Constraint-Governed Unfolding Structure
About this pattern
This is a generated FPF pattern page projected from the published FPF source. It is canonical FPF content for this ID; it is not a FPF Reference product feature page.
How to use this pattern
Read the ID, status, type, and normativity first. Use the content for exact wording, the relations for adjacent concepts, and citations to keep active work grounded without pasting the whole specification.
Type: A.22 specialization of
U.StructureStatus: Stable Normativity: Normative unless explicitly marked informative
Use this pattern when a diagram or explanation shows several possible next actions, but readers may mistake one displayed path for the required work sequence. Start with one ordinary question:
Relations
Content
Use This When
Use this pattern when a diagram or explanation shows several possible next actions, but readers may mistake one displayed path for the required work sequence. Start with one ordinary question:
Which alternatives are available now, and what condition blocks each one?
Name the decision or question, the visible alternatives, the condition for each alternative, and the facts available now. If a needed fact or rule is missing, mark that alternative unknown and stop when this answers the practical question. A useful explanation need not first become a formal record or an admitted structure.
Open the formal branch only when the team must qualify, persist, compare, publish, or rely more strongly on the structure. A ConstraintGovernedUnfoldingStructure (CGUS) is one A.22 U.Structure whose locally named loci, constituents, obtaining relations, and constraints define at least two potential continuations across the cases allowed by those constraints. A separate result says which alternatives are enabled, disabled, or unknown for one case and time window.
Do not use CGUS merely because a card, graph, table, narrative, prompt path, or README line looks route-shaped. A single recommendation or displayed sequence is not enough. The structure may branch, join, cycle through subject relations, remain partially ordered, or leave several alternatives live at once. A result with zero or one enabled alternative can still concern that same branching structure.
What changes in practice. Practitioners correct the visible alternatives and their conditions before completing formal fields. They keep potential structure separate from the result for the present case, and they mark an affected alternative unknown at its first unresolved fact or rule instead of inventing a continuation. Display order alone neither prescribes nor performs Work.
Problem Frame
FPF often needs to explain how several identified things and relations constrain what may follow without turning that explanation into a workflow. The shared object is one A.22 structure. CGUS adds local loci and a membership test for potential branching; a continuation judgement then evaluates one case.
Descriptions, publication forms, evidence, assurance, authorization, work plans, performed Work, architecture claims, and mathematical models can be used alongside that structure. They remain separate objects and claims under the patterns that define or test them.
Problem
A route-shaped explanation can hide the relations and constraints that make an alternative available. Readers then follow the displayed order as if it were a required procedure, or they treat a condition label as proof that the condition is true now.
The opposite repair is also harmful: authors replace the simple decision question with a large admission, replay, publication, and assurance package. The formal package becomes harder to use than the misleading card it was meant to correct.
Forces
Solution
Ordinary branch
Write the smallest useful answer in domain language:
- name the decision or question;
- list the real alternatives;
- state the condition for each alternative;
- state the facts known for this case;
- mark each alternative
available,blocked, orunknown, and name the first missing fact or rule.
For example, a design review has two alternatives: accept the design or repair it. Acceptance needs both checks to pass. Repair needs at least one failed check and a repair proposal that concerns this design.
That corrected card is already useful. It keeps both potential alternatives visible and refuses to invent the missing relation. Continue only if a named later use needs formal structure identity or replayable results.
Formal qualification branch
Use the four A.22 discriminators to identify one U.Structure:
- its constituent references;
- the obtaining relation occurrences it selects;
- the applied constraint claims;
- the named selection-use frame: the question, admissible action, and stop or return condition. Any optional explanatory overread follows F.19:4 and remains outside the identity basis.
CGUS membership adds locally declared loci and bindings that expose how those constituents matter to the unfolding question. The selected relations and constraints must define at least two potential continuation candidates across allowed cases. The current continuation result, a description, or a publication field adds no structure-identity discriminator.
forbiddenOverread? and groundedForbiddenOverread? name the same optional explanation. Use F.19:4's plausible-reader test to decide whether it is useful here.
A CGUS locus belongs to this structure, not to a reusable relation declaration:
The constituent must already belong to the A.22 identity basis. A locus binding neither changes that constituent's kind nor creates a relation. Do not use an A.6.5 SlotSpec as a free-standing structure position.
When replay must identify one participant in a relation occurrence, retain the direct relation definition, the occurrence, the participant order, and the participant binding:
Add a RelationSignature and its declaration-local SlotSpec together only when an existing reusable declaration is itself needed for replay. Neither declaration value substitutes for the obtaining occurrence. The CGUS has no ambient context field.
Judge each continuation separately. An immediate local use may keep the following values in the explanation; persistence or replay may place them in an ordinary C.2.1 result episteme.
A claim reference identifies the claim being applied; it does not show that the test applies or that its condition is satisfied. An obtaining relation is not a condition claim. Keep these two basis branches distinct and derive the case result only from completed judgements.
The membership test concerns potential topology. Changed facts, evidence, test outcomes, or time windows normally change a judgement and the current set, not the structure. Reidentify the A.22 structure when a constituent, selected obtaining relation occurrence, applied constraint, or named use frame changes. Reapply CGUS membership when a locus binding or potential-continuation row changes.
Four separate decisions
Do not turn qualification, case evaluation, description adequacy, and downstream reliance into one score.
Potential branches and joins remain part of the structure even when the present case enables one or none. A linear teaching slice neither removes the other topology nor fixes the order of performed Work.
Explanations, descriptions, and the non-workflow boundary
Before qualification, an ordinary explanation is about the domain question or proposed alternatives. If persistence is needed, its C.2.1 EntityOfConcern remains that question or proposed set, not a CGUS that has not yet qualified.
After qualification, a whole-structure description may describe loci, bindings, relations, constraints, potential branches, case results, and relevant omissions. A separate demonstrative slice may show one traversal for a declared teaching or comparison use. That slice is a C.2.1 episteme: its exact claim content, the qualified CGUS as EntityOfConcern, and its effective U.ReferenceScheme jointly recover its identity. DemonstrativeUnfoldingSlice@Context is readable lineage for this possibility, not a U.Kind or an exact slice by itself. The slice neither creates nor reidentifies the structure. Use C.33 only when hidden or lost structure in its carrier matters to a declared architecture use within C.33's scope.
Displayed words such as move, next, and path remain ordinary language unless a stronger claim requires another kind. A proposed action, a plan item, a U.WorkPlan, dated U.Work, and an actual U.Transformation are different values. Use E.10.MOVE, A.15, and A.3 only when that distinction changes the claim; a display performs and authorizes nothing.
For a transformation-flow use, apply E.18.3. It owns the choice among one TFS, one parent-relative SubflowRef, or an E.18.NET network and the corresponding position and demonstration locators. CGUS keeps only its local locus bindings and potential topology; it does not copy the network's members, positions, valuations, Work, transformations, or tags.
Cite another pattern only when its content supplies a needed definition, constraint, test, method, evidence rule, or assurance rule. For example, use C.30 for an architecture claim and C.32 for architecture-candidate synthesis, E.23 for repeated quality improvement under a declared evaluation, G.11 for source currentness, C.29 for a mathematical-lens claim, A.10 for claim-bound evidence reliance, or B.3 for an actual named assurance claim.
If a durable name or a relation between local senses is the question, use F.17, F.18, or F.9 after the value has been recovered. Do not copy their naming or Bridge procedures into this pattern. Entry cards and publication faces remain under E.11 and E.17.
Replay and change localization
Replay structure identity from the four A.22 discriminators. Replay CGUS membership from the local locus bindings and potential topology. Replay the case result from each candidate's basis, inputs, facts, polarity, dependent occurrences, time window, outcome, and reason.
Localize change before reopening wider work. A changed constituent, selected occurrence, constraint, or use frame can reidentify the A.22 structure. A changed locus binding or potential-continuation row reopens CGUS membership. A changed fact, evidence item, test result, or time window normally reopens only the affected judgement and current set. A changed omission reopens the affected description use. A changed neighboring claim stays with the pattern that defines or tests it.
Complete Worked Case
Return to the design review from 4.1. The ordinary card becomes formal only because the team now needs to retain and compare the review basis across editions.
The structure has two potential continuations although this case enables only repair. The relation rows state their predicates and ordered participants; the judgement rows state the tests, applicability, inputs, facts, polarity, dependent occurrences, window, outcomes, and reasons.
If RepairProposalTargetsCandidate@DR-27 or its participant binding is missing, the repair result becomes unknown — proposal target not established. If the structure's identity was established on another sufficient basis, only this case result is incomplete. If that occurrence belongs to the claimed identity basis, this structure claim also remains provisional.
If a later thermal check passes while the service check still passes, acceptance becomes enabled and repair becomes disabled. If the constituents, selected occurrences, constraints, use frame, locus bindings, and potential topology have not changed, the CGUS keeps its identity and membership. A replacement result episteme or relation occurrence must first be compared under the A.22 discriminators.
Bias-Annotation
Conformance Checklist
Common Anti-Patterns And Repairs
Consequences
CGUS preserves the usefulness of a route-shaped explanation without making it a workflow. Ordinary use is cheap: decision, alternatives, conditions, facts, honest result, and stop. Formal use costs more because structure identity, CGUS membership, the case result, and any description or neighboring claim must remain separately checkable.
This separation prevents a changed fact from reidentifying a stable structure and prevents a missing publication, evidence, or assurance value from erasing a useful ordinary answer.
Rationale
The recurring object is a thin specialization of A.22 U.Structure, not a new root kind. Constraint-based process modeling, object-centric querying, artifact-centric modeling, acausal modeling, and FPF pattern use all distinguish a constraint-bearing structure from a performed trace, work order, view, publication, solver run, or example path.
The same distinction appears in acausal engineering models: component relations and constraints can be stated before an analysis chooses a calculation direction. FPF adopts only that general separation. Mathematical models, analyses, executions, results, and publications keep their own kinds and rules.
SoTA-Echoing
As of 2026-08-04, OCPQ and Dyad are the current comparators used here. Modelica, Declare, DCR, artifact-centric/GSM, and CMMN remain lineage where their distinctions are useful. Reopen this choice when a newer object-centric constraint method or relation-first engineering language changes the treatment of objects, relations, analyses, or live alternatives.
Relations
Specializes: one A.22 U.Structure whose local locus bindings, obtaining relations, and constraints define at least two potential continuations across allowed cases. Continuation judgements, descriptions, slices, loss notes, and neighboring claims are separate results or uses.
Specialized by: E.18.3 when the same structure also satisfies its transformation-flow condition. Local applications include architecture, abduction, improvement, narrative, grounding, currentness, and first-entry uses only when their own constituents, relations, constraints, and use frames are recoverable.
Coordinates with: A.6.P and A.6.5 for relation occurrence and reusable declaration precision; E.18, E.18.NET, and E.18.3 for transformation-flow substrates; A.3 and A.15 for Method, plan, Work, and Transformation claims; A.10 for claim-bound evidence reliance, B.3 for an actual named assurance claim, A.20 for internal-constraint validity, and A.21 for gate decisions; C.30 for architecture claims and C.32 for architecture-candidate synthesis; E.23 for repeated quality improvement under a declared evaluation; G.11 for currentness; C.29 for mathematical-lens use; C.33 for description loss that affects a declared architecture use; E.11 for entry, E.17 for a source-backed reader face, and E.24.PUB for publication occurrence, form, carrier, audience, bounded use, and availability; and F.17, F.18, and F.9 for source-local sense, durable naming, and Bridge claims.
Does not replace any pattern that supplies the definition, constraint, test, method, evidence rule, or assurance rule for a neighboring claim.
Use F.19 for ordinary precise-plain-language repair and the plausible-reader test for an optional explanatory overread.
A.22.CGUS:End
Last Updated: 2026-09-10 — upstream FPF commit a87d0ef4 (github.com/ailev/FPF)