Add a Language
Adding a production language is primarily a GF/RGL + RuntimeSet release workflow, not a new SA core implementation.
Target flow
GF language development
↓
build + test
↓
PGF grammar artifact
↓
SA↔GF bridge artifact
↓
lexical/locale artifacts
↓
capability profile
↓
SA conformance suite
↓
immutable evidence
↓
RuntimeSet RELEASED
↓
activation
1. Develop the language in GF
Morphology, agreement, syntax and language-specific grammar belong in the GF language repository/RGL work.
GF Wordbench can be used for build, diagnostics and language-development workflows. SA does not patch grammar gaps at runtime. See GF-Ecosystem.
2. Build the PGF
Produce an immutable PGF artifact for the concrete language.
3. Implement the SA↔GF bridge
Map the SA operations required by the profile to the grammar's admitted GF functions/expressions.
The bridge must consume all semantic slots/features for a unit or fail.
4. Build lexical artifacts
Provide exact bindings for semantic concepts required by the profile/workload.
5. Define the capability profile
Declare the operation/block/register requirements applications may rely on.
6. Run conformance
The exact grammar + bridge + lexicon composition must pass the profile suite completely.
7. Release a RuntimeSet
Hash all artifacts, pin conformance evidence and mark the language/profile RELEASED only after the gate passes.
8. Activate
If more than one released RuntimeSet exists for the same language/profile, explicitly activate one or pin it in the request.
Scaling rule
The 300th language should require new validated artifacts and conformance evidence, not a new branch in SA core.