AEON v1 Addressing and References Reference

Status: official v1 companion reference Scope: canonical paths, reference forms, quoted segments, attribute selectors, and reference legality.

1. Canonical Path Model

Canonical path identity in AEON v1 uses only:

• member segments
• index segments

Examples:

Identity rules:

• a.b means two member segments;
• ["a.b"] means one member key that contains a dot;
• attribute selectors are valid in addressing expressions, but are not canonical path identity segments.

Authoritative grammar sketch:

Canonical notes:

• non-bare keys are rendered in double-quoted bracket form;
• lists and tuples use numeric index segments;
• node children also use numeric index segments;
• canonical path identity is deterministic across equivalent quoted spellings.
• quoted root-member traversal uses the explicit dot form $.["..."];
• bare root-bracket member traversal such as $["a"] is not part of the Core v1 path grammar.

2. Key Segment Forms

2.1 Member Traversal

Bare member traversal:

Quoted member traversal:

Nuances:

• ~a.b resolves member a, then member b;
• ~"a.b" and ~["a.b"] are equivalent initial-segment forms for one quoted member key literally named a.b;
• ~["a.b"] resolves a single key literally named a.b;
• ~a.["b.c"] resolves member a, then a quoted member literally named b.c;
• ~$.["a.b"] is the explicit root-prefixed quoted-member form;
• ~$["a.b"] is not part of the Core v1 reference grammar;
• quoted segment decoding occurs before identity comparison.

2.2 Indexed Traversal

List/tuple/node-child addressing:

Nuances:

• index segments are positional and zero-based;
• index segments participate in canonical path identity;
• index segments may be followed by member traversal or attribute selectors.
• node-child addressing uses the same bracket index form as list and tuple addressing.

3. Attribute Selectors

Attributes live in a distinct namespace from data members.

Attribute selector forms:

Grammar sketch:

Nuances:

• @key addresses a bare attribute key;
• @["key with spaces"] addresses a quoted attribute key;
• quoted member and attribute-key segments must not be empty;
• quoted bracket member segments may follow attribute selectors using .[\"...\"];
• quoted attribute selectors may be followed by ordinary member traversal, quoted member traversal, or index traversal;
• mixed traversal such as ~a@["x.y"].z, ~a@meta.["x.y"], and ~$.a@["profile.name"][0] is valid when each traversed segment exists and is otherwise legal;
• malformed or incomplete forms such as ~a@, ~$.a@[, and ~.["a"] are invalid;
• attribute selectors are part of reference/addressing syntax, not canonical path identity;
• data namespace and attribute namespace are distinct and must not be merged.

4. Reference Forms

AEON v1 supports:

• clone/reference form: ~path
• alias/pointer form: ~>path

Examples:

Nuances:

• ~path references the target value;
• ~>path preserves alias/pointer intent in the AST/AES model;
• ASCII inter-token whitespace may appear between ~ or ~> and the following reference path, but canonical formatting removes it;
• both forms use the same path grammar after the introducer.

AES notes:

• clone and alias remain distinct value kinds;
• path text is preserved structurally rather than flattened to one raw string.

5. Disambiguation Rules

5.1 Dot vs Quoted Key

Then:

5.2 Attribute vs Data Namespace

Then:

These are not the same target.

Attachment-scope note:

• attributes attach to bindings (or node heads), not to already-completed literal values;
• binding-attached attributes remain addressable wherever the binding itself is addressable;
• therefore a@{b=1} = [0] exposes $.a@b, and a = [{x@{b=0}=1}] exposes $.a[0].x@b;
• postfix literal forms such as a = [0]@{b=2} are invalid Core v1 syntax.

6. Reference Legality

Reference legality is deterministic in Core v1.

Invalid:

• forward references
• missing targets
• self-references

Examples:

Nuances:

• no-forward and no-self checks apply independently per namespace;
• explicit attribute references use the same legality model;
• quoted member traversal after attribute selection does not change legality rules; it continues to resolve against the selected attribute value using ordinary member/index traversal;
• legality is a Core concern, not AEOS/schema responsibility.

7. Duplicate Identity and Decoding

Quoted spellings are decoded before identity comparison.

Implications:

• duplicate-key checks use decoded identity;
• duplicate canonical-path checks also use decoded identity;
• escaped equivalents must not create distinct addresses.

Example:

These resolve to the same decoded key identity and should conflict.

8. Canonical Rendering Notes

Canonical output conventions:

• root marker is $;
• bare member segments use .name;
• non-bare members use ["..."];
• attribute selectors use @name or @["..."];
• numeric indices use [n].

Examples:

9. Node Model Boundary

Node children use numeric index segments in canonical paths, matching the ordered child-slot model used by lists and tuples.

Implications:

• the binding that owns the node has a canonical path for the node value;
• node child ordering is structural and reflected through bracket indices such as $.page[0];
• object members inside node children continue normal member traversal, for example $.page[0].title;
• node tags are not independent canonical path segments.

10. Minimum Conformance Reminders

Implementations targeting Core v1 must at minimum:

• support quoted member segments in paths;
• support quoted attribute segments in paths;
• preserve dot-vs-quoted-key disambiguation;
• reject forward references deterministically;
• reject missing targets deterministically;
• reject self-references deterministically;
• report canonical paths consistently in diagnostics and emitted events.