Repository navigation
ogar-vocab: mint typed_field (0x080A), ship W4 document ActionDefs - #282
Conversation
5+3 council-ratified (spec v1 -> v2 -> v3, per .claude/agents/5plus3-council.md) implementation of OGAR-DOC-W4-BUILD-SPEC.md SS W4-3/W4-4: - Mint class_ids::TYPED_FIELD = 0x080A (key/value/confidence/geometry per-field decomposition), confirm class_ids::DOCUMENT = 0x080B unchanged. Both wired into the private CODEBOOK table, class_ids::ALL, all_promoted_classes(), and the ConceptDomain::Ocr domain-count fuse (93->94 classids, 10->11 OCR-domain concepts). - New crates/ogar-vocab/src/document_actions.rs: persist_document / read_document / reconstruct_document ActionDefs (facts only, External kausal), object_class = document (0x080B). Lands in a SEPARATE domain table from ocr_actions.rs (Deviation D-1 from W4-4's literal text, council-verified via the cross-domain-synthesizer savant): resolve_hotplug gates per contributing table, so folding these rows into ocr_actions.rs would have forced paperless-kv into OCR_EXPECTED_EXECUTORS and permanently entangled tesseract-ogar's and paperless-kv's hot-plug resolutions. Expected executor: paperless-kv. - capability_registry::domain_tables() gains a 4th entry (document_entries, same entries_from_actions() derivation as ocr/geo/healthcare). New hot-plug test proves paperless-kv resolves the document table with zero contamination of tesseract-ogar's existing OCR resolution. - Doc staleness corrected append-only across DISCOVERY-MAP.md, OGAR-DOC-LAYER-PROPOSAL.md, DOCIR-COMPOSITION-GROUNDING.md, OCR-ACTIONS-V2-PROPOSAL.md: every "typed_field mint deferred / council-pending" claim regraded in place with a dated Status paragraph, none deleted. Deliberately NOT in this landing: the persist_document executor body (walking a DocIr into GUID-keyed SoA nodes) -- facts only per W4-4's own rule. Cross-repo companions tracked separately: lance-graph-contract's ogar_codebook mirror needs the matching typed_field row; paperless-rs's paperless-kv needs its HOT_PLUG activation against this table. 164 unit tests + 6 doctests green, clippy -D warnings clean, fmt clean (cargo fmt -p ogar-vocab / cargo clippy -p ogar-vocab, scoped). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ArVbbq3DsToBM7r79zGeEf
…the spider convergence plan Follow-up to 3fbd9f3 (the W4 mint + document_actions landing). The mandatory closing exhaustiveness grep sweep for 0x080A/typed_field across OGAR/lance-graph/paperless-rs turned up two more sites in this repo, both regraded append-only: - ogar-doc-ir/src/lib.rs's "This crate mints nothing" section named the ActionDefs' home as "ogar-vocab / ogar-doc (W4)" -- the council landed them in ogar-vocab's own document_actions module, NOT a separate ogar-doc crate (W4-5 SS A1's "executor lives in the assembly repo" ruling). Corrected; crate's own 16 tests still green. - DOC-IR-SPIDER-CONVERGENCE-PLAN.md's W0 council gate gains a Status paragraph: the mints half is closed, W2/W3 (the producers actually persisting through persist_document) remain open. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ArVbbq3DsToBM7r79zGeEf
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
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_e8c1f50e-d67d-4b16-be27-da502129892e) |
![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: 894a901d59
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".
| attr("key", "String"), | ||
| attr("value", "String"), | ||
| attr("confidence", "u8"), | ||
| attr("geometry", "Geometry"), |
There was a problem hiding this comment.
When a ogar_doc_ir::TypedField is projected into this newly minted class, its location cannot round-trip: the IR defines bbox: BBoxRail and has no per-field geometry member (geometry belongs to DocIr). Consequently, ClassView/persistence code driven by this canonical schema will omit the field's bbox rails and request a nonexistent attribute, losing the position required to reconstruct or render harvested fields. Replace this attribute with the matching bbox/BBoxRail shape.
Useful? React with 👍 / 👎.
Summary
5+3 council-ratified (spec v1 → v2 → v3, per
.claude/agents/5plus3-council.md) implementation ofOGAR-DOC-W4-BUILD-SPEC.md§W4-3/§W4-4 — the doc-layer capstone this session's council was convened for.class_ids::TYPED_FIELD = 0x080A(key/value/confidence/geometry per-field decomposition), confirmsclass_ids::DOCUMENT = 0x080Bunchanged. Both wired into the privateCODEBOOKtable,class_ids::ALL,all_promoted_classes(), and theConceptDomain::Ocrdomain-count fuse (93→94 classids, 10→11 OCR-domain concepts).crates/ogar-vocab/src/document_actions.rs:persist_document/read_document/reconstruct_documentActionDefs (facts only,Externalkausal),object_class = document (0x080B). Lands in a separate domain table fromocr_actions.rs(Deviation D-1 from §W4-4's literal text, council-verified via the cross-domain-synthesizer savant):resolve_hotpluggates per contributing table, so folding these rows intoocr_actions.rswould have forcedpaperless-kvintoOCR_EXPECTED_EXECUTORSand permanently entangledtesseract-ogar's andpaperless-kv's hot-plug resolutions. Expected executor:paperless-kv.capability_registry::domain_tables()gains a 4th entry (document_entries, sameentries_from_actions()derivation as ocr/geo/healthcare). New hot-plug test provespaperless-kvresolves the document table with zero contamination oftesseract-ogar's existing OCR resolution.DISCOVERY-MAP.md,OGAR-DOC-LAYER-PROPOSAL.md,DOCIR-COMPOSITION-GROUNDING.md,OCR-ACTIONS-V2-PROPOSAL.md,DOC-IR-SPIDER-CONVERGENCE-PLAN.md, andogar-doc-ir/src/lib.rs— every "typed_field mint deferred / council-pending" claim regraded in place with a dated Status paragraph, none deleted.Deliberately not in this PR: the
persist_documentexecutor body (walking aDocIrinto GUID-keyed SoA nodes) — facts only, per §W4-4's own rule. Cross-repo companions (both merged as part of this same council landing):lance-graph-contract'sogar_codebookmirror row,paperless-rs'spaperless-kv::HOT_PLUGactivation.Test plan
cargo test -p ogar-vocab— 164 unit tests + 6 doctests greencargo clippy -p ogar-vocab --all-targets -- -D warningscleancargo fmt -p ogar-vocab -- --checkcleandocument_hotplug_resolves_vocab_and_actions_and_does_not_contaminate_ocrtest re-verifiestesseract-ogar's OCR resolution is byte-identical before/aftercargo test -p ogar-doc-ir— 16/16 green after the doc-comment correctionGenerated by Claude Code