The graph is the program.
Typed nodes and edges carry the meaning. A text listing is generated from the structure, never the other way round.
EML-NOVA is the family’s structure-first line. To an AI, a program is not a page of text but a typed graph with a stable identity. EML-NOVA builds, changes and verifies programs as that structure; text becomes one projection among several.
33 functions written as NOVA graphs, each one verified against an independent reference in exact arithmetic. This is NOVA’s practice-first corpus.
Typed nodes and edges carry the meaning. A text listing is generated from the structure, never the other way round.
A canonical form and a semantic hash identify a program by what it is, not by how it happens to be written down.
Edits are proposed as patches to the graph and applied as transactions, so every change is explicit and reviewable.
When an AI asks to build or change a program, the request is checked (types, shapes, tests) before anything is committed.
The sum of squares from the family page, drawn as the kind of structure EML-NOVA works with: typed nodes, typed edges, and an identity computed from the structure itself.
EML-U keeps the meaning.
EML-P runs it today.
EML-NOVA is the structure AIs build with.
This page describes the idea; the standard library shows it running, with every result verified. Releases will be announced here.
Back to the family