Continuation Metadata
Purpose
Continuation metadata provides a deterministic, structural description of potential future parser paths.
Continuation metadata is descriptive only.
Continuation metadata does not imply resumability.
Ownership and lifecycle
The current ownership flow is:
Grammar
→ Preparation
→ Scheduler
→ Execution
→ Observation
→ Export
→ Analysis
Preparation owns continuation metadata production before scheduling starts.
The scheduler transports metadata and does not gain preparation or execution authority.
Descriptor model
Current continuation descriptors carry only structural metadata:
- origin alternative (
AlternativeIndex); - continuation category (
ParserContinuationCategory); - continuation position (
SequencePosition); - optional shallow expected token names.
They do not carry parser frames, callbacks, parse results, or execution delegates.
Conservative categories
Current categories are intentionally conservative:
Terminal;Sequential;SharedPrefixCandidate.
These categories are deterministic labels for metadata inspection and testing. They do not imply replay, resume, execute, or schedule semantics.
Distinctions
| Concept | Meaning |
|---|---|
| LookAhead | Observation |
| SharedPrefix | Structural grouping |
| Continuation | Structural future path |
| Alternative | Execution candidate |
Limitations
- Continuation metadata is non-authoritative.
- Continuation metadata does not grant runtime control.
- Continuation metadata can be discarded without changing parse acceptance authority.
- Under-detection is preferred over over-classification.