Use this issue to track InQL RFC 029, proposed in PR #60 at docs/rfcs/029_metadata_attachments.md.
Area
- Specification (RFCs)
- Documentation
Summary
RFC 029 defines typed metadata attachments for InQL semantic targets. Attachments let authors, library code, sessions, adapters, and inspection tooling associate structured metadata with plans, fields, expressions, read roots, quality checks, policy checkpoints, and execution evidence without collapsing those facts into comments or untyped maps.
Motivation
Relational evidence needs a common attachment model. Without one, lineage, policy, data quality, adapter coverage, and export metadata will each invent their own ad hoc key/value structure, making evidence hard to validate, compare, serialize, or consume.
Proposal sketch
Define typed metadata attachment records that bind a target, attachment type, source, lifecycle, and structured payload. The RFC should distinguish authored metadata from derived or observed metadata and require versioned, inspectable schemas for serialized evidence.
Alternatives considered
The draft rejects untyped string maps, embedding all metadata directly into every plan node, and letting downstream systems attach arbitrary facts without a shared InQL target model.
Impact / compatibility
This is additive. Existing plans can remain metadata-free. Tools that require metadata must distinguish absent metadata from an empty metadata set.
Implementation notes (optional)
Review PR #60 for the draft. This work depends on RFC 028 semantic targets and feeds the quality, governance, bundle, and export RFCs.
Checklist
Use this issue to track InQL RFC 029, proposed in PR #60 at
docs/rfcs/029_metadata_attachments.md.Area
Summary
RFC 029 defines typed metadata attachments for InQL semantic targets. Attachments let authors, library code, sessions, adapters, and inspection tooling associate structured metadata with plans, fields, expressions, read roots, quality checks, policy checkpoints, and execution evidence without collapsing those facts into comments or untyped maps.
Motivation
Relational evidence needs a common attachment model. Without one, lineage, policy, data quality, adapter coverage, and export metadata will each invent their own ad hoc key/value structure, making evidence hard to validate, compare, serialize, or consume.
Proposal sketch
Define typed metadata attachment records that bind a target, attachment type, source, lifecycle, and structured payload. The RFC should distinguish authored metadata from derived or observed metadata and require versioned, inspectable schemas for serialized evidence.
Alternatives considered
The draft rejects untyped string maps, embedding all metadata directly into every plan node, and letting downstream systems attach arbitrary facts without a shared InQL target model.
Impact / compatibility
This is additive. Existing plans can remain metadata-free. Tools that require metadata must distinguish absent metadata from an empty metadata set.
Implementation notes (optional)
Review PR #60 for the draft. This work depends on RFC 028 semantic targets and feeds the quality, governance, bundle, and export RFCs.
Checklist