The origin.
EML-U is the complete theory EML started from. Meaning can attach wherever a symbol can carry it (above it, below it and at each corner), programs can be read in two dimensions, and one semantic core projects to many hosts. EML-P, the language you can run today, is its practical subset.
EML-P lives inside EML-U
Every EML-P program has an EML-U meaning. Going the other way, projecting EML-U down to EML-P follows one rule with three outcomes.
- ΠP · 1Fully expressiblebecomes EML-P
- ΠP · 2Not executable, but preservableEML-P plus metadata
- ΠP · 3Cannot be expressed safelymarked unsupported, explicitly
No silent loss.
Capabilities preserved from the original design
None of these is implemented yet. They are kept on record so that engineering EML-P never quietly narrows what EML means.
- 01
Semantic attachment at the top-right, and at the other corners
- 02
Semantic layers above and below a symbol
- 03
Two-dimensional syntax and a non-linear reading order
- 04
Semantic graphs and intent nodes
- 05
A host-neutral semantic IR
- 06
One meaning, many projections: Python, C++, JS, natural language, agent JSON
From preservation to an AI-native interface
- U0Theory preservation
- U1Semantic ontology
- U2Two-dimensional and multi-position syntax
- U3Across hosts
- U4AI-native interface
The full picture
The definition, the capability clusters, the projection rule and the source documents are on the Origins page.