Cohesive Systems logoCOHESIVE SYSTEMS

Search Cohesive Systems

Ready

Search Cohesive Systems

Find product pages, building blocks, technical articles, and graph definitions.

Control Flow

Control flow describes how one possible action, operation, step, or continuation becomes enabled, selected, and followed by another at a declared boundary.

The term is overloaded across domain processes, system graphs, programs, runtimes, interactions, and operations. Cohesive therefore treats control flow as a qualified family of relations rather than one universal edge type. A model should state what progresses, what selects the successor, which boundary observes that progression, and what an arrow or dependency means.

Control does not necessarily imply a central controller. A successor may be selected by a deterministic transition, data-dependent branch, participant interaction, nondeterministic choice, scheduler, distributed protocol, or environmental event.

Qualified Meanings

MeaningWhat progresses or is controlledCanonical home
process control flowSteps, branches, tokens, forks, joins, waits, recurrence, cancellation, and terminal outcomes within a semantic process structure.process graphs and workflow patterns
program or evaluation control flowExpressions, instructions, calls, returns, exceptions, continuations, loops, and evaluation paths within a program or runtime.reduction and evaluation, runtimes, and scheduling
interaction control flowWhich participant actively pushes, fetches, polls, calls, or delivers at an interaction boundary.interaction control flow
control-flow synchronyWhether one logical operation waits for a result, observation, or completion before continuing.synchrony and asynchrony
operational controlAuthorized intervention that pauses, resumes, routes, drains, replays, or otherwise changes running-system behavior.operational control
flow controlHow much work may be admitted or remain in flight in response to capacity.flow control
feedback controlHow observations of behavior regulate future action through a controller and actuator.control theory and control models

The boundary qualifier is essential. One process-progression edge may lower into several program evaluations and distributed interactions. A synchronous-looking program call may lower into queues and callbacks. A pull interaction may drive data opposite to the interaction-control direction without reversing process progression.

Control flow should not be inferred from:

  • Data flow, which says where a value moves.
  • Causal flow, which says which occurrence may have influenced another.
  • Ordering, which constrains relative positions without necessarily selecting the next step.
  • Scheduling, which selects among enabled work but does not necessarily define which successors are semantically valid.
  • Orchestration authority, which says who owns a process decision surface.
  • Flow control, which regulates capacity rather than branch or token progression.
  • Feedback control, which regulates a controlled variable rather than selecting the next semantic or program step.

These relations may align in a particular realization, but an unqualified arrow must not stand for all of them. Flow views should label whether an edge means carried-value movement, process progression, causality, interaction drive, or another relation.

Modeling Discipline

When control flow matters, state:

  • The abstraction layer, subject, and boundary.
  • The current position or continuation state and the possible successors.
  • What enables, selects, rejects, suspends, or cancels a successor.
  • Whether choice is deterministic, policy-driven, nondeterministic, scheduled, or externally triggered.
  • How concurrency, forks, joins, recurrence, and termination are represented.
  • Whether waiting is logical, durable, or physically blocking.
  • Which state must survive interruption for progression to resume correctly.
  • What each control-flow edge claims and which related data, causal, interaction, and ordering edges are separate.

This qualification preserves correspondence across layers without collapsing a semantic process into a program counter, an interaction driver, a workflow engine, or a capacity protocol.

External References

Related concepts: process theories, control theory, control models, process, process graphs, workflow patterns, flow views, interaction, interaction control flow, flow control, synchrony and asynchrony, operational control, reduction, evaluation, and confluence, nondeterminism and choice, runtimes, scheduling, ordering, causality, orchestration and choreography, trace and feedback.