FPF Reference

    One connected reading and repair

    Preface node heading:one-connected-reading-and-repair:100500

    What this page is

    This is generated FPF reference text from the specification preface or supporting sections. It helps interpret FPF; it is not FPF Reference product documentation.

    Methodology

    Use it to understand how the specification wants to be read, then return to a route, pattern, or work packet for active work. Cite generated IDs only when the wording changes the task decision.

    Content

    1. State the governing message. Say what object, claim, action, event, or distinction the span needs to convey. Mark process traces, status language, reference boilerplate, quality proof, defensive caveats, ornamental detail, and other apparatus that may be displacing it. Apparatus receives no protection merely because it is true or polished.
    2. Recover the predicate and its participants. Identify the operation and every participant that changes it. This may be an actor, object, source, target, result, or another required operand. Apply the same question to verbal and relational nouns: recover of what, for what, between what, or another required participant. Leave an argument implicit only when one intended value is cheaply and uniquely recoverable from the local span.
    3. Check predicate compatibility. Recover what the complete claim asserts under its negation, modality, or conditions, and test the subject and participants under the intended literal or metonymic reading. A positive assertion must assign its predicate to a compatible kind. A denial may correct an evidenced type mistake under the plausible-reader test below; without that ground, “evidence does not notice the error” introduces an idle alternative. Keep ordinary metonymy when the relation is established and the capable participant or work remains recoverable: a diagram may show, a framework may help, a reminder may cue, and a constraint may limit.
    4. Resolve referents and kinds. Pronouns, demonstratives, omitted heads, and repeated labels must select one locally appropriate referent. Preserve the object kind, claim or relation kind, slot or relation position, use and publication boundary, and flow distinction whenever one changes the claim. A shared grammatical position does not make different FPF kinds interchangeable.
    5. Test contribution. Try deleting every optional contrast, guard, modifier, example, coordinated member, and extra proposition. Remove it when the plausible intended reader can still recognize, understand, decide, and act in the same way. Local truth and grammatical fit do not earn a phrase a place by themselves.
    6. Resolve coordination and lists on two axes. First ask whether the receiving use needs a series at all. A list or parallel construction earns its form only when the reader must distinguish or retain its members together; otherwise select the governing claim, relation, or representative case. If a series is needed, determine its membership semantics. State the proposition or action it serves; use one kind or predicate only when it fits every member; distinguish a closed set, illustrative examples, alternatives, a sequence, several direct relations, and a failed ontology. A closed set needs its kind, membership rule, and closure. Illustrative examples need the proposition or kind first and a non-exhaustive cue when a plausible reader could mistake them for a classification. Then test discourse load: keep a member only when it adds a distinct consequence; reduce coordination repeated at several grammatical levels and modifier chains that make the reader retain needless branches or postpone the governing message. Length is evidence to inspect, not a verdict. If a list hides an FPF kind, relation, or structure that ordinary reading cannot recover, use the pattern that defines or tests it, or return the unresolved meaning as a blocker.
    7. Foreground, rewrite, and compare. Put the governing event, claim, requested action, or decision before optional atmosphere, examples, caveats, and catalogues. A prerequisite may come first when it is needed for safe interpretation or action. Write the shortest ordinary technical sentence that preserves every live predicate and participant, established term, polarity, and action-changing detail. Such detail can include quantity or threshold, sequence or timing, criterion or tolerance, exception, and applicability. Compare before and after: any unsupported change of kind, relation, scope, use, currentness, or operational effect is a loss and blocks the rewrite unless another accepted decision authorizes it.

    Keep ontology visible only where it carries the sentence. A term-source or type annotation is needed only when it changes how the reader identifies the object, kind, relation, slot, use, publication boundary, admissible use, or applicable rule. A record, card, table, schema, data structure, dashboard, or named form remains apparatus unless it carries one of those values. If ordinary domain wording already preserves them, keep the ordinary sentence. "The aircraft flies" is better than a typed expansion unless the flight function, system kind, or slot relation is under repair.

    Precision before a coarsened rendering. When head kind, qualifier claim, or comparison basis remains unresolved in FPF-governed prose, use this working order:

    Restore the head kind first; a narrowing qualifier such as comparative, safe, interactive, or reliable does not by itself restore that kind. Then unpack the qualifier claim, then check whether the comparison or escalation basis is homogeneous. Only after that may a later Plain, didactic, or coarsened rendering admissibly relax the sentence, while keeping the more precise upstream interpretation recoverable.

    Judge homogeneity against the comparison or condition rule actually used; recover distinct relations separately when that rule combines them. The basis may be a homogeneous claim-kind criterion, threshold, or named defining, constraining, or source-relation condition.

    Treat exact, direct, current, governed, subject, owner, defining, and similar qualifiers as content only when they distinguish live alternatives. Remove them when no such contrast changes the truth, action, stop, or reliance. A PatternID may remain an ordinary citation; expand it into a claim-bearing episteme, ClaimGraph, U.MethodDescription, U.Method, actor, assignment, U.Work, or another formal identity only when the current claim or a named later use depends on that distinction.

    Keep ordinary practitioner action and instrumental pattern-use wording ordinary when it does not assert a particular dated Work occurrence. “Use E.9 to record the decision” and “the framework maintainer compares the editions” need no invented Method, MethodDescription, performer, assignment, or Work identity.

    Open the identity-bearing branch only when the sentence deliberately asserts a particular dated U.Work occurrence. Then point to its basis: A.13 first, independent A.15.1 Work admission second, and F.6 afterward only for precise assignment-bound attribution. Add a local system-role kind or a separate System-classification judgment only when that neighboring claim matters. Treat a pattern episteme as a U.MethodDescription only after A.3.2 establishes that it has an already admitted Method as its EntityOfConcern and explains how that Method is performed. Otherwise cite the applicable pattern content as guidance and use A.3.1 for the Method itself.

    Last Updated: 2026-09-03 — upstream FPF commit b999972c (github.com/ailev/FPF)