Semantic Formats

Semantic source and transport currently use several related representations.

`.se`

.se is the readable Semantic language used by the selfhosting/compiler experiments in this repository. It is the format currently fed into the native MSPLC object pipeline.

For the selfhosting work, .se files are treated as compiler source modules rather than as opaque containers for a prebuilt executable.

`.sp`

.sp is a readable, versioned transport representation of the existing SemanticProgram / Universal AST.

Canonical envelope:

sp 1
program {
    schema = 1
    source_language = "go"
    field.evaluation = "eager_left_to_right"
    field.universal_ast = { ... }
}

The important property is that .sp does not introduce a second IR. It serializes the canonical semantic program.

`.spz`

.spz is the compressed Semantic transport form. It is intended to decode back into the same canonical .sp semantics.

flowchart LR U[SemanticProgram / UAST] U <--> SE[.se] U <--> SP[.sp] SP <--> SPZ[.spz] U <--> J[JSON debug/interchange]

JSON

JSON remains useful for inspection, interchange and validation. It exposes the structured program graph, node fields, semantic evidence and relation matrices.

Format rule

The representation may change, but program meaning must not:

Decode(Encode(P)) ≡ P

Unknown extensions or semantics must be preserved deliberately or rejected explicitly.