UTF-4088 Code-Point and Serialization Contract
Status: experimental implementation contract, version 1.
1. Scope and representation
The current implementation stores both InputState and CodePoint in unsigned 64-bit integers. It does not implement a 4088-bit integer or a byte-stream encoding of that width. The accepted experimental identifier interval is:
- Minimum:
0x110000(inclusive). - Maximum:
0x1FFFFFFFF(inclusive). - Width required for this interval: 33 bits.
The interval deliberately starts above Unicode's maximum code point. These values are UTF-4088 project identifiers, not Unicode scalar values. APIs must not label them Unicode characters or silently pass them to Unicode encoders.
2. Canonical registry record
A registry record contains integer_id, stage, language, codepoint, shape_id, and meaning_id. Registry generation is deterministic under UTF4088-RG-1. The manifest is the source of truth for the record count, enumeration order, and hash seeds.
3. Canonical glyph representation
A glyph is 12 rows of 8 bits each. Row order is top-to-bottom; within each row, bit x represents horizontal pixel x, where x=0 is the least significant bit. Unused bits are not permitted because each row is exactly 8 bits. This stores exactly 96 bitmap bits.
Seed-derived glyphs are reproducible experimental placeholders. They are not automatically valid writing, language-specific characters, or historically attested symbols.
4. Determinism and validation
Implementations must:
- Use the fixed integer widths and limits documented above.
- Reject experimental identifiers outside the defined interval at the public code-point boundary.
- Avoid undefined behavior when floating-point inputs are NaN or infinite.
- Clamp values before quantization and mask each packed field before shifting.
- Preserve directed edge identity as the ordered
(from, to)pair; a hash value is an index aid, never the edge's identity. - Keep generated IDs stable within a manifest version.
- Record source and licensing/provenance details before labeling a glyph or meaning as historically sourced.
5. Versioning
Any change to identifier bounds, field widths, bit order, registry count, hash seeds, or record interpretation requires a new format/manifest version. This experimental contract makes no claim of Unicode compatibility or standards approval.
6. Input-map traversal metadata
Input-map coordinates and pressure iterations are separate from registry identifiers and glyph serialization. Follow INPUT-MAP-AND-PRESSURE.md for top-left, left-to-right, top-to-bottom, then next-z traversal and pressure levels 1–50. The National Lien subject order is specified separately in NATIONAL-LIEN-REGISTER.md. Do not encode coordinates, alphabetic ranks, or pressure levels into an identifier unless a future version explicitly defines a versioned serialization format.