one frame and rebuilt when the frame check refuses. Stated against the zero-copy law's pointer-table carve-out and Q0's accounted prefetch cost. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
…lance-graph-dqfjlx
…atch fold #1814 and #1819 have merged, so the plan now cites them by section from this repo instead of as open PRs by line number. That line-number row was the one citation-decay failure on c440fab: the file did not exist locally, so the local run reported UNVERIFIABLE where CI reported DECAYED. Reading both in full showed two contradictions, now fixed: - the dispatch step's two-valued `OR over w in G` fold is restricted to candidate DO presence; selection evaluates typed guards in K3 (#1814 kill test I03); - open question 3 now follows #1819: several mappings and convergence are normal; a conflict needs incompatible source-qualified results. Also corrects the quoted Ruff role names to #1814's list. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
The plan and board entry said the merged plans do not use ON / WHEN / DO. That holds only for the three DT64 plans. ON×WHEN×DO is established in merged lance-graph research docs: the D-CFA series (docs/research/counterfactual-behavior/) and the EAW series (docs/research/epistemic-awareness/), plus #1819. All twelve files read in full; the grounding row and the non-goal line now say so. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
… CR-H0/H1) #1760's D-SAG-6 plan is unchanged since merge; its successor is the crai-rs CR-H0 / CR-H1 map (crai-rs #1, #2), which makes D-SAG-6 the motor and Ruff + DT64 the transmission. The plan now grounds both, places the per-cell ON×WHEN×DO registry as that transmission's behaviour binding, ties D5/D8 to D-SAG-6 F06, F07 and F11, and requires behaviour registrations to reference ordered effects: a DO set or deduplicated SPO bag cannot distinguish two effect orders. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
… registry Searched odoo-rs, OGAR, crai-rs, lance-graph and ndarray for the #1760 lineage. Besides crai-rs CR-H0/H1, odoo-rs #46 (Odoo Codegraph harvest schema, docs) is the one successor; OGAR and ndarray have none. Read #46's doc and schema in full. Its `dt64_candidates` array (source nodes -> existing algebra_ref, no seat mint) is the registry's Odoo intake; every emitted_by MethodId is kept; effective behaviour needs a registry_snapshot in the frame. Open question 2 (inheritance before the cell lookup) is narrowed by #46's _inherit / _inherits rules. Notes that #46's doc lists dt64_candidates as a graph layer while its schema does not. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
…ract) - The two OGIT readers see disjoint verb sets: ttl_parse `classify` accepts the 12 rdfs:Class verbs (all WorkOrder, all ogit:from-to), OGAR `parse_verb` the 511 owl:ObjectProperty verbs, of 523. Measured over an unpinned OGIT clone (b5731a8, receipt only). - Add D-DT64-6-1a: classify every verb file in both readers first. - D-DT64-6-2 builds (src, verb, dst) from the entity ogit:allowed blocks (960 files, none in verbs/) plus the 12 from-to verbs, reads OGAR's OgitTables::Allowed instead of adding a second parser, and puts the standing parity test beside tests/ontology_cache.rs. - 518 / 2,756 is OGAR #363's reader output (eaw-5), not OGIT drift. - crai-rs CR-H0/H1 is a sidecar consumer of #1760, not its successor; its open-PR list is dated 2026-10-27. - Ruff has no Rust frontend; role names checked in ruff role.rs. - Cite D-DT64-Q0 by section, not line 9; INTEGRATION_PLANS says eight decisions. Docs only. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
Add the D-DT64-6-1a row and restate D-DT64-6-2 as reading OGAR OgitTables::Allowed rather than a second expansion in lance-graph-ontology, matching plan §5 after the #1831 review. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMPWTwovjYJEUMuGPb4
…= main) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
…-dqfjlx' into claude/clone-ndarray-lance-graph-dqfjlx
…h-dqfjlx Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
Two methods on one cell with equal read/write sets and a different effect order stay two routed entries; so do two methods whose effects commute. A declared guard without its witness refuses. DT-owned values (method key, guard witness, effect-order reference) are carried opaque; the test's DT stub and the ported ndarray #395 order law are the oracle. Lab only: no DT resolver (D-CMH-1 owns it), no MethodKey mint, no contract change. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
STATUS_BOARD: D-DRCA-1 and D-DRCA-2 marked Shipped (#1834, #1836); D-DRCA-3 In PR. Entries index regenerated (503 rows). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMNiyjwqFke9CfmoBgn
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars.
 |
AdaWorldAPI
marked this pull request as ready for review
October 30, 2026 02:06
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
The first DR #4 falsifier from D-DRDT §5 (#1835), as a lab module in
dr-math-binding-probe:src/dr4.rs:route(store, register, resolution, op, observed_guards)over one admittedBoundDr64.tests/dr4_method_routing.rs: 12 tests..claude/plans/d-drca-3-dr4-method-routing-v1.md.Contract
DR carries DT-owned values opaque.
MethodKey,GuardWitnessandEffectOrderRefare 16-byte references, compared for equality and never interpreted.Kept: one output entry per input method, in input order. Key, signature revision, read/write field sets, guard witness and effect-order reference pass through unchanged, and the result carries the register's frame. Nothing is merged, including methods whose effects commute; merging is the law owner's decision.
Refused, in order:
Frame(..): stale register frame.ResolvedUnderOtherFrame: the methods were resolved under another frame issue, including another store at equal revision numbers.ForeignClass: the resolution or one of its methods belongs to another class.DuplicateMethod: the same key twice. Equal read/write sets are not a duplicate.OutOfRegister: coordinates outside 64 × 64.UnsupportedAlgebra: a numeric fold.NotBoolean: a presence read over a three-valued register.GuardDropped/GuardMismatch: a declared guard with no witness, or with another one. DR checks witnesses; it does not evaluate guards.Oracle
The test's DT stub holds each method's real effects and guard. The effect-order law is ndarray #395's
Effect/run_effects/commutes/check_reorder(examples/pof_antennas/order_law.rs), ported verbatim.[Set(1), Add(1)]and[Add(1), Set(1)]give 2 and 1, and the swap is refused. So the two references DR keeps apart are provably different methods.[Add(2), Add(3)]and its reverse give equal results and the swap is legal. Routing still keeps both.Evidence
cargo test --release --manifest-path crates/dr-math-binding-probe/Cargo.toml: all suites green, including the 12 new tests.cargo fmt --checkandcargo clippy --all-targets -- -D warnings(the crate's CI steps) are clean.7 disable runs, each restored from the commit, each red on its named test:
Scope and ownership
PreparedClass(class-method-schema-hotplug-v1.md).Resolutionis a lab stand-in for DT's answer.MethodKeymint. The key is DT's to mint (D-CMH §2.1, decision ledger item 2); the test mints its own, named as such.Board: new entry
entries/2026-10-30-d-drca-3-dr4-method-routing.md; STATUS_BOARD D-DRCA-3 moved to In PR, and D-DRCA-1 / D-DRCA-2 marked Shipped (#1834, #1836); INTEGRATION_PLANS gains a row; entries index regenerated (503 rows).Generated by Claude Code