Run a module asynchronously — run

Executes a module and returns a promise that resolves to the result. Useful for running multiple modules in parallel. Ordinary PredictModule objects use the provider's native async path. ProgramOfThought, CodeAct, and RLM modules use an isolated background process when configured with interpreter_factory. Caller-owned runners are rejected because they cannot be safely shared across concurrent invocations.

Usage

run_async(module, ..., .llm = NULL)

Arguments

module: A dsprrr Module object
...: Named inputs matching the module's signature
.llm: Optional ellmer Chat object

Value

A promise that resolves to the structured output

Examples

if (FALSE) { # \dontrun{
# Run multiple modules in parallel
promises <- list(
  run_async(mod1, question = "Q1"),
  run_async(mod2, question = "Q2"),
  run_async(mod3, question = "Q3")
)

# Wait for all to complete
promises::promise_all(.list = promises) |>
  promises::then(function(results) {
    # Process results
  })
} # }