feat(spo-3d): Three-Axis Content-Addressable Graph with Causal Chain Discovery#130
feat(spo-3d): Three-Axis Content-Addressable Graph with Causal Chain Discovery#130AdaWorldAPI merged 10 commits intomainfrom
Conversation
![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: 02923750c3
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".
| extra_flags: u16, | ||
| ) -> Result<CogRecord, SpoError> { | ||
| // Pack sparse axes into content container | ||
| let (content, mut desc) = pack_axes(x, y, z)?; |
There was a problem hiding this comment.
assemble_record packs whatever SparseContainer::from_dense produced, but the current builders feed it dense fingerprints (label_fp and Container::random-derived values), so each axis typically has ~128 non-zero words and violates the 6 + Nx + Ny + Nz <= 128 packing limit. In practice this makes build_node, build_edge, and build_meta_awareness return AxesOverflow for normal inputs instead of producing records.
Useful? React with 👍 / 👎.
| use super::sparse::{pack_axes, unpack_axes, AxisDescriptors, SparseContainer, SpoError}; | ||
|
|
||
| /// SPO geometry ID (matches ContainerGeometry::Spo = 6). | ||
| const GEOMETRY_SPO: u8 = 6; |
There was a problem hiding this comment.
The builder hardcodes SPO geometry as 6, but ContainerGeometry::from_u8 currently only recognizes 0..=5 and unknown values are treated as default Cam by MetaView::geometry(). Records produced here will therefore be misclassified by shared metadata readers, so downstream code cannot reliably detect or handle SPO records.
Useful? React with 👍 / 👎.
| let bitmap = [content.words[offset], content.words[offset + 1]]; | ||
| let words: Vec<u64> = content.words[offset + 2..offset + 2 + count].to_vec(); |
There was a problem hiding this comment.
unpack_one_axis directly indexes content.words[offset] and slices offset + 2..offset + 2 + count without checking bounds. If metadata descriptors are corrupted (or a non-SPO record is scanned), queries will panic instead of returning a SpoError, which can crash the store during read paths.
Useful? React with 👍 / 👎.
1 participant
SPO 3D: Three-Axis Content-Addressable Graph
What
Adds
ContainerGeometry::Spo— three sparse axes (Subject/Predicate/Object) packed into one 2KB CogRecord. Forward, reverse, and relation queries become native axis scans with no join or edge table.Architecture
Files
Docs (
docs/spo_3d/):INTEGRATION_PLAN.md— Master plan, phases, decisionsCONTRACTS.md— Trait definitions, invariants, compile-time guaranteesSCHEMA.md— Byte-level layout of every fieldNATURES_CAM.md— Biological foundations (DNA, immune system, Piaget)Rust (
src/graph/spo/):sparse.rs— SparseContainer + PackedAxes (506 lines)scent.rs— NibbleScent 48-byte histogram (204 lines)store.rs— SpoStore three-axis queries + causal chains (434 lines)builder.rs— Node/edge/meta-awareness construction (340 lines)tests.rs— 6 ironclad tests (348 lines)mod.rs— Module root + re-exportsWhat does NOT change
6 Ironclad Tests