Rule

Base R6 class for reactive production rules.

Rule

Rules use a two-phase input model for performance:

• planning/scheduling phase (inputs(), bind(), requirements()) works with ResourceSpec objects, not loaded resource payloads;
• execution phase (run()) receives materialized input values loaded by the execution engine worker just-in-time.

This avoids repeated scheduler-side loading of large resources (for example GWAS summary statistics) and reduces data copying across async workers.

Practical guidance:

• keep requirements() based on output.spec and bound specs/metadata;
• if bind() needs cached metadata (e.g. genome build), prefer store$load.meta() over full store$load();
• reserve heavy data access for run().

name — Character. Rule name.

Public methods

• Rule$new()
• Rule$matches()
• Rule$inputs()
• Rule$bind()
• Rule$requirements()
• Rule$run()
• Rule$print()
• Rule$clone()

Method new()

Create a Rule base class

Usage

Rule$new(name)

Arguments

name — Character; rule name

Method matches()

Check if the rule can produce an output

Usage

Rule$matches(output.spec)

Arguments

output.spec — ResourceSpec

Returns

Logical

Method inputs()

Build inputs for an output

Usage

Rule$inputs(output.spec)

Arguments

output.spec — ResourceSpec

Returns

Named list of inputs (ResourceSpec or NULL)

Method bind()

Bind derived inputs.

Usage

Rule$bind(inputs, store, output.spec = NULL)

Arguments

inputs — Named list of inputs.

store — ResourceStore.

output.spec — Output specification being prepared.

Returns

Named list of inputs.

Method requirements()

Compute requirements

Usage

Rule$requirements(output.spec, inputs)

Arguments

output.spec — ResourceSpec

inputs — Named list of bound input specifications.

Returns

Named list with cores and memory (GiB, gibibytes -- the same unit as ExecutionEngine$max.memory and workspace$config$max.memory). A value strictly between 0 and 1 is rejected at task-ready time as an almost-certain unit error.

Method run()

Execute the rule

Usage

Rule$run(output.spec, inputs, logger = pg.logger())

Arguments

output.spec — ResourceSpec

inputs — Named list of materialized resource values loaded by the execution engine worker.

logger — PolyGeniusLogger

Returns

List with data, meta, logs

Method print()

Print rule summary

Usage

Rule$print(...)

Arguments

... — Unused

Method clone()

The objects of this class are cloneable with this method.

Usage

Rule$clone(deep = FALSE)

Arguments

deep — Whether to make a deep clone.

Other engine-core: ExecutionEngine, ExecutionScheduler, ResourceResolver(), ResourceSpec, ResourceSpecSet(), RuleRegistry, TaskExecutor(), execute(), resources.factory, resources.type