Validation
Behavior

Validation Is Now Executable for the Solved Core

Specification Sections 01–07 now define deterministic candidate validity and normalization rules for the current Mandatory Core. Survey Domain validity is exact but remains parameterized by the governing half-span H until FMD-04B closes.

Conformance may therefore test exact input classification, normalization eligibility, semantic preservation, invalidity, and canonical serialization behavior without relying on tolerance or implementation-specific guesswork.

Accept Authoritative Form

An input that satisfies the governing Specification and is already expressed in the canonical form required for its type, operation, or encoding shall be accepted as canonical valid.

Its mathematical and spatial meaning shall be interpreted exactly and without modification.

Normalize Equivalent Meaning

A valid input that expresses legitimate canonical meaning but is not yet in canonical representation may be classified as valid noncanonical where the Specification defines an exact normalization path.

The implementation shall normalize that input deterministically to the required canonical form while preserving identical mathematical meaning.

A valid noncanonical state exists only where exact equivalence and normalization are defined. It shall not be used as a general repair category for malformed or ambiguous input.

Invalid Means No Canonical Meaning Under the Claimed Profile

An input shall be classified invalid when it violates the grammar, mathematical constraints, geometry rules, canonical type requirements, Domain validity, encoding rules, or other mandatory conditions of the governing Specification profile.

Invalid input shall not be promoted into canonical truth through repair, inference, tolerance, or fallback interpretation.

Exact Coordinate Scalars Must Be Structurally and Mathematically Valid

CRPC validation shall enforce the exact rational representation required by the Specification.

Invalid conditions include prohibited denominator form, malformed integer representation, noncanonical zero where canonical form is required, or any other representation that cannot resolve to a valid exact CRPC scalar under the claimed profile.

Equivalent reducible values may be valid noncanonical only where the input form is permitted and deterministic reduction preserves exact value.

Reject Reference Outside the Governed Survey Domain

A mathematically structured coordinate or geometry whose required spatial content falls outside the governed Survey Domain shall fail Domain validation according to the applicable Specification rules. Under the current candidate model, the Domain is D_H = {(x,y) : -H ≤ x ≤ H and -H ≤ y ≤ H}.

It shall not be reinterpreted as Exterior, neighboring territory, alternate World-space, or another semantic category merely because its syntax can be parsed.

Survey validity is resolved before any higher-layer World-membership meaning is considered. FMD-04A supplies the closed-square geometry and boundary semantics; FMD-04B leaves the numerical value of H open. Conformance applies the governing value or parameterized rule; it does not select the capacity.

The earlier ±1,000,000 Pang half-span is a superseded working candidate and shall not be used by Conformance as a canonical Domain limit.

Primitive Geometry Must Satisfy Exact Structural Rules

A Point shall contain valid exact coordinate values and satisfy applicable Domain rules.

A Segment shall contain two valid Point endpoints and satisfy the structural and validity rules established by the Specification. A zero-length or otherwise prohibited Segment shall be invalid rather than silently collapsed or repaired.

Polygonal Extent Validity Is Exact

A candidate SCPE shall be validated through the exact geometry rules of the Specification.

Required conditions include sufficient distinct vertices, valid Points, valid consecutive Segments, nonzero area, one simple closed cycle, no prohibited self-intersections, and compliance with the governed Survey Domain.

Exact redundant collinear middle vertices may be valid noncanonical where normalization removes them without changing geometry. Structural invalidity shall not be repaired into a different polygon.

Reject Invalid FSF-CJSON Structure

FSF-CJSON-1.0 input that violates required structural or byte-level rules shall fail validation as canonical serialization.

Examples include malformed JSON, prohibited numeric-literal use for exact coordinate values, duplicate keys, prohibited or unknown fields under the governing format, invalid type structures, noncanonical exact-value strings, or other violations defined by Specification 06.

A parser shall not infer intended canonical spatial meaning from structurally invalid canonical interchange.

ECEM Rejects Coarse/Fine Point Invention

A valid canonical Point denotes one exact position under ECEM.

Inputs that attempt to assign foundational parent/child Point semantics, coarse/fine Point status, uncertainty-area meaning, or implicit descendant sets shall not be accepted as canonical Point semantics unless a future Specification explicitly introduces such a governed type.

Compatible representational extension may add newly expressible exact values without changing old Point meaning.

Do Not Guess Canonical Meaning

An input that permits more than one possible canonical interpretation under the applicable rules shall not be accepted as authoritative Survey truth.

Ambiguity shall result in deterministic validation failure unless the Specification defines one exact normalization rule that resolves the representation without changing meaning.

Approximation Is Not Canonical Truth

Estimated, rounded, approximate, pending, disputed, or incomplete values shall not be accepted as canonical Survey truth merely because they are spatially plausible.

Higher Architecture domains may preserve such states for their own purposes, but the Foundational Survey Fabric shall distinguish them from authoritative exact input.

Detect Loss Rather Than Hide It

A transformation or conversion is lossless only when the exact mathematical object remains unchanged.

Truncation, rounding, rasterization, reduced storage precision, approximate simplification, or other information-discarding conversion shall be recognized as lossy and shall not be treated as fully equivalent canonical representation.

Normalization Must Preserve Meaning Exactly

Validation may normalize CRPC form, Segment endpoint order, SCPE traversal, exact redundant collinear representation, canonical key order, or other explicitly governed representation differences only where exact semantic preservation is guaranteed.

Validation shall not relocate a Point, alter an Extent, change precision meaning, substitute a nearby representable value, or convert invalid input into a different valid object.

The Same Input Receives the Same Judgment

Independent conforming implementations shall reach the same validity classification for the same input under the same Specification profile.

Where normalization is permitted, they shall also produce the same canonical normalized result.

Unsupported Is Not the Same as Invalid Mathematics

Where the Specification has not yet defined canonical semantics or output closure for a capability, Conformance shall not invent a validation rule that makes the capability canonical.

Such behavior may be classified as unsupported or outside the claimed profile rather than falsely accepted as canonical.

Current examples include unresolved general Boolean/composite result forms, arbitrary exact rotation, general distance/path-length scalar forms, and future geometry not yet formally selected.

Reject for Defined Reasons

Conformance shall distinguish validation failure conditions sufficiently for deterministic testing, including malformed syntax, invalid exact-value structure, invalid geometry, out-of-Domain reference, unsupported profile capability, invalid serialization, version mismatch where applicable, and other governed failures.

The complete canonical machine-readable error taxonomy and final error identifiers remain open.

Human-readable diagnostic wording may vary unless a future profile explicitly standardizes it.

Invalid Input Must Not Be Silently Corrected

A conforming implementation shall not silently repair invalid canonical input through snapping, rounding, truncation, coercion, inferred defaults, guessed intent, fallback geometry, or implementation-specific substitution.

Any convenience layer that proposes a correction must remain explicitly separate from canonical FSF validation and must not present the corrected object as the original canonical input.

Every Validation Class Must Be Demonstrable

Validation behavior shall be tested through Reference Vectors or equivalent executable fixtures covering canonical valid input, valid noncanonical normalization, invalid input, unsupported profile behavior, and canonical serialization validity where applicable.

The solved core is sufficiently deterministic for validation fixtures to be authored now. Reference Vectors demonstrate the governing rules; they do not create new validity semantics.

Validation Has Moved From General Principle to Exact Candidate Behavior

The original section correctly required deterministic acceptance, normalization, and rejection. The integrated Specification now provides enough exact mathematics to define those classifications concretely for the solved Mandatory Core.

The remaining work is primarily executable fixture coverage, formal error taxonomy, profile/version handling, and future validation rules for mathematics that are still open in the Specification.

Primary Referenced Findings

This Conformance section is derived especially from Requirements Findings #30, #57, #71–#75, #78, #80–#83, with additional support from the permanent addressability and exactness requirements.