Run a multi-phase workflow project — RunProject • workr

RunProject() runs multiple workflow phase folders in sequence, sharing one lData object across phases. Each subdirectory of strPath is treated as a phase. Phases are executed in sorted order (or in the order specified by strPhases). Within each phase, workflows are loaded via MakeWorkflowList() and run via RunWorkflows().

Usage

RunProject(
  strPath,
  lData = NULL,
  lConfig = NULL,
  strPhases = NULL,
  bRecursive = FALSE,
  bReturnResult = TRUE,
  bKeepInputData = FALSE,
  strResultNames = c("Type", "ID"),
  bContinueOnError = FALSE
)

Arguments

strPath: character Path to the project directory. Must contain one or more subdirectories, each holding workflow YAML files.
lData: list Initial named list of data objects. Default NULL.
lConfig: list Runtime configuration hooks. Top-level LoadData and SaveData hooks are passed through to RunWorkflows() for each phase and run for each workflow in that phase. Use lConfig$phases for phase-specific workflow hooks and lConfig$project for hooks that run once for the whole project. Phase handoff settings such as input.from_phases belong in per-phase _config.yaml files, not in lConfig$phases.
strPhases: character Optional vector of phase folder names to run. If NULL (the default), all sorted subdirectories of strPath are used.
bRecursive: logical Passed to MakeWorkflowList(). Default FALSE.
bReturnResult: logical Passed to RunWorkflows(). Default TRUE.
bKeepInputData: logical Passed to RunWorkflows(). Default FALSE.
strResultNames: character Vector of length two passed to RunWorkflows(). Default c("Type", "ID").
bContinueOnError: logical Passed to RunWorkflows(). If TRUE, failed workflows are recorded and later workflows/phases still run. Default FALSE retains the existing fail-fast behavior from RunWorkflows().

Value

A named list with one element per phase. Each element contains the result returned by RunWorkflows() for that phase. When bContinueOnError = TRUE, returns a summary with results, status, and failures.

Details

Phase folders and order:

strPath must exist and be a directory; otherwise execution stops.
By default phases are the immediate subdirectories of strPath, executed in alphabetical order.
When strPhases is supplied, phases run in the exact order given (caller order is preserved, not re-sorted).
A strPhases entry that does not exist on disk stops execution.
A phase folder that contains no workflow YAMLs is skipped with a workflow log message; execution proceeds with the remaining phases.

Per-phase _config.yaml files may define input and output maps. Unknown keys stop execution. Without _config.yaml, all prior phase outputs are available to the next phase.

Think of _config.yaml and lConfig$phases as two phase-scoped settings at different layers. _config.yaml is project orchestration config: use input.from_phases or input.from_results to select earlier phase data, input.include_workflows to pass workflow definitions, and input.extra for static values. Use output.wrap_as to store a phase result under one name or output.transform to apply a function before later phases use the result. lConfig$phases[[phase_name]] is workflow runtime config: it is merged into the lConfig passed to RunWorkflows() for that phase, so use it for phase-specific hooks such as LoadData and SaveData.

Load/save hooks:

Put LoadData or SaveData at the top level of lConfig to pass those hooks to RunWorkflows() for each phase. RunWorkflows() then applies them to each workflow in that phase.
Put hooks under lConfig$phases[[phase_name]] to use them only for that phase. Phase hooks override top-level hooks with the same name.
Put LoadData or SaveData under lConfig$project to run once before or after the whole project. Project LoadData must return an lData list; project SaveData receives the project result.

Examples

if (FALSE) { # \dontrun{
results <- RunProject("inst/example_workflows")
} # }