Omnist Specification, v0.1

Abstract

Omnist is a data-modeling system built on one small formalism. A Document is an ordered list of labeled edges. A Schema is a graph of named, closed records constraining those edges. Because both models are small and closed, a set of operations over schemas — inclusion, equivalence, minimization, extraction, inference — are decidable and total, not best-effort.

graph LR
    JSON["JSON"] --> Doc["Document<br/>(ordered edge list)"]
    YAML["YAML"] --> Doc
    TOML["TOML"] --> Doc
    XML["XML"] --> Doc
    OML["OML"] --> Doc

    Doc -->|schema, stage 2| Val["validate / materialize"]
    Schema["Schema<br/>(root + closed records)"] --> Val
    Schema --> Algebra["Schema Algebra"]
    Algebra --> compat["compatible_with"]
    Algebra --> equiv["equivalent"]
    Algebra --> norm["normalize"]
    Algebra --> extr["extract"]
    Algebra --> infer["infer"]

This specification defines:

1. the Document model, and the invariants every implementation must preserve;
2. the Schema model, and what it deliberately cannot express;
3. OML, the text format for Documents;
4. OSD, the text format for Schemas;
5. the Schema Algebra: the operations over schemas and their exact semantics;
6. the ingestion pipeline that turns JSON, YAML, TOML, and XML into Documents;
7. a conformance protocol and error taxonomy.

The target is byte-identical conformance results across independent implementations. Where a rule is underspecified, that is a defect in this document, not license to choose.

Key principles

One model, many formats. JSON, YAML, TOML, XML, and OML all read into the same Document. Conversion is read-one, write-another. No format-specific model exists.

Repeated labels are the array. There is no array type in the Document model and no array type in the Schema model. {"tag": ["x","y"]} is the label tag occurring twice. This is what lets XML's interleaving survive the round trip that a map-of-arrays would destroy.

Order is data, never a constraint. A Document preserves edge order because it is a faithful record of its input. Schema validation ignores order entirely; only counts matter.

Closed by default, open only where written. Records name every label they allow. Scalar types are never composed. There are no maps, no wildcard keys, no unions, no enums. The single opening is the any type, which is visible in the schema text at a fixed label.

Decidability is the constraint that shapes the rest. Every refusal in this spec exists to keep the Schema Algebra total and exact. A feature that would force an operation to answer "maybe" is not added.

Relationship to prior work

The Schema Algebra descends from Lee and Cheung, XML Schema Computations (CIKM 2010). Omnist generalizes the paper in one direction and simplifies it in three. Chapter 6 states the deltas precisely; that is the right place to read them, since they only matter once the operations are on the table.

Keywords

The key words MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD, SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL in this specification are to be interpreted as described in RFC 2119 and RFC 8174, and only when they appear in all capitals.

Text in blockquotes, and any section marked Non-normative, is commentary. It explains reasoning and does not impose requirements.

Reading order

Chapter	Read it for
01 Glossary	Terms used everywhere else
02 Document model	The data structure
03 Schema model	The constraint structure
04 OML grammar	Writing Documents as text
05 OSD grammar	Writing Schemas as text
06 Schema Algebra	The operations
07 Codecs	Reading other formats
08 Conformance and errors	Passing the test suite
09 Divergence ledger	What implementations may differ on
10 Governance	How this document changes