nav plane + concept re-keying: the transcode-doctrine Phase 0/1/3/5 machinery#72
Merged
Merged
Conversation
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: serverGenReqId_38cb950a-00aa-42ac-bad7-87882acad52f) |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f0adeb8b35
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jul 10, 2026
…on surface precedence - Program.cs: screen classification now compares the LAST dotted segment of a base type (LastDotSegment), so `System.Windows.Forms.Form` / `global::...UserControl` bases and namespace-qualified screen creations match the bare screenBases/corpus registry instead of slipping past the fixpoint and silently dropping the screen's nav triples. - surface_schema.rs: when a name carries multiple surface tokens (`add_option_widget_template`), the LAST match in name order wins — the trailing table-kind token names the actual storage, so get/add_option_*_template classify as TemplateSource, never EnumSource. All matched surface tokens leave the residue (concept material only). 131 crate tests green; furnace exam re-run unchanged (6 concepts, 2154 residuals).
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: serverGenReqId_16a93d43-e49c-4635-960b-96842b2cd38a) |
…cates + Roslyn nav arm) The "ERB/RB split": before a concept re-keying pass can untangle a God-object DAL, the pipeline needs the labyrinth map — which screens (rooms) exist, how they navigate, which concept each surfaces. This lands the language-agnostic NavGraph plane + its first frontend arm (C#/WinForms). ruff_spo_triplet (closed vocab 64 -> 68, council-reviewed shape): - navigates_to (screen -> screen | nav-method IRI) Inferred - surfaces_concept (screen -> concept, config-declared) Authoritative - handles_event (control -> "<Event>:<handler-IRI>") Authoritative - contains_control (parent -> control) Authoritative surfaces_concept's tier is deliberate: the binding is a corpus-owner config-table row (data-as-config) — the config IS the claim. ruff_csharp_spo harvester (opt-in --nav, zero change without the flag): - screen classification to fixpoint (UserControl/Form bases transitively + --room-prefixes uc_,mod_,con_), two-pass; - EmitNavArm walks each screen type: Designer `+=` wiring -> handles_event, `Controls.Add` -> contains_control, screen-type object creation + nav-pattern dispatch (--nav-method-prefixes/-suffixes, default Show_*/*_Show/*_Hide) -> navigates_to; - --room-aliases dir=concept rows -> surfaces_concept. All convention knobs are config, no corpus-specific tokens in code. - examples/validate_load.rs: generic ndjson-through-load validator. Verified on a real WinForms corpus (279 files / 167K LOC): 103416 triples, 6239 nav-plane rows (4496 contains_control, 1403 handles_event, 189 surfaces_concept across 93 screens, 151 navigates_to), all accepted by ruff_csharp_spo::load. Tests: 98/98 + 4/4 (new loads_and_validates_nav_arm_ ndjson pin); count-lock test renamed predicate_count_locked_at_68.
…de-doctrine machinery
Three modules that turn a God-object harvest into re-keyed concepts, with
every unresolved method landing on a slag ledger instead of silently
dropping (doctrine: "mess -> config DTO -> proposer placement -> residual
DTO -> next pass"):
- concept_split.rs (Phase 1): fixes the CONCEPT = CLASS give-up point.
tokenize_method_name ('_' + case boundaries, concept_key discipline),
ConceptConvention (verbs / scopes / concept_aliases — pure data-as-config,
zero domain tokens shipped), split_method_name, rekey_model over
functions+helpers, ResidualMethod{NoVerbPrefix,EmptyResidue} slag rows.
Boundary-4: SplitName::concept is concept_key-fixed-point on BOTH branches
(alias values are folded through concept_key; tested).
- surface_schema.rs (Phase 3): classify_surface — getter-over-lookup-table
shapes (combo/option/template) are SCHEMA, not behavior; classified before
action lifting so enum plumbing stops polluting the capability surface.
SubtabSurface/GridSurface reserved for the coming convention rows.
- structured_names.rs (Phase 5): NameGrammar-driven parse of positional
name-encoded trees (f_2_3 -> form 2 / section 3) + part_of_edges for the
FacetCascade forest.
- examples/rekey_exam.rs: the furnace exam — corpus-agnostic, fully
config-driven (verb/scope/alias/codebook/expect rows at runtime; no corpus
tokens or data committed). Exercised against a real 103k-triple WinForms
God-object harvest: all expected concepts re-derived nonzero, 594 methods
keyed, 2154 residuals ledgered, zero silent drops.
- examples/validate_load.rs: generic ndjson-through-load validator.
131 crate tests green (33 new). Reviewed: P0 doc-identifier scrub, P1
alias-branch fixed-point fold + test, P2 doc/reserved notes applied.
…on surface precedence - Program.cs: screen classification now compares the LAST dotted segment of a base type (LastDotSegment), so `System.Windows.Forms.Form` / `global::...UserControl` bases and namespace-qualified screen creations match the bare screenBases/corpus registry instead of slipping past the fixpoint and silently dropping the screen's nav triples. - surface_schema.rs: when a name carries multiple surface tokens (`add_option_widget_template`), the LAST match in name order wins — the trailing table-kind token names the actual storage, so get/add_option_*_template classify as TemplateSource, never EnumSource. All matched surface tokens leave the residue (concept material only). 131 crate tests green; furnace exam re-run unchanged (6 concepts, 2154 residuals).
AdaWorldAPI
force-pushed
the
claude/nav-plane-labyrinth-recon
branch
from
d6684bd to
68c976a
Compare
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: serverGenReqId_ec00975c-d322-42b3-85b5-39121b6a2e14) |
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jul 10, 2026
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jul 10, 2026
…on (gap b)
Closes the routes.rb stratum (E-CLICKWEG-CHOREOGRAPHY-1 gap b): an
AST-walk (lib-ruby-parser) over config/routes.rb + modules/*/config/
routes.rb that resolves a route-helper stem (+ verb) to its
controller#action, making the InvokesAction/NavigatesTo join resolvable
for the first time. Two additive predicates RoutesTo (stem ->
verb:controller#action) + RouteScope (member/collection/canonical/
standalone), both Authoritative; predicate count-lock 71 -> 73.
Faithful Rails semantics per the council-hardened v3 spec
(openproject-nexgen-rs .claude/ruff-expansions/2026-07-10-routes-arm-spec.md):
canonical-seven with collection_token/member_token (news -> news_index
when singular==plural), singleton-collection-singular, absolute
/-controller strip, controller:-kwarg ranking, module:-scope-wrap,
naked-verb-in-resources nesting, string-path auto-naming, as:"",
decl-order first-wins + duplicate_stem_conflicts, multi-name replay,
escaped_dynamic for non-literal args, only/except coercion, shallow
member-drop. 19 unit fixtures + an env-gated (RAILS_CORPUS_SRC) drift
fuse pinned to the measured corpus: 29 files, 1625 declared, 1534
emitted, escaped {mounts=9 via_all=6 dynamic=13}, escaped_other exactly
["use_doorkeeper"].
Central-review fixes over the first implementation (gated in the shared
target, verified on the real corpus):
- namespace-path controller fallback: a bare verb with explicit action
in a namespace (get "plugin/:id", action: :show_plugin in
namespace :admin;:settings) resolves to admin/settings#show_plugin
(the real Admin::SettingsController) instead of leaking the verb name
into escaped_other. escaped_other is now exactly the Doorkeeper macro.
- escaped_dynamic pinned to the measured 13 (not the grep-estimated 2):
the walker counts every conservatively-declined declaration, all
counted, none silently dropped.
- as: value used verbatim for the collection name (Rails guide:
resources :foos, as: :bar -> bar, not bars) — test corrected.
- 3-level nesting spot-check uses new_project_meeting_agenda_item_outcome
(the except: %i[index show] filter removes the show helper) — the
walker correctly respects except:.
Edit allowlist honored: routes.rs (new) + lib.rs (register) + schema.rs
(pub(crate) singularize) + triple.rs (2 variants). Base ruff main #72 +
W3. Pre-existing #72-file fmt/pedantic warnings left untouched (out of
scope).
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
2 participants
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.
Summary
Three commits landing the language-general transcode-doctrine machinery ("mess → config DTO → proposer placement → residual DTO → next pass"), rebased onto main post-#69/#70/#71 and converged with the parallel UI-navigation plane that landed there (
navigates_to/selects_view/invokes_action/renders_as).Commit 1 — UI-config plane (Phase 0 labyrinth recon, the room map). Main's nav plane carries the Klickweg EDGES; this adds the ROOM MAP a concept re-keying pass navigates by:
ruff_spo_triplet: closed vocab 68 → 71 —surfaces_concept(corpus-owner room→concept config rows, Authoritative: the config IS the claim) /handles_event(Designer+=wiring) /contains_control(control-tree containment).NavigatesTounified with main's (its "pure screen→screen" discipline kept — my method-pattern variant dropped). Count-lock test renamedpredicate_count_locked_at_71.ruff_csharp_spoharvester, converged with main's arm: main'sEmitNavArm(Show/ShowDialog, field-instantiation SPA idiom,selects_viewribbon, CommonDialog exclusion) kept wholesale; grafted on top: screen classification extended to fixpoint through corpus base chains + optional--room-prefixes,--room-aliasesdir→concept binding, andEmitUiConfigArm(event wiring + containment, type-level walk so constructors/Designer partials are covered).Commit 2 — the re-keying machinery (Phases 1/3/5).
concept_split.rs: fixes the CONCEPT = CLASS give-up point (entries_from_actionsresolving a God-object class to classid 0 for every row). Tokenizer (underscore + case boundaries,concept_keydiscipline),ConceptConvention(verbs/scopes/aliases — pure data-as-config, zero domain tokens shipped),split_method_name,rekey_model,ResidualMethodslag rows. Boundary-4:SplitName::conceptisconcept_key-fixed-point on both branches (alias values folded; tested).surface_schema.rs: getter-over-lookup-table shapes (combo/option/template) classified as SCHEMA, not behavior, before action lifting.structured_names.rs:NameGrammar-driven parsing of positional name-encoded trees (f_2_3→ form 2 / section 3) +part_of_edges.examples/rekey_exam.rs: the corpus-agnostic furnace exam (all convention/codebook/expectation rows arrive as runtime config; no corpus tokens or data committed).Commit 3 — codex P2 fixes (template-over-option surface precedence: last surface token in name order wins; the qualified-base fix folded into the converged prepass, which uses main's
LastSegment).Exercised end-to-end against a real 103.5k-triple WinForms God-object corpus on the CONVERGED tree: 173
navigates_to+ 22selects_view(main's arm) + 4 551contains_control+ 1 452handles_event+ 191surfaces_concept(this PR's config plane); all triplesload()-validated; furnace exam green — all expected domain concepts re-derived with nonzero classids, 594 methods keyed, 2 154 residuals ledgered, zero silent drops.Test Plan
cargo test -p ruff_spo_triplet -p ruff_csharp_spo— 133 + 5 passing on the rebased tree (new: tokenizer/split/rekey/residual reasons, alias fixed-point, name-grammar tiers +part_ofedges, surface classification incl. last-match precedence, UI-config-arm loader pin, count-lock 71).concept_keyfold + fixed-point test, both codex P2 threads addressed.