nx_asset_record.nx

nx_asset_record.nx -- UNIVERSAL ORGANIZATION TOOLING arc, R0. THE unified, content-addressed ASSET RECORD: every artifact (a generated image, a downloaded video, a doc, a dataset, an audio clip, a product) BECOMES one of these -- one model, all kinds (org_research.tsv: unified-base-type, schema.org CreativeWork is the base, subtypes specialize). Keyed by its CID (content-addressed-id): same bytes -> same id everywhere; any byte change -> a different id (the inventory's self-proof of integrity). THE BACKBONE NEEDS ZERO NEW STORAGE/IDENTITY/CRYPTO -- this organ is pure COMPOSITION: * identity + canonical serialize -> nx_canon_cid (canon_encode / cid_of) * tolerant decode -> nx_uxf_decode (canon_decode) [house-of-cards safe] * self-describing profiled CID -> nx_uxf_cid (uxf_cid_profiled, codec=UXF_DATA) * durable segmented KV store -> nx_seg_store (ss_begin/ss_add/ss_build_keys/ss_commit/ ss_open/ss_hget), key = the record's CID A record is a SET OF (key,value) string fields fed to canon_encode -> exactly one canonical byte string -> exactly one CID (insertion order is irrelevant by construction; canon_encode key-sorts). Field ACCESS is by key name over a decoded (keys[],vals[]) pair -- never struct-field aliasing (heap-mmap + buffer/offset idiom only, per the build rules). RECORD MODEL (cited to org_research.tsv CONFIRMED claims): type image|video|doc|dataset|audio|product (unified-base-type) -- Dublin Core core spine (dublin-core-core): ----------------------------------------- title dc:title creator dc:creator date dc:date (created/issued, ISO-8601 string) identifier dc:identifier (external/native id; CID is the PRIMARY id, this is a sidecar) subject dc:subject (free-text subject line; tags[] is the keyword facet) rights dc:rights -- W3C PROV provenance class (provenance-class-three; entity<-activity<-agent shape): ----- prov_class human | machine | downloaded prov_agent person name | software name | "external" prov_tool (machine) the generating tool/app prov_model (machine) the model id prov_source (downloaded) the source URL prov_date (downloaded) the fetch date -- type-specific MEDIA block (schema-mediaobject; only meaningful for image/video/audio): -- width / height / duration / encoding_format -- organization facets: --------------------------------------------------------------- tags comma-joined keyword facet (faceted-findability) location where the bytes live (path / NAS volume / URL)